research topics in diabetic retinopathy

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• v.15(11); 2023 Nov
• PMC10739189

Advancement in Understanding Diabetic Retinopathy: A Comprehensive Review

Sharad chaurasia.

1 Medicine and Surgery, Jawaharlal Nehru Medical College, Datta Meghe Institute of Higher Education and Research, Wardha, IND

Archana R Thool

2 Ophthalmology, Jawaharlal Nehru Medical College, Datta Meghe Institute of Higher Education and Research, Wardha, IND

Khizer K Ansari

Azeem i saifi.

Diabetic retinopathy (DR) is a significant global health concern, with its prevalence and severity increasing alongside the rising incidence of diabetes. DR is a leading cause of vision impairment among working-age adults, resulting in substantial economic and healthcare burdens. This article explores the epidemiology and pathophysiology of DR, highlighting the global variation in its prevalence and the associated systemic risk factors. It delves into the complex relationship between glycemic control, duration of diabetes, and medication use in the context of DR development and progression. The review also discusses current screening methods and their implications, emphasizing the need for efficient and scalable approaches. Furthermore, it investigates the various treatment strategies available for DR, including laser photocoagulation, vitreous body excision, and anti-vascular endothelial growth factor (VEGF) therapy, while underlining their limitations and potential side effects. In conclusion, this article underscores the urgency of developing novel preventive and therapeutic approaches for DR. It highlights the potential role of cytokines and growth factors as treatment targets and emphasizes the importance of glycemic control and management of systemic risk factors in mitigating the impact of this vision-threatening complication of diabetes. The article serves as a comprehensive resource for understanding the challenges posed by DR and the need for innovative strategies to address this growing public health concern.

Introduction and background

Diabetes mellitus is a set of diseases in which there is an increase and imbalance in blood glucose levels, which may be due to either impaired insulin production or systemic resistance to the effects of insulin. With > 22 million folks (7%) in the United States having diabetes mellitus, it is remarkably a health burden. Over 176 billion dollars are spent on treating diabetes economically in the United States each year, with ocular problems accounting for more than 20% of the total [ 1 ]. By 2050, it is anticipated that between a quarter and a third of all Americans will have diabetes due to an increase in its prevalence [ 2 ]. According to estimates, 5.5 million adults over 40 had diabetic retinopathy (DR) in 2005; by 2050, that figure is anticipated to rise to 16 million. In 2005, there were 1.2 million cases of vision-threatening DR; by 2050, that number will rise to 3.4 million [ 3 ].

Millions of people's everyday lives are impacted by visual difficulty, which in adults of working age is severe in industrialised nations. The majority of patients with this visual consequence were found despite rigorous glycemia, blood pressure, and lipid-lowering medication treatment by fundus examination. The count of DR patients persists in climbing, and treatment options are scarce. The present treatments come with considerable drawbacks and negative outcomes. Therefore, there is a vital necessity for the creation of novel DR prevention techniques and treatment approaches. Evidence suggests that a particular level of cytokines changes as DR develops but before clinical presentation [ 4 ].

Methodology

We conducted an extensive search across electronic databases, including PubMed, MEDLINE, Embase, Google Scholar, and ResearchGate, and explored the available English-language literature. Additionally, it was the focus of a separate inquiry. The MeSH terms were "Diabetic retinopathy" OR "Diabetes-related blindness"; "diabetic retinopathy prevalence" OR "risk factors; "development" OR "advancement"; "Pharmacological management of diabetic retinopathy" OR "Anti-VEGF therapy in diabetic retinopathy"; “Surgical management of diabetic retinopathy”; "Screening strategies for diabetic retinopathy" OR "telemedicine in diabetic retinopathy". The articles included in this review adhere to the following criteria: they encompass studies solely focused on progress in comprehending DR and novel treatment approaches, and they are studies conducted in the English language within the last two decades. Figure ​ Figure1 1 illustrates the utilization of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) methodology in the research process.

PRISMA: Preferred Reporting Items for Systematic Reviews and Meta-Analyses

Image credits: Sharad Chaurasia

Epidemiology and pathophysiology

Globally, there were 103 million cases of diabetes in 2020; by 2045, that number is predicted to rise to 161 million. The primary cause of this increase is the world's rapidly expanding diabetic population, which is concentrated in Africa, the Western Pacific, the Middle East and North Africa. There have been reports of a significant frequency of DR in North America and the Caribbean (33.30%), Middle East and North Africa (32.90%) and Africa (35.90%). The remaining areas had the following rates of DR prevalence: Western Pacific, 19.20%; South East Asia, 16.99%; South and Central America, 13.37% and Europe, 18.75% [ 5 ]. Despite the gravity of this matter, the frequency of diabetes is escalating, particularly in developing Asian nations like India (9.3%) and China (5%) by 2030 [ 6 , 7 ]. With rates fluctuating from 17.6% in an Indian study to 33.2% in an extensive U.S. investigation, preceding isolated studies have indicated significant variation in DR frequency estimates in individuals with diagnosed and undiagnosed diabetes [ 8 ]. Between 2005 and 2014, Taiwan had a prevalence of 0.29 to 0.35% for blindness and poor vision and 3.75 to 3.95% for diabetic eye disease [ 9 ]. From 14.3 per cent in 2006 to 15.9 per cent in 2013, DR was more common in Korea [ 10 ]. Both analyses demonstrated that females having type 2 diabetes mellitus (4%) and (16.6-17%) had a greater preponderance of DR than males (3.5%) and (12.7-14.3%), respectively. The extent of DR not only minimized well-being but also strongly anticipated death from all causes, vascular disease, and non-cancer [ 11 ].

The fact that DR is a microvascular disease has long been known. Hyperglycemia is thought to exert a major influence on the development of damage in the retinal microvasculature. A number of metabolic mechanisms have been connected to hyperglycemia-induced vascular damage, including the hexosamine route, the polyol pathway, the PKC pathway, and the accumulation of advanced glycation end products. Blood flow alterations and blood vessel dilation are the retinal blood vessels' initial reactions to hyperglycemia. In diabetic patients, these changes are regarded as a form of metabolic autoregulation designed to enhance retinal metabolism. Another feature of the initial phases of DR is the depletion of pericytes. There is evidence from both in vitro and in vivo investigations that excessive hyperglycemia causes pericytes to undergo apoptosis. Loss of pericytes causes localised capillary wall outpouching because they are necessary for capillaries' structural stability. Endothelial cell apoptosis and basement membrane thickening are also detected during the development of DR, in addition to pericyte loss, these processes jointly impair the blood-retina barrier (Figure  2 ) [ 12 ].

The advanced glycoxidation products are connected to the altered physiological states and the likelihood of DR, while the C-reactive protein (CRP) and homocysteine are markers of inflammation [ 13 , 14 ]. The plasma proteomic method has helped identify many more indicators. In DR, for instance, diphosphoinositol polyphosphohydrolase 3 alpha, CD 160 antigen, retinol-binding protein 1 (RBP1), haemoglobin subunit gamma 2 (HBG2) and neuroglobin (NGB) were downregulated and increased, respectively. The plasma level of neuroglobin, one of the five proteins listed above, shows potential utility as a diagnostic marker for DR because of the substantive variation between the groups with & without the disease. Furthermore, substantial research is being done on metabolomics, micro RNA, and genetic biomarkers [ 11 ].

Pro-angiogenic cytokines and vascular endothelial growth factor (VEGF) are thought to be the main contributors to DR neovascularization. The cytokine with profibrotic action, a connective tissue growth factor, is another potential cause of fibrosis in PDR. These are interconnected with retinal fibrosis and PDR [ 15 ]. Erythropoietin (EPO) administration and levels appeared to be most closely associated with the occurrence and severity of PDR, according to analysis by Diskin et al. [ 16 ]. A correlation between hematocrit and, most importantly, the total dose of EPO administered is suggested by the deterioration of retinopathy following the start of hemodialysis. According to another study, EPO is a strong retinal angiogenic factor that can stimulate hypoxia-driven retinal angiogenesis while functioning independently of VEGF. Inhibiting these molecular processes in retinal angiogenesis may be a fresh therapeutic approach to stop or prevent abnormal angiogenesis in DR, in accordance with study outcomes [ 4 ].

The commencement of DR is tightly connected with the type of diabetes, duration of diabetes, the level of hyperglycemia, and hypertension. Intense glycemic management lowers the incidence and worsening of DR, which is highly correlated with a higher HbA1c level [ 17 , 18 ]. Recent research has shown that in type 2 diabetes, DR is closely correlated with glycemic fluctuation [ 19 ]. Postprandial hyperglycemia needs to be managed in an effort to avert diabetic retinopathy. Furthermore, the link between hypertension and diabetic retinopathy is clearly demonstrated by the available research [ 20 ]. A strict blood pressure regimen slows the progression of retinopathy. Dyslipidemia, smoking, and a higher body mass index (BMI) are additional risk factors that can be changed to stop the progression of DR (Figure  3 ) [ 21 , 22 ].

Diabetes mellitus and DR

Glycemic Control

A typical measure for glycemic management monitoring is glycated haemoglobin. Numerous studies have repeatedly established that HbA1c is an autonomous risk element for DR [ 23 ]. An increased onset and advancement of DR are linked to a higher HbA1c. The LALES study discovered a one per cent increment in HbA1c and a twenty-two per cent increment in the occurrence of DR. However, the data indicated that the curve peaked at an HbA1c of about eleven per cent [ 24 ]. Elevated HbA1c is a sign of prediabetes mellitus, thalassemia, sickle cell anaemia, and uncontrolled diabetes, which is one of the prominent factors of diabetes mellitus issues, entailing DR. Even an on-target HbA1c level of seven per cent, however, was associated with an absolute risk of seven point nine per one thousand patient-years for retinal laser photocoagulation, in accordance with the United Kingdom Prospective Diabetes Study (UKPDS) [ 25 ].

Duration of Diabetes

This risk factor cannot be avoided; it has been consistently illustrated that long-term diabetes increases the odds of developing DR. According to the LALES trial, the chance of developing DR increased by 8% for every year added to diabetes history. Prolonged exposure to the hyperglycemic condition can explain this link, which may raise the risk of vascular injury and other consequences such as DR [ 24 , 25 ].

Drug Use in Diabetes

Individuals with DR are at a higher likelihood of needing medications to manage their diabetes, such as oral hypoglycemics or insulin. In contrast to diabetes that is undiagnosed and untreated, diabetes that is known and managed is a predictor of DR. Ninety percent of diabetics without retinopathy who participated in an exhaustive study in the Chinese population were either untreated, under diet management, or taking oral hypoglycemic medications, according to the results [ 26 ]. In contrast, oral hypoglycemic medications or insulin injections were necessary for the diabetic management of nearly 80% of individuals with DR [ 27 ]. The extent and scale of blood glucose control achieved by patients may help to explain the correlation between insulin use and DR. 

Systemic risk factors and DR

Hypertension

Time and again, studies have shown that hypertension is in conjunction with the genesis of DR. In comparison to diabetic patients who are normotensive, the Hoorn study calculated that hypertensive individuals had a prospect of retinopathy occurrence after ten years that was more than twice as high [ 28 ]. The clinical observation that diabetes and hypertension usually co-exist may help to explain this clear correlation between hypertension and DR. Hard exudates, cotton-wool patches, and retinal haemorrhages are morphological aberrations in retinal vascular system bearing similarities to those observed in mild-to-moderate NPDR and can be brought on by hypertension. Patients with DR must have their blood pressure under control, according to the influential UKPDS 69 research. The ABCD Trial authors discovered that despite having excellent blood pressure control, both groups had poor glycaemic control, which may have contributed to the advancement of DR. This would underline the significance of glycemic management in DR even more [ 21 , 24 , 29 ].

Another risk factor that is frequently linked to cardiovascular disease is obesity. BMI, waist-hip ratio, and waist circumference can all be used to characterize it. Positive correlations exist between increased waist circumference and DR, as well as a higher waist-hip ratio [ 30 ]. Studies on the connection between BMI and DR have produced a range of conclusions. The primary risk factor for DR was shown to be obesity, defined as having a BMI of >30 kg/m 2 in a study that primarily focused on individuals with type 1 DM, even after accounting for other risk variables, including HbA1c and the use of cardioprotective medications [ 30 ]. A separate study found that obesity and retinopathy were associated with retinopathy patients being more likely to be fat [ 30 ]. However, after adjusting for confounding factors like blood pressure, this association was no longer present. Elevated BMI value, however, was discovered to be positively correlated to deteriorating retinal vasculature condition that endangers eyesight in retinopathy patients [ 30 ].

Hyperlipidemia

Various associations between high cholesterol and DR have been unveiled in studies. A greater frequency of diabetic macular oedema and vision-jeopardizing DR was shown to be related to raised total blood cholesterol, according to Yau et al. [ 30 ]. According to the Hoorn research, there is no correlation between total cholesterol levels and the prevalence of DR. However, it did show that raised serum lipid levels are linked to a higher prevalence of the hard exudates that are characteristic of NPDR [ 28 , 30 ]. Elevated lipid and serum cholesterol levels are widely understood manifestations of the metabolic syndrome. Routine primary care dyslipidemia screening for diabetes patients is now feasible due to the risk variables' well-established link with events involving the cardiovascular system and mortality, additionally, the favourable impacts of medication on these risk factors are also notable. Elevated blood lipid and cholesterol levels have been linked to a prolonged-term risk of loss of vision resulting from DR. One study found that those who experienced a persistent reduction in eyesight to 5/200 or worse had an initial cholesterol score of 244 as opposed to 228 in individuals who did not [ 31 ].

Current Standards and Practices

The most up-to-date protocols and standards for the development of DR were released by the International Council of Ophthalmology in 2018 as established recommendations for diabetic eye care. In the same year, a position paper on DR was also released by the American Diabetes Association. The American Diabetes Association suggests a certain initial eye exam at a particular time based on what sort of diabetes and reasonable (level B) evidence (Table  1 ). A retinal examination appropriate for detecting DR and vision screening would need to be included in the standard screening test in order to ensure proper referral to an ophthalmologist.

ADA: American Diabetes Association [ 32 ]

Opportunistic Versus Systematic Screening

Opportunistic screening happens occasionally and takes place when a patient requests a test from their physician or another healthcare provider. Not all individuals at risk may be included in opportunistic screening, and it may not be subjected to quality assurance checks. As previously said, systematic screening, on the other hand, is comprised of quality-assured preset screening procedures that involve actively identifying individuals who are at risk, keeping track of eligible subjects, and inviting them to participate in the screening programme. Every participant in the systematic screening process goes through the same screening process. The processes for invitation, selection, and follow-up are pre-established and constitute a system that provides invitations and/or notifications, as well as calls and reminders for screening at specific intervals [ 33 ].

Screening Methods

Different techniques have been used to screen for DR. This comprises wide-angle digital photography and direct ophthalmoscopy, as well as dilated stereoscopic fundoscopy, analogue fundus photography, and direct ophthalmoscopy. Many nations, including Iceland, the UK's ENSP, and France's OPHDIAT system (a telemedical network for DR screening), have implemented countrywide screening programmes. Using non-mydriatic cameras, technicians at satellite screening centres first take fundus photos for the OPHDIAT programme. After that, ophthalmologists receive these pictures via a telemedicine network for assessment. In the same manner, South India uses telescreening to detect DR in India. It entails taking 45° single-field digital fundus pictures, which are then sent digitally to retina experts for assessment. The people in the UK who grade DR go through extensive training provided by ophthalmologists, but they are not required to have medical expertise [ 33 ].

Future Perspectives of Screening

Currently, skilled specialists like ophthalmologists, optometrists, or particularly trained graders screen for DR. The use of automated grading is now being investigated due to its time-consuming nature and requirement for screening big populations. In this, DR lesions are found using a computer system that processes images and recognizes patterns. The neural network method and the digital image processing technology are two different ways to recognize patterns. For recognizing and counting the initial DR lesions, such as hard exudates, haemorrhages, cotton wool patches and microaneurysms, image processing is useful (Figure ​ (Figure4) 4 ) [ 33 ]. According to one study, DR was seen in 22.5% of the individuals in this cohort, ranging from 20% to 28%. Out of this group of patients with DR, 1.8% needed an urgent referral within 30 days because their DR was so severe, and 0.6% needed an urgent referral for non-DR-related causes. Additionally, 8.7% of people, needed ophthalmologic care within six months due to DR, and another 8.1% to 19.5% of the population needed care between 6 months and 1 year. In 23% of cases, incidental abnormalities were found throughout the screening process; most of these findings had to do with disorders such as cataracts and dry macular degeneration. In 0.6% of the screened eyes, there were incidental discoveries that were urgent or clinically significant [ 34 ]. According to one study, the cost analysis reveals that telemedicine-based DR screening is less expensive than traditional retinal examinations, which costs $49.95 and $77.80 respectively. This strategy, which makes use of digital retinal imaging and telemedicine, maybe a more practical and accessible option, especially for underprivileged and distant communities. It is imperative to encourage the broad use of telemedicine in light of the rising incidence of diabetes mellitus and DR in the US and around the world. For diabetic patients who require immediate access to eye care specialists, this can retain affordability while also greatly improving access to care and enhancing adherence to annual examinations [ 35 ].

Treatment strategies

Pharmacotherapy

As intravitreous glucocorticoids have anti-angiogenic and anti-inflammatory qualities, they are mostly used to treat diabetic macular oedema. These benefits also extend to proliferative DR, since they result in the stabilisation of the inner blood-retina barrier. Despite the absence of clinical trial data, triamcinolone acetonide is becoming widely used off-label to treat diabetic macular oedema, demonstrating its therapeutic effectiveness against the condition. Different dosages are given, varying from 4 to 25 mg, but the disadvantage is that the outcome just temporarily lasts for three months and repeated injections are required. Furthermore, about one-third of individuals might get secondary glaucoma. Dexamethasone is therefore seen as an alternate course of therapy, but with the potential adverse effect of secondary cataracts. Vascular endothelial growth factor (VEGF) plays a role in the degradation of the inner blood-retinal barrier and vascular leakage. VEGF inhibitors can inhibit the proliferation and leakage in cases of diabetic macular oedema. They are well-known for their efficacy in treating wet age-related macular degeneration. However, the duration of their impact is brief, ranging from four to six weeks. The effects of pegaptanib (an aptamer), ranibizumab (a recombinant, humanised monoclonal antibody fragment), and bevacizumab (a humanised monoclonal antibody) are now being studied in prospective multicenter studies. In individuals with diabetic macular oedema receiving treatment, randomised, double-blind research conducted in 2005 showed a substantial increase in visual acuity and a decrease in retinal thickness. There are difficulties in using ranibizumab and pegaptanib off-label because they are only approved for the treatment of wet age-related macular degeneration. Additionally, these drugs are very pricey; each injection of ranibizumab, for example, costs 1300 euros. By the end of 2011, pegaptinib and ranibizumab should be authorized for the treatment of diabetic macular oedema. In addition, research is being done on VEGF Trap Eye, a recombinant protein with an extended half-life when contrasted with ranibizumab. The USA has already approved Ozurdex, an injectable glucocorticoid with a sustained impact that lasts up to 12 months, regarding the treatment of central retinal vascular occlusion (CRVO). Additionally, investigations are underway to assess its applicability in diabetic macular oedema treatment. A range of problems, such as endophthalmitis, retinal detachment, and lens damage, are associated with intravitreous injections, sometimes referred to as intravitreous operational medication (IVOM). The complication rate is still far lower than 1%, though. IVOM operations should be carried out in aseptic operating rooms to reduce the risk of infection. In controlled research, other therapy options employing oral or injectable drugs such as somatostatin analogues (octreotide) or protein kinase C inhibitors (ruboxistaurin ) have not produced the expected outcomes. A randomised controlled trial (CALDIRET) with 635 originally recruited participants found that calcium dobesilate, when taken orally for vascular problems including venous insufficiency, could not prevent clinically significant macular oedema in people with type 2 diabetes. Only through posthoc subgroup analysis, that is, those females who had poorly managed hypertension and HbA1c levels above 9%-was a protective effect seen [ 36 ].

Laser Photocoagulation

Based on data from the prospective, randomised, controlled Early therapy Diabetic Retinopathy (ETDR) trial, which included 3,711 patients overall and was published in 1991, laser photocoagulation is the established intervention for DR and diabetic macular oedema. As a result, the Working Group on Diabetes and the Eye (Arbeitsgemeinschaft Diabetes und Auge, AGDA) and the Initiative Group for the Early Detection of Diabetic Eye Diseases (IFdA) have recommended treatments, and Germany has released national guidelines in this regard. Usually operating at a wavelength of 532 nm, the double-frequency neodymium:yttrium-aluminium-garnet (Nd:YAG) laser produces the laser utilised in this therapy. The therapy is administered through the support of contact lens applied to the cornea and a split-lamp microscope that is connected to the laser. However, Nd:YAG laser therapy might not be possible if extensive opacification of the cornea or lens obscures vision and disperses the healing light beam. In these situations, an 810 nm diode laser can be used, or the cataract can be treated initially and then a few days later, the laser can be used. When used for proliferative DR, peripheral laser photocoagulation restores the partial oxygen pressure in the non-vascular regions of the retina to normal levels, which attempts to regress newly created capillaries. Thus, there is a decreased chance of membrane development and vitreous haemorrhage. Usually, the whole surface of the retina is covered by up to 2,500 laser foci, each with a 500 μm diameter that are dispersed over the periphery but leave the centre intact. A prospective, randomised research with 1,732 eyes. The Diabetic Retinopathy Study (DRS) demonstrated that this drug decreased the risk of severe vision loss by more than 50% in 1976. 56 untreated eyes and 129 treated eyes both experienced severe vision loss. Targeted focused laser coagulation is used to close leaky microaneurysms and capillaries surrounding the fovea in cases of clinically severe diabetic macular oedema. In this instance, the laser focus sizes differed from 100 to 200 micrometers. The ETDR trial from 1985 showed a substantial reduction in the likelihood of impaired vision because of severe macular oedema. The study comprised 754 eyes that underwent targeted laser coagulation and 1,490 eyes in an untreated control group. Following rapid diagnosis and treatment, laser coagulation was promptly provided to the control group as a result of the study's findings, and this continues to be the gold standard for individuals with clinically severe macular oedema. It's crucial to remember that laser therapy seldom results in better visual acuity. Because declining visual acuity is sometimes irreversible, early diagnosis through preventative screenings is essential to maintaining acuity when the eye still has adequate vision [ 36 ].

In the case of non-resolving vitreous haemorrhage, subhyaloid haemorrhage, tractional macular oedema, ghost-cell glaucoma and tractional retinal detachment, pars plana vitrectomy (PPV) is advised. With PPV, the retina may be repositioned, scar tissue, membranes, and hazy vitreous material can be removed. An ideal laser photocoagulation therapy can also be used. A prospective, randomised, controlled trial known as the Diabetic Retinopathy Vitrectomy trial (DRVS) demonstrated the efficacy of PPV and indicated the best time to do this treatment. Individuals who had their vitreous removed early on had far better eyesight than those who had the surgery done a year later. Recent years have seen the routine and increased efficiency of vitrectomy as a result of developments in microsurgical methods. Advancements in technology have shortened the duration of surgery and done away with the requirement for sutures. The sizes of the tools used for vitrectomy have similarly shrunk, from 1.0 to 0.6 mm. Thus, at least minimal vision is currently maintained in individuals having advanced proliferative DR. Surgical excision of the non-functioning eye may be explored as a last-ditch method of pain relief in severe situations [ 36 ].

The articles included in the review are shown in Table ​ Table2 2 .

AER: Albumin excretion rate; GV: Glycemic variability; LADA: Latent autoimmune diabetes of adults; DBP: Diastolic blood pressure; DR: Diabetic retinopathy; Hcy: Homocysteine; CRP: C-reactive protein; CTGF: Connective tissue growth factor; PDR: Proliferative diabetic retinopathy; VEGF: Vascular endothelial growth factor; BMI: body mass index

Conclusions

In this comprehensive review article, the authors conducted an extensive search across multiple databases, summarizing key findings and developments in the understanding and treatment of DR over the preceding years. They covered a wide range of critical aspects related to the epidemiology, pathophysiology, systemic risk factors, screening methods, and treatment strategies for DR. The epidemiological data presented highlights the alarming rise in diabetes cases globally, with a particular concentration in regions like Africa, the Middle East, North America, and the Caribbean. The prevalence of DR varies significantly across different populations, with notable disparities between genders and types of diabetes. Furthermore, DR is not only a vision-threatening condition but also a predictor of overall health outcomes. The pathophysiology section elucidates the intricate mechanisms underlying DR, emphasizing the role of hyperglycemia in retinal microvascular damage. The review delves into various metabolic processes associated with hyperglycemia-induced vascular injury, leading to alterations in blood flow, pericyte loss, endothelial cell apoptosis, basement membrane thickening, and blood-retina barrier impairment. The article also discusses emerging biomarkers, such as advanced glycoxidation products, C-reactive protein, and neuroglobin, which offer potential diagnostic and prognostic value in DR. Pro-angiogenic cytokines and VEGF are identified as central contributors to DR neovascularization. The influence of glycemic control, duration of diabetes, drug use, hypertension, obesity, and hyperlipidemia on DR is comprehensively examined. The review underscores the significance of managing these risk factors to prevent the onset and progression of DR. The authors provide valuable insights into the current standards and practices for DR screening, comparing opportunistic and systematic screening approaches. They also discuss the various screening methods, including telemedicine-based solutions, highlighting their cost-effectiveness and potential to improve access to care. The treatment strategies section explores pharmacotherapy, laser photocoagulation, and surgery as viable options for managing DR. Intravitreous glucocorticoids and anti-VEGF agents are discussed as key interventions for diabetic macular oedema, while laser photocoagulation remains the gold standard for the treatment of DR. Pars plana vitrectomy is considered for cases of non-resolving vitreous haemorrhage and retinal detachment. In conclusion, this comprehensive review article provides a thorough overview of the current state of knowledge regarding DR. It emphasizes the importance of early screening, risk factor management, and advances in treatment modalities. The article underscores the need for ongoing research and innovation to address the growing global burden of DR and its significant impact on public health.

The authors have declared that no competing interests exist.

Diabetic retinopathy

Showing results 1 - 10 of 88

Sorted by most recent

21 March 2024

12 February 2024

30 January 2024

24 October 2023

16 October 2023

13 September 2023

4 August 2023

24 April 2023

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Review Article
• Published: 19 January 2021

Current understanding of the molecular and cellular pathology of diabetic retinopathy

• David A. Antonetti   ORCID: orcid.org/0000-0003-1130-6577 1 ,
• Paolo S. Silva   ORCID: orcid.org/0000-0001-8024-9348 2 , 3 &
• Alan W. Stitt 4  

Nature Reviews Endocrinology volume  17 ,  pages 195–206 ( 2021 ) Cite this article

7775 Accesses

220 Citations

21 Altmetric

Metrics details

• Cell signalling
• Diabetes complications

Diabetes mellitus has profound effects on multiple organ systems; however, the loss of vision caused by diabetic retinopathy might be one of the most impactful in a patient’s life. The retina is a highly metabolically active tissue that requires a complex interaction of cells, spanning light sensing photoreceptors to neurons that transfer the electrochemical signal to the brain with support by glia and vascular tissue. Neuronal function depends on a complex inter-dependency of retinal cells that includes the formation of a blood–retinal barrier. This dynamic system is negatively affected by diabetes mellitus, which alters normal cell–cell interactions and leads to profound vascular abnormalities, loss of the blood–retinal barrier and impaired neuronal function. Understanding the normal cell signalling interactions and how they are altered by diabetes mellitus has already led to novel therapies that have improved visual outcomes in many patients. Research highlighted in this Review has led to a new understanding of retinal pathophysiology during diabetes mellitus and has uncovered potential new therapeutic avenues to treat this debilitating disease.

Diabetic retinopathy is a leading cause of blindness that disrupts the normal interaction of the retinal neural and vascular components leading to vascular permeability, neovascularization and loss of proper neural function.

Current effective therapeutic approaches target vascular endothelial growth factor, while a host of new therapies targeting vascular endothelial and pericyte signalling and inflammatory cytokines are being tested for diabetic retinopathy.

Stem cell therapy for vascular regeneration holds potential for restorative therapeutic approaches in diabetic retinopathy.

Understanding the neuronal and glial changes that drive loss of vision is rapidly emerging, and targeted approaches to directly test the relationship between the neurovascular unit and alteration in diabetic retinopathy are needed.

This is a preview of subscription content, access via your institution

Access options

Access Nature and 54 other Nature Portfolio journals

Get Nature+, our best-value online-access subscription

24,99 € / 30 days

cancel any time

Subscribe to this journal

Receive 12 print issues and online access

195,33 € per year

only 16,28 € per issue

Buy this article

• Purchase on Springer Link
• Instant access to full article PDF

Prices may be subject to local taxes which are calculated during checkout

Similar content being viewed by others

The role of microglia in the development of diabetic retinopathy

Perspectives of diabetic retinopathy—challenges and opportunities

Retinal non-perfusion in diabetic retinopathy

Ogurtsova, K. et al. IDF diabetes atlas: global estimates for the prevalence of diabetes for 2015 and 2040. Diabetes Res. Clin. Pract. 128 , 40–50 (2017).

CAS   PubMed   Google Scholar  

Ajlan, R. S., Silva, P. S. & Sun, J. K. Vascular endothelial growth factor and diabetic retinal disease. Semin. Ophthalmol. 31 , 40–48 (2016).

PubMed   Google Scholar  

Li, A. S., Veerappan, M., Mittal, V. & Do, D. V. Anti-VEGF agents in the management of diabetic macular edema. Expert Rev. Ophthalmol. 15 , 285–296 (2020).

CAS   Google Scholar  

[No authors listed]. Grading diabetic retinopathy from stereoscopic color fundus photographs–an extension of the modified Airlie House classification. ETDRS report number 10. Early Treatment Diabetic Retinopathy Study Research Group. Ophthalmology 98 , 786–806 (1991).

Google Scholar  

Solomon, S. D. et al. Diabetic retinopathy: a position statement by the American Diabetes Association. Diabetes Care 40 , 412–418 (2017).

PubMed   PubMed Central   Google Scholar  

Kohner, E. M., Patel, V. & Rassam, S. M. Role of blood flow and impaired autoregulation in the pathogenesis of diabetic retinopathy. Diabetes 44 , 603–607 (1995).

Roy, S., Ha, J., Trudeau, K. & Beglova, E. Vascular basement membrane thickening in diabetic retinopathy. Curr. Eye Res. 35 , 1045–1056 (2010).

Enge, M. et al. Endothelium-specific platelet-derived growth factor-B ablation mimics diabetic retinopathy. EMBO J. 21 , 4307–4316 (2002).

CAS   PubMed   PubMed Central   Google Scholar  

[No authors listed]. Fundus photographic risk factors for progression of diabetic retinopathy. ETDRS report number 12. Early Treatment Diabetic Retinopathy Study Research Group. Ophthalmology 98 , 823–833 (1991).

Klein, R. et al. The Wisconsin Epidemiologic Study of Diabetic Retinopathy XXII. The twenty-five-year progression of retinopathy in persons with type 1 diabetes. Ophthalmology 115 , 1859–1868 (2008).

Browning, D. J. et al. Relationship between optical coherence tomography-measured central retinal thickness and visual acuity in diabetic macular edema. Ophthalmology 114 , 525–536 (2007).

Gardner, T. W., Larsen, M., Girach, A. & Zhi, X. Diabetic macular oedema and visual loss: relationship to location, severity and duration. Acta Ophthalmol. 87 , 709–713 (2009).

[No authors listed]. Photocoagulation for diabetic macular edema. Early treatment diabetic retinopathy study report number 1. Early Treatment Diabetic Retinopathy Study Research Group. Arch. Ophthalmol. 103 , 1796–1806 (1985).

Cunha-Vaz, J. Mechanisms of retinal fluid accumulation and blood-retinal barrier breakdown. Dev. Ophthalmol. 58 , 11–20 (2017).

Samuels, I. S., Bell, B. A., Pereira, A., Saxon, J. & Peachey, N. S. Early retinal pigment epithelium dysfunction is concomitant with hyperglycemia in mouse models of type 1 and type 2 diabetes. J. Neurophysiol. 113 , 1085–1099 (2015).

Bentley, K. et al. The role of differential VE-cadherin dynamics in cell rearrangement during angiogenesis. Nat. Cell Biol. 16 , 309–321 (2014).

Desjardins, D. M. et al. Progressive early breakdown of retinal pigment epithelium function in hyperglycemic rats. Invest. Ophthalmol. Vis. Sci. 57 , 2706–2713 (2016).

Lutty, G. A. Effects of diabetes on the eye. Invest. Ophthalmol. Vis. Sci. 54 , ORSF81–ORSF87 (2013).

Wilkinson, C. P. et al. Proposed international clinical diabetic retinopathy and diabetic macular edema disease severity scales. Ophthalmology 110 , 1677–1682 (2003).

Abramoff, M. D. et al. Approach for a clinically useful comprehensive classification of vascular and neural aspects of diabetic retinal disease. Invest. Ophthalmol. Vis. Sci. 59 , 519–527 (2018).

Klein, B. E. Overview of epidemiologic studies of diabetic retinopathy. Ophthalmic Epidemiol. 14 , 179–183 (2007).

Klein, R., Klein, B. E., Moss, S. E. & Cruickshanks, K. J. The Wisconsin Epidemiologic Study of Diabetic Retinopathy. XV. The long-term incidence of macular edema. Ophthalmology 102 , 7–16 (1995).

Duh, E. J., Sun, J. K. & Stitt, A. W. Diabetic retinopathy: current understanding, mechanisms, and treatment strategies. JCI Insight 2 , e93751 (2017).

PubMed Central   Google Scholar  

Klein, R. in Diabetic Retinopathy Ch. 3 (ed. Duh, E.) 67–107 (Humana, 2008).

Kady, N. M. et al. ELOVL4-mediated production of very long-chain ceramides stabilizes tight junctions and prevents diabetes-induced retinal vascular permeability. Diabetes 67 , 769–781 (2018).

Tikhonenko, M. et al. Remodeling of retinal fatty acids in an animal model of diabetes: a decrease in long-chain polyunsaturated fatty acids is associated with a decrease in fatty acid elongases Elovl2 and Elovl4. Diabetes 59 , 219–227 (2010).

Silva, P. S. et al. Peripheral lesions identified by mydriatic ultrawide field imaging: distribution and potential impact on diabetic retinopathy severity. Ophthalmology 120 , 2587–2595 (2013).

Rasmussen, M. L. et al. Comparison between early treatment diabetic retinopathy study 7-field retinal photos and non-mydriatic, mydriatic and mydriatic steered widefield scanning laser ophthalmoscopy for assessment of diabetic retinopathy. J. Diabetes Complications 29 , 99–104 (2015).

Silva, P. S. et al. Peripheral lesions identified on ultrawide field imaging predict increased risk of diabetic retinopathy progression over 4 years. Ophthalmology 122 , 949–956 (2015).

Talks, S. J., Manjunath, V., Steel, D. H., Peto, T. & Taylor, R. New vessels detected on wide-field imaging compared to two-field and seven-field imaging: implications for diabetic retinopathy screening image analysis. Br. J. Ophthalmol. 99 , 1606–1609 (2015).

Aiello, L. P. et al. Comparison of early treatment diabetic retinopathy study standard 7-field imaging with ultrawide-field imaging for determining severity of diabetic retinopathy. JAMA Ophthalmol. 137 , 65–73 (2019).

Wessel, M. M. et al. Ultra-wide-field angiography improves the detection and classification of diabetic retinopathy. Retina 32 , 785–791 (2012).

Silva, P. S. et al. Diabetic retinopathy severity and peripheral lesions are associated with nonperfusion on ultrawide field angiography. Ophthalmology 122 , 2465–2472 (2015).

Korot, E., Comer, G., Steffens, T. & Antonetti, D. A. Algorithm for the measure of vitreous hyperreflective foci in optical coherence tomographic scans of patients with diabetic macular edema. JAMA Ophthalmol. 134 , 15–20 (2016).

Jia, Y. et al. Split-spectrum amplitude-decorrelation angiography with optical coherence tomography. Opt. Express 20 , 4710–4725 (2012).

Vujosevic, S. et al. Early microvascular and neural changes in patients with type 1 and type 2 diabetes mellitus without clinical signs of diabetic retinopathy. Retina 39 , 435–445 (2019).

Spaide, R. F., Fujimoto, J. G., Waheed, N. K., Sadda, S. R. & Staurenghi, G. Optical coherence tomography angiography. Prog. Retin. Eye Res. 64 , 1–55 (2018).

Haines, N. R., Manoharan, N., Olson, J. L., D’Alessandro, A. & Reisz, J. A. Metabolomics analysis of human vitreous in diabetic retinopathy and rhegmatogenous retinal detachment. J. Proteome Res. 17 , 2421–2427 (2018).

Liew, G. et al. Metabolomics of diabetic retinopathy. Curr. Diab. Rep. 17 , 102 (2017).

Moss, S. E., Klein, R. & Klein, B. E. The 14-year incidence of visual loss in a diabetic population. Ophthalmology 105 , 998–1003 (1998).

Klein, R., Klein, B. E., Moss, S. E. & Cruickshanks, K. J. The Wisconsin Epidemiologic Study of Diabetic Retinopathy: XVII. The 14-year incidence and progression of diabetic retinopathy and associated risk factors in type 1 diabetes. Ophthalmology 105 , 1801–1815 (1998).

Mishra, A. & Newman, E. A. Inhibition of inducible nitric oxide synthase reverses the loss of functional hyperemia in diabetic retinopathy. Glia 58 , 1996–2004 (2010).

Nguyen, T. T. et al. Correlation of light-flicker-induced retinal vasodilation and retinal vascular caliber measurements in diabetes. Invest. Ophthalmol. Vis. Sci. 50 , 5609–5613 (2009).

Pannicke, T. et al. Diabetes alters osmotic swelling characteristics and membrane conductance of glial cells in rat retina. Diabetes 55 , 633–639 (2006).

Thompson, K. et al. Advanced glycation end (AGE) product modification of laminin downregulates Kir4.1 in retinal Muller cells. PLoS ONE 13 , e0193280 (2018).

McDowell, R. E. et al. Muller glial dysfunction during diabetic retinopathy in rats is reduced by the acrolein-scavenging drug, 2-hydrazino-4,6-dimethylpyrimidine. Diabetologia 61 , 2654–2667 (2018).

Portillo, J. C. et al. CD40 in retinal Muller cells induces P2X7-dependent cytokine expression in macrophages/microglia in diabetic mice and development of early experimental diabetic retinopathy. Diabetes 66 , 483–493 (2017).

Karlstetter, M. et al. Retinal microglia: just bystander or target for therapy? Prog. Retin. Eye Res. 45 , 30–57 (2015).

Mohamed, Q., Gillies, M. C. & Wong, T. Y. Management of diabetic retinopathy: a systematic review. JAMA 298 , 902–916 (2007).

Simo, R., Sundstrom, J. M. & Antonetti, D. A. Ocular anti-VEGF therapy for diabetic retinopathy: the role of VEGF in the pathogenesis of diabetic retinopathy. Diabetes Care 37 , 893–899 (2014).

Titchenell, P. M. & Antonetti, D. A. Using the past to inform the future: anti-VEGF therapy as a road map to develop novel therapies for diabetic retinopathy. Diabetes 62 , 1808–1815 (2013).

Nguyen, Q. D. et al. Ranibizumab for diabetic macular edema: results from 2 phase III randomized trials: RISE and RIDE. Ophthalmology 119 , 789–801 (2012).

Do, D. V. et al. One-year outcomes of the DA VINCI study of VEGF trap-eye in eyes with diabetic macular edema. Ophthalmology 119 , 1658–1665 (2012).

Heier, J. S. et al. Intravitreal aflibercept for diabetic macular edema: 148-week results from the VISTA and VIVID studies. Ophthalmology 123 , 2376–2385 (2016).

Sun, J. K. et al. Rationale and application of the protocol S anti-vascular endothelial growth factor algorithm for proliferative diabetic retinopathy. Ophthalmology 126 , 87–95 (2019).

Wells, J. A. et al. Aflibercept, bevacizumab, or ranibizumab for diabetic macular edema. N. Engl. J. Med. 372 , 1193–1203 (2015).

Arima, M. et al. Claudin-5 redistribution induced by inflammation leads to anti-VEGF-resistant diabetic macular edema. Diabetes 69 , 981–999 (2020).

Potente, M., Gerhardt, H. & Carmeliet, P. Basic and therapeutic aspects of angiogenesis. Cell 146 , 873–887 (2011).

Miloudi, K. et al. NOTCH1 signaling induces pathological vascular permeability in diabetic retinopathy. Proc. Natl Acad. Sci. USA 116 , 4538–4547 (2019).

Daneman, R., Zhou, L., Kebede, A. A. & Barres, B. A. Pericytes are required for blood-brain barrier integrity during embryogenesis. Nature 468 , 562–566 (2010).

Lindblom, P. et al. Endothelial PDGF-B retention is required for proper investment of pericytes in the microvessel wall. Genes Dev. 17 , 1835–1840 (2003).

Park, D. Y. et al. Plastic roles of pericytes in the blood-retinal barrier. Nat. Commun. 8 , 15296 (2017).

Geraldes, P. et al. Activation of PKC-delta and SHP-1 by hyperglycemia causes vascular cell apoptosis and diabetic retinopathy. Nat. Med. 15 , 1298–1306 (2009).

Geraldes, P. & King, G. L. Activation of protein kinase C isoforms and its impact on diabetic complications. Circ. Res. 106 , 1319–1331 (2010).

Xu, Q. et al. Vascular development in the retina and inner ear: control by Norrin and Frizzled-4, a high-affinity ligand-receptor pair. Cell 116 , 883–895 (2004).

Ye, X. et al. Norrin, frizzled-4, and Lrp5 signaling in endothelial cells controls a genetic program for retinal vascularization. Cell 139 , 285–298 (2009).

Wang, Y. et al. Norrin/Frizzled4 signaling in retinal vascular development and blood brain barrier plasticity. Cell 151 , 1332–1344 (2012).

Zhou, Y. et al. Canonical WNT signaling components in vascular development and barrier formation. J. Clin. Invest. 124 , 3825–3846 (2014).

Tokunaga, C. C., Chen, Y. H., Dailey, W., Cheng, M. & Drenser, K. A. Retinal vascular rescue of oxygen-induced retinopathy in mice by norrin. Invest. Ophthalmol. Vis. Sci. 54 , 222–229 (2013).

Ohlmann, A. et al. Norrin promotes vascular regrowth after oxygen-induced retinal vessel loss and suppresses retinopathy in mice. J. Neurosci. 30 , 183–193 (2010).

Zeilbeck, L. F. et al. Norrin mediates angiogenic properties via the induction of insulin-like growth factor-1. Exp. Eye Res. 145 , 317–326 (2016).

Diaz-Coranguez, M., Lin, C. M., Liebner, S. & Antonetti, D. A. Norrin restores blood-retinal barrier properties after vascular endothelial growth factor-induced permeability. J. Biol. Chem. 295 , 4647–4660 (2020).

Simo-Servat, O., Hernandez, C. & Simo, R. Usefulness of the vitreous fluid analysis in the translational research of diabetic retinopathy. Mediators Inflamm. 2012 , 872978 (2012).

Vujosevic, S. & Simo, R. Local and systemic inflammatory biomarkers of diabetic retinopathy: an integrative approach. Invest. Ophthalmol. Vis. Sci. 58 , BIO68–BIO75 (2017).

Joussen, A. M. et al. Nonsteroidal anti-inflammatory drugs prevent early diabetic retinopathy via TNF-α suppression. FASEB J. 16 , 438–440 (2002).

Huang, H. et al. TNFα is required for late BRB breakdown in diabetic retinopathy, and its inhibition prevents leukostasis and protects vessels and neurons from apoptosis. Invest. Ophthalmol. Vis. Sci. 52 , 1336–1344 (2011).

Joussen, A. M. et al. A central role for inflammation in the pathogenesis of diabetic retinopathy. FASEB J. 18 , 1450–1452 (2004).

Demircan, N., Safran, B. G., Soylu, M., Ozcan, A. A. & Sizmaz, S. Determination of vitreous interleukin-1 (IL-1) and tumour necrosis factor (TNF) levels in proliferative diabetic retinopathy. Eye 20 , 1366–1369 (2005).

Koleva-Georgieva, D. N., Sivkova, N. P. & Terzieva, D. Serum inflammatory cytokines IL-1β, IL-6, TNF-α and VEGF have influence on the development of diabetic retinopathy. Folia Med. 53 , 44–50 (2011).

Schoenberger, S. D. et al. Increased prostaglandin E2 (PGE2) levels in proliferative diabetic retinopathy and correlation with VEGF and inflammatory cytokines. Invest. Ophthalmol. Vis. Sci. 53 , 5906–5911 (2012).

Yoshimura, T. et al. Comprehensive analysis of inflammatory immune mediators in vitreoretinal diseases. PLoS ONE 4 , e8158 (2009).

Ogata, N. et al. Pigment epithelium-derived factor in the vitreous is low in diabetic retinopathy and high in rhegmatogenous retinal detachment. Am. J. Ophthalmol. 132 , 378–382 (2001).

Aveleira, C. A., Lin, C. M., Abcouwer, S. F., Ambrosio, A. F. & Antonetti, D. A. TNF-α signals through PKCζ/NF-κB to alter the tight junction complex and increase retinal endothelial cell permeability. Diabetes 59 , 2872–2882 (2010).

Titchenell, P. M. et al. Novel atypical PKC inhibitors prevent vascular endothelial growth factor-induced blood-retinal barrier dysfunction. Biochem. J. 446 , 455–467 (2012).

Lin, C. M. et al. Inhibition of atypical protein kinase C reduces inflammation-induced retinal vascular permeability. Am. J. Pathol. 188 , 2392–2405 (2018).

Gardner, T. W. & Sundstrom, J. M. A proposal for early and personalized treatment of diabetic retinopathy based on clinical pathophysiology and molecular phenotyping. Vis. Res. 139 , 153–160 (2017).

Sun, J. K. et al. Protection from retinopathy and other complications in patients with type 1 diabetes of extreme duration: the Joslin 50-Year Medalist Study. Diabetes Care 34 , 968–974 (2011).

Praidou, A. et al. Angiogenic growth factors and their inhibitors in diabetic retinopathy. Curr. Diabetes Rev. 6 , 304–312 (2010).

Yokomizo, H. et al. Retinol binding protein 3 is increased in the retina of patients with diabetes resistant to diabetic retinopathy. Sci. Transl. Med. 11 , eaau6627 (2019).

Garcia-Ramirez, M. et al. Interphotoreceptor retinoid-binding protein (IRBP) is downregulated at early stages of diabetic retinopathy. Diabetologia 52 , 2633–2641 (2009).

Gao, B. B. et al. Extracellular carbonic anhydrase mediates hemorrhagic retinal and cerebral vascular permeability through prekallikrein activation. Nat. Med. 13 , 181–188 (2007).

Clermont, A. et al. Plasma kallikrein mediates retinal vascular dysfunction and induces retinal thickening in diabetic rats. Diabetes 60 , 1590–1598 (2011).

Sodhi, A. et al. Angiopoietin-like 4 binds neuropilins and cooperates with VEGF to induce diabetic macular edema. J. Clin. Invest. 129 , 4593–4608 (2019).

Xin, X. et al. Hypoxic retinal Muller cells promote vascular permeability by HIF-1-dependent up-regulation of angiopoietin-like 4. Proc. Natl Acad. Sci. USA 110 , E3425–E3434 (2013).

Bouleti, C. et al. Protective effects of angiopoietin-like 4 on cerebrovascular and functional damages in ischaemic stroke. Eur. Heart J. 34 , 3657–3668 (2013).

Cho, D. I. et al. Antiinflammatory activity of ANGPTL4 facilitates macrophage polarization to induce cardiac repair. JCI Insight 4 , e125437 (2019).

Cerani, A. et al. Neuron-derived semaphorin 3A is an early inducer of vascular permeability in diabetic retinopathy via neuropilin-1. Cell Metab. 18 , 505–518 (2013).

Wang, X. et al. LRG1 promotes angiogenesis by modulating endothelial TGF-β signalling. Nature 499 , 306–311 (2013).

Kallenberg, D. et al. A humanized antibody against LRG1 that inhibits angiogenesis and reduces retinal vascular leakage. Preprint at bioRxiv https://doi.org/10.1101/2020.07.25.218149 (2020).

Article   Google Scholar  

Duraisamy, A. J., Mishra, M., Kowluru, A. & Kowluru, R. A. Epigenetics and regulation of oxidative stress in diabetic retinopathy. Invest. Ophthalmol. Vis. Sci. 59 , 4831–4840 (2018).

Kowluru, R. A. Diabetic retinopathy, metabolic memory and epigenetic modifications. Vis. Res. 139 , 30–38 (2017).

Barber, A. J. et al. Neural apoptosis in the retina during experimental and human diabetes. Early onset and effect of insulin. J. Clin. Invest. 102 , 783–791 (1998).

Sohn, E. H. et al. Retinal neurodegeneration may precede microvascular changes characteristic of diabetic retinopathy in diabetes mellitus. Proc. Natl Acad. Sci. USA 113 , E2655–E2664 (2016).

Gardner, T. W., Abcouwer, S. F., Barber, A. J. & Jackson, G. R. An integrated approach to diabetic retinopathy research. Arch. Ophthalmol. 129 , 230–235 (2011).

Lynch, S. K. & Abramoff, M. D. Diabetic retinopathy is a neurodegenerative disorder. Vis. Res. 139 , 101–107 (2017).

Miller, W. P. et al. Deletion of the Akt/mTORC1 repressor REDD1 prevents visual dysfunction in a rodent model of type 1 diabetes. Diabetes 67 , 110–119 (2018).

Bogdanov, P. et al. Topical administration of bosentan prevents retinal neurodegeneration in experimental diabetes. Int. J. Mol. Sci. 19 , 3578 (2018).

Santos, A. R. et al. Functional and structural findings of neurodegeneration in early stages of diabetic retinopathy: cross-sectional analyses of baseline data of the EUROCONDOR project. Diabetes 66 , 2503–2510 (2017).

Biessels, G. J., Staekenborg, S., Brunner, E., Brayne, C. & Scheltens, P. Risk of dementia in diabetes mellitus: a systematic review. Lancet Neurol. 5 , 64–74 (2006).

Kopf, D. & Frolich, L. Risk of incident Alzheimer’s disease in diabetic patients: a systematic review of prospective trials. J. Alzheimers Dis. 16 , 677–685 (2009).

Cheung, C. Y., Ikram, M. K., Chen, C. & Wong, T. Y. Imaging retina to study dementia and stroke. Prog. Retin. Eye Res. 57 , 89–107 (2017).

Sundstrom, J. M. et al. Proteomic analysis of early diabetic retinopathy reveals mediators of neurodegenerative brain diseases. Invest. Ophthalmol. Vis. Sci. 59 , 2264–2274 (2018).

Stitt, A. W. et al. The progress in understanding and treatment of diabetic retinopathy. Prog. Retin. Eye Res. 51 , 156–186 (2016).

Wei, Y. et al. Nrf2 in ischemic neurons promotes retinal vascular regeneration through regulation of semaphorin 6A. Proc. Natl Acad. Sci. USA 112 , E6927–E6936 (2015).

Xu, Z. et al. NRF2 plays a protective role in diabetic retinopathy in mice. Diabetologia 57 , 204–213 (2014).

Xu, Z. et al. Neuroprotective role of Nrf2 for retinal ganglion cells in ischemia-reperfusion. J. Neurochem. 133 , 233–241 (2015).

Abcouwer, S. F. et al. Effects of ischemic preconditioning and bevacizumab on apoptosis and vascular permeability following retinal ischemia-reperfusion injury. Invest. Ophthalmol. Vis. Sci. 51 , 5920–5933 (2010).

Muthusamy, A. et al. Ischemia-reperfusion injury induces occludin phosphorylation/ubiquitination and retinal vascular permeability in a VEGFR-2-dependent manner. J. Cereb. Blood Flow. Metab. 34 , 522–531 (2014).

Hui, Q. et al. Inhibition of the Keap1-Nrf2 protein-protein interaction protects retinal cells and ameliorates retinal ischemia-reperfusion injury. Free Radic. Biol. Med. 146 , 181–188 (2020).

Stitt, A. W. et al. Vascular stem cells and ischaemic retinopathies. Prog. Retin. Eye Res. 30 , 149–166 (2011).

Mohan, R. & Kohner, E. M. Retinal revascularisation in diabetic retinopathy. Br. J. Ophthalmol. 70 , 114–117 (1986).

Takahashi, K., Kishi, S., Muraoka, K. & Shimizu, K. Reperfusion of occluded capillary beds in diabetic retinopathy. Am. J. Ophthalmol. 126 , 791–797 (1998).

Abaci, A. et al. Effect of diabetes mellitus on formation of coronary collateral vessels. Circulation 99 , 2239–2242 (1999).

Hernandez, S. L. et al. Characterization of circulating and endothelial progenitor cells in patients with extreme-duration type 1 diabetes. Diabetes Care 37 , 2193–2201 (2014).

Altabas, V. Diabetes, endothelial dysfunction, and vascular repair: what should a diabetologist keep his eye on? Int. J. Endocrinol. 2015 , 848272 (2015).

Patel, J. et al. Functional definition of progenitors versus mature endothelial cells reveals key SoxF-dependent differentiation process. Circulation 135 , 786–805 (2017).

Naito, H. et al. Endothelial side population cells contribute to tumor angiogenesis and antiangiogenic drug resistance. Cancer Res. 76 , 3200–3210 (2016).

Iba, T. et al. Isolation of tissue-resident endothelial stem cells and their use in regenerative medicine. Inflamm. Regen. 39 , 9 (2019).

Ingram, D. A. et al. Vessel wall-derived endothelial cells rapidly proliferate because they contain a complete hierarchy of endothelial progenitor cells. Blood 105 , 2783–2786 (2005).

Kovacic, J. C. & Boehm, M. Resident vascular progenitor cells: an emerging role for non-terminally differentiated vessel-resident cells in vascular biology. Stem Cell Res. 2 , 2–15 (2009).

Naito, H., Kidoya, H., Sakimoto, S., Wakabayashi, T. & Takakura, N. Identification and characterization of a resident vascular stem/progenitor cell population in preexisting blood vessels. EMBO J. 31 , 842–855 (2012).

Wakabayashi, T. et al. CD157 marks tissue-resident endothelial stem cells with homeostatic and regenerative properties. Cell Stem Cell 22 , 384–397 (2018).

Sekiguchi, H., Ii, M. & Losordo, D. W. The relative potency and safety of endothelial progenitor cells and unselected mononuclear cells for recovery from myocardial infarction and ischemia. J. Cell Physiol. 219 , 235–242 (2009).

Grant, M. B. et al. Adult hematopoietic stem cells provide functional hemangioblast activity during retinal neovascularization. Nat. Med. 8 , 607–612 (2002).

Otani, A. et al. Rescue of retinal degeneration by intravitreally injected adult bone marrow-derived lineage-negative hematopoietic stem cells. J. Clin. Invest. 114 , 765–774 (2004).

Otani, A. et al. Bone marrow-derived stem cells target retinal astrocytes and can promote or inhibit retinal angiogenesis. Nat. Med. 8 , 1004–1010 (2002).

O’Neill, C. L. et al. Endothelial cell-derived pentraxin 3 limits the vasoreparative therapeutic potential of circulating angiogenic cells. Cardiovasc. Res. 112 , 677–688 (2016).

Medina, R. J. et al. Endothelial progenitors: a consensus statement on nomenclature. Stem Cell Transl. Med. 6 , 1316–1320 (2017).

O’Neill, C. L. et al. Therapeutic revascularisation of ischaemic tissue: the opportunities and challenges for therapy using vascular stem/progenitor cells. Stem Cell Res. Ther. 3 , 31 (2012).

Park, S. S. et al. Intravitreal autologous bone marrow CD34+ cell therapy for ischemic and degenerative retinal disorders: preliminary phase 1 clinical trial findings. Invest. Ophthalmol. Vis. Sci. 56 , 81–89 (2015).

CAS   PubMed Central   Google Scholar  

Chambers, S. E. J. et al. The vasoreparative function of myeloid angiogenic cells is impaired in diabetes through the induction of IL1β. Stem Cells 36 , 834–843 (2018).

Medina, M. J. et al. Outgrowth endothelial cells characterisation and their potential for reversing ischemic retinopathy. Invest. Ophthalmol. Vis. Sci. 51 , 5906–5913 (2010).

Medina, R. J. et al. Ex-vivo expansion of human endothelial progenitors leads to IL-8-mediated replicative senescence and impaired vasoreparative function. Stem Cells 31 , 1657–1668 (2013).

Yoder, M. C. Defining human endothelial progenitor cells. J. Thromb. Haemost. 7 , 49–52 (2009).

Heo, S. C. et al. WKYMVm-induced activation of formyl peptide receptor 2 stimulates ischemic neovasculogenesis by promoting homing of endothelial colony-forming cells. Stem Cells 32 , 779–790 (2014).

Bertelli, P. M. et al. Vascular regeneration for ischemic retinopathies: hope from cell therapies. Curr. Eye Res. 45 , 372–384 (2020).

Prasain, N. et al. Differentiation of human pluripotent stem cells to cells similar to cord-blood endothelial colony-forming cells. Nat. Biotechnol. 32 , 1151–1157 (2014).

Reid, E. et al. Preclinical evaluation and optimization of a cell therapy using human cord blood-derived endothelial colony-forming cells for ischemic retinopathies. Stem Cell Transl. Med. 7 , 59–67 (2018).

Sakimoto, S. et al. CD44 expression in endothelial colony-forming cells regulates neurovascular trophic effect. JCI Insight 2 , e89906 (2017).

Cahoon, J. M. et al. Intravitreal AAV2.COMP-Ang1 prevents neurovascular degeneration in a murine model of diabetic retinopathy. Diabetes 64 , 4247–4259 (2015).

Mendel, T. A. et al. Pericytes derived from adipose-derived stem cells protect against retinal vasculopathy. PLoS ONE 8 , e65691 (2013).

Hajmousa, G. et al. Human adipose tissue-derived stromal cells act as functional pericytes in mice and suppress high-glucose-induced proinflammatory activation of bovine retinal endothelial cells. Diabetologia 61 , 2371–2385 (2018).

Langhe, R. & Pearson, R. A. Rebuilding the retina: prospects for Muller glial-mediated self-repair. Curr. Eye Res. 45 , 349–360 (2020).

Ahmad, I., Teotia, P., Erickson, H. & Xia, X. Recapitulating developmental mechanisms for retinal regeneration. Prog. Retin. Eye Res. 76 , 100824 (2019).

Zerti, D., Collin, J., Queen, R., Cockell, S. J. & Lako, M. Understanding the complexity of retina and pluripotent stem cell derived retinal organoids with single cell RNA sequencing: current progress, remaining challenges and future prospective. Curr. Eye Res. 45 , 385–396 (2020).

Galloway, C. A. et al. Drusen in patient-derived hiPSC-RPE models of macular dystrophies. Proc. Natl Acad. Sci. USA 114 , E8214–E8223 (2017).

Kiamehr, M. et al. Compromised barrier function in human induced pluripotent stem-cell-derived retinal pigment epithelial cells from type 2 diabetic patients. Int. J. Mol. Sci. 20 , 3773 (2019).

Kim, K. et al. Epigenetic memory in induced pluripotent stem cells. Nature 467 , 285–290 (2010).

Download references

Author information

Authors and affiliations.

Department of Ophthalmology and Visual Sciences, Department of Molecular and Integrative Physiology, Kellogg Eye Center, University of Michigan, Ann Arbor, MI, USA

David A. Antonetti

Department of Ophthalmology, Harvard Medical School, Boston, MA, USA

Paolo S. Silva

Beetham Eye Institute, Joslin Diabetes Center, Boston, MA, USA

Centre for Experimental Medicine, Queen’s University, Belfast, UK

Alan W. Stitt

You can also search for this author in PubMed   Google Scholar

Contributions

The authors contributed equally to all aspects of the article.

Corresponding author

Correspondence to David A. Antonetti .

Ethics declarations

Competing interests.

D.A.A. has received grant support from F. Hoffmann-La Roche, Inc. The other authors declare no competing interests.

Additional information

Peer review information.

Nature Reviews Endocrinology thanks R. Simó and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Related links

UKRI: https://gtr.ukri.org/projects?ref=MR%2FN006410%2F1

The clear gel filling the space between the retina and the lens.

The vascular bed adjacent to the retinal pigmented epithelium (RPE) supporting the rods and cones of the outer retina that are on the opposing side of the RPE.

The posterior segment of the human retina visible during ophthalmoscopy made up of the optic disc (optic nerve head) and macula, or avascular central area including the fovea, or thinned retina with high cone density responsible for high visual acuity.

Rights and permissions

Reprints and permissions

About this article

Cite this article.

Antonetti, D.A., Silva, P.S. & Stitt, A.W. Current understanding of the molecular and cellular pathology of diabetic retinopathy. Nat Rev Endocrinol 17 , 195–206 (2021). https://doi.org/10.1038/s41574-020-00451-4

Download citation

Accepted : 18 November 2020

Published : 19 January 2021

Issue Date : April 2021

DOI : https://doi.org/10.1038/s41574-020-00451-4

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

This article is cited by

Wnt-inhibitory factor 1-mediated glycolysis protects photoreceptor cells in diabetic retinopathy.

Journal of Translational Medicine (2024)

Porous Se@SiO2 nanospheres alleviate diabetic retinopathy by inhibiting excess lipid peroxidation and inflammation

Molecular Medicine (2024)

Cellular communication network factor 1 promotes retinal leakage in diabetic retinopathy via inducing neutrophil stasis and neutrophil extracellular traps extrusion

• Yanming Chen

Cell Communication and Signaling (2024)

Rejuvenation of diabetic macular edema with senolytic therapy

• Tomoaki Murakami
• Thomas W. Gardner

Nature Medicine (2024)

Modeling early pathophysiological phenotypes of diabetic retinopathy in a human inner blood-retinal barrier-on-a-chip

• Thomas L. Maurissen
• Alena J. Spielmann
• Héloïse Ragelle

Nature Communications (2024)

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

• Get new issue alerts Get alerts

Secondary Logo

Journal logo.

Colleague's E-mail is Invalid

Your message has been successfully sent to your colleague.

Save my selection

Mapping research trends in diabetic retinopathy from 2010 to 2019

A bibliometric analysis.

Editor(s): Wane., Daryle

a Tianjin Eye Hospital, Tianjin Key Lab of Ophthalmology and Visual Science, Tianjin Eye Institute, Clinical College of Ophthalmology Tianjin Medical University

b Ophthalmology Department, Baodi Clinical College of Tianjin Medical University, Tianjin Baodi Hospital, Tianjin, China.

∗Correspondence: Yi Dong, Tianjin Eye Hospital, Tianjin Key Lab of Ophthalmology and Visual Science, Tianjin Eye Institute, Clinical College of Ophthalmology Tianjin Medical University, Gansu Road 4, Tianjin 300020, China (e-mail: [email protected] ).

Abbreviations: AMD = age-related macular degeneration, BRB = blood-retinal barrier, DME = diabetic macular edema, DR = diabetic retinopathy, FA = fluorescein angiography, FAZ = foveal avascular zone, MAs = microaneurysms, MKD = mapping knowledge domain, OCT = optic coherence tomography, OCTA = optical coherence tomography angiography, PDR = proliferative diabetic retinopathy, PRP = panretinal photocoagulation, VA = visual acuity, VEGF = vascular endothelial growth factor, WoS = Web of Science, WoSCC = Web of Science Core Collection.

How to cite this article: Dong Y, Liu Y, Yu J, Qi S, Liu H. Mapping research trends in diabetic retinopathy from 2010 to 2019: a bibliometric analysis. Medicine . 2021;100:3(e23981).

YD and YL contributed equally to this work

The authors declare no conflicts of interest.

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

This is an open access article distributed under the terms of the Creative Commons Attribution-Non Commercial License 4.0 (CCBY-NC), where it is permissible to download, share, remix, transform, and buildup the work provided it is properly cited. The work cannot be used commercially without permission from the journal. http://creativecommons.org/licenses/by-nc/4.0

Background: 

Although many publications in diabetic retinopathy (DR) have been reported, there is no bibliometric analysis.

Purpose: 

To perform a bibliometric analysis in the field of diabetic retinopathy (DR) research, to characterize the current international status of DR research, to identify the most effective factors involved in this field, and to explore research hotspots in DR research.

Methods: 

Based on the Web of Science Core Collection (WoSCC), a bibliometric analysis was conducted to investigate the publication trends in research related to DR. Knowledge maps were constructed by VOSviewer v.1.6.10 to visualize the publications, the distribution of countries, international collaborations, author productivity, source journals, cited references and keywords, and research hotspots in this field.

Results: 

In total, 11,839 peer-reviewed papers were retrieved on DR from 2010 to 2019, and the annual research output increased with time. The United States ranks highest among countries with the most publications. The most active institution is the University of Melbourne. Wong, TY contributed the largest number of publications in this field. Investigative Ophthalmology & Visual Science was the most prolific journal in DR research. The top-cited references mainly investigated the use of anti-vascular endothelial growth factor (VEGF) medications in the management of DR, and the keywords formed 6 clusters:

• 1. pathogenesis of DR;
• 2. epidemiology and risk factors for DR;
• 3. treatments for DR;
• 4. screening of DR;
• 5. histopathology of DR; and
• 6. diagnostic methods for DR.

Discussion: 

With the improvement of living standard, DR has gradually become one of the important causes of blindness, and has become a hot spot of public health research in many countries. The application of deep learning and artificial intelligence in diabetes screening and anti-VEGF medications in the management of DR have been the research hotspots in recent 10 years.

Conclusions: 

Based on data extracted from the WoSCC, this study provides a broad view of the current status and trends in DR research and may provide clinicians and researchers with insight into DR research and valuable information to identify potential collaborators and partner institutions and better predict their dynamic directions.

1 Introduction

Diabetic retinopathy (DR) is an important complication of diabetes that affects blood vessels in the retina and can cause vision loss and blindness. [1] It is quickly becoming a worldwide public health challenge. [2,3] A large number of research papers related to DR have been published in academic journals in recent decades. In recent decades, many reviews on the pathology, metabolomics, imaging, biomarkers, and treatment of DR have been published. [1,4–8]

However, to our knowledge, the global research trend and other related topics in DR have not yet been well studied. It is difficult to read all the publications. Therefore, there is a need to use a method and tool to investigate the global status of the research in DR. Bibliometric methods and mapping knowledge domain (MKD) methods have been used in various fields to visually highlight the most influential countries, authors, journals, publications, and identify main research topics. [9] Bibliometric analysis is a method for analyzing the literature and its accompanying citation counts over time with mathematical statistics. The MKD method provides a new way to conduct literature mining and reveal the core structure of scientific knowledge. It also enables researchers to determine the range of research topics and identify new topics and assists them in planning their research direction and predicting research trends. [10] This study aimed to use bibliometric tools to analyze DR articles retrieved from the Web of Science (WoS) (Thomson Reuters Company) database and assess the research development status of DR throughout the world. This analysis could help us to uncover the current status and global trends of DR. It was hoped that our research results could provide meaningful help to the current researchers of DR. Recently, some systematic reviews have been conducted to evaluate the efficacy of PD Other systematic reviews have analyzed the adverse events in patients treated with PD However, the status of research in the area of PD-1 and PD-L1 in the cancer field and other related topics have not been investigated.

The remainder of the paper is structured as follows. The data collection and analytical methods are described in Section 2. The distribution of publications, countries, research organizations, journals, and research hotspots are presented in Sections 3. Global trends, document citation analysis, and research frontiers are discussed in Section 4. The final section, Section 5, summarizes the findings and concludes the paper.

2 Materials and methods

This study followed the tenets of the Declaration of Helsinki and was approved by the Medical Ethics Committee of Tianjin Eye Hospital and Tianjin Baodi Hospital. The search for papers to be included in this study was carried out on April 20, 2020 using the Science Citation Index Expanded (SCI-EXPANDED) database via the Web of Science Core Collection (WoSCC) provided by Thomson Reuters (Philadelphia, PA, USA). The database was searched using the term “diabetic retinopathy” in terms of “topic” (title, abstract, author's keywords, and WoS-assigned keywords, called Keywords Plus) to retrieve all articles where the expression “diabetic retinopathy” appeared, as well as other relevant expressions (e.g., diabetic retinopathies). The time span was set to between 2000 and 2019. Only articles were included as document types (nonarticle documents such as reviews, meeting abstracts, editorial materials, proceedings papers, letters were excluded). Journal articles were used for the analysis because they accounted for the majority of document types that also included complete research ideas and results. Data were downloaded from the WoS in “Full record and cited references” formats.

Visualization software can produce node-link maps that allow us to intuitively observe the publication outputs, hotspots, and other aspects of a research field. In this study, the data were imported into VOSviewer v.1.6.10 and analyzed systematically. VOSviewer ( www.vosviewer.com ), developed by van Eck and Waltman, is a literature visualization software that has the advantages of displaying cluster analysis results. [11] In the knowledge maps generated using VOSviewer, items are represented as nodes and links. The nodes and their labels, such as countries, organizations, authors, co-citation literature, and keywords, are proportional to the weight of the analysis components. The links between the nodes reflect the relationship between the components. CiteSpace IV (Drexel University, Philadelphia, PA) was used to capture keywords with strong citation bursts, which could be considered as predictors of research frontiers.

3.1 Yearly quantitative distribution of publications

According to the selection criteria, we identified and included 11,839 publications on DR that were indexed in the WoSCC from 2010 to 2020. The number of publications showed a gradually increasing trend over time, from 857 in 2010 to 1573 in 2019 ( Fig. 1 A). Through keyword burst detection analysis ( Fig. 1 B), we detected 28 keywords that represented citation bursts; among these keywords, “machine learning” showed citation bursts in 2019, which is consistent with the increase in published papers.

3.2 Distribution of productive countries in DR

According to the retrieved results, the 11,839 articles originated from 128 countries. As presented in Table 1 , the top 10 countries engaged in DR research published 10,419 articles, accounting for 88.0% of the total number of publications. The United States contributed the most publications (3280, 27.7%), followed by China (2222, 18.8%) and Japan (811, 6.9%). According to citation analysis, the United States had 79,761 citations, followed by China (26,304 citations) and Japan (15,670 citations).

Country co-authorship analysis reflects the degree of communication between countries as well as the most influential countries in this field. The larger nodes represent the more influential countries; the thickness and distance of the links between nodes represent the strength of the cooperative relationships among countries. Figure 2 shows that the United States intensely cooperated with many countries in the DR field, such as England, Australia, Germany, France, and Denmark. Although China has published a large number of articles, there is little cooperation with other countries. This indicates that geographical distance is not the primary influencing factor of cooperative relationships.

3.3 Distribution of main research organizations

According to the retrieved results, 11,839 articles were published by 8642 organizations. The top 10 organizations published 1852 articles, accounting for 15.64% of the total number of publications ( Table 2 ). Based on co-authorship analysis, Figure 3 displays the knowledge domain map of the research organizations’ distribution in DR research. The size of the node corresponds to the number of published articles. The links between nodes represent the collaborations. The thicker and longer the node-link, the closer the collaboration is between the 2 organizations.

3.4 Distribution of authors and co-authorship of research groups

According to the retrieved results, over 111,933 authors contributed to DR research. Among all authors, Wong Tienyin (91 publications) ranked first, followed by Wong Tieny (82 publications) and Klein Ronald (79 publications), indicating their productive contribution to DR research. Information on author co-citations was analyzed as well. Among all co-cited authors, Klein, R (3199 co-citations) ranked first, followed by Kowluru, RA (1618 co-citations), and Aiello, LP (1423 co-citations), indicating their relative influence on DR research ( Table 3 ).

According to the co-authorship analysis, Figure 4 displays the knowledge domain map of the authors in DR research. The size of the node corresponds to the number of published articles. The links between nodes represent the cooperative relationship between authors.

The greater the link strength, the greater the density of cooperation was between the linked authors.

3.5 Distribution of source journals

Based on the retrieved results, articles on DR research were published in 1524 journals. The top 10 journals that publish on this topic are listed in Table 4 . Investigative Ophthalmology & Visual Science published the greatest number of articles (782, 6.6%), followed by PLOS ONE (464, 3.9%) and Retina-The Journal of Retinal and Vitreous Diseases (455, 3.8%). Articles published in these 3 journals accounted for 14.36% of all publications included in this study.

3.6 Distribution of cited references: knowledge bases of DR research

Through co-citation analysis of the cited references, a knowledge base of DR research can be efficiently constructed. The minimum number of citations for a cited reference was set to 200. Of the 337,162 cited references, 305 met the threshold. The top 10 co-cited references are presented in Table 5 .

3.7 Distribution of keywords hotspots of DR research

Through the co-occurrence analysis of high-frequency keywords, the research hotspots of DR were identified. The minimum number of co-occurrences of a keyword was set to 20. Of the 13,415 extracted keywords involved in DR, 226 met the threshold. Based on the network, the keywords with similarities were clustered, and the 6 main clusters were denoted using the colors red, green, brown, yellow, purple, and blue, respectively ( Fig. 5 ). The top 10 keywords for each cluster are listed in Table 6 .

4 Discussion

4.1 global trends in research on dr.

The variation in the number of academic papers is an important research index that can reflect the development trend of the corresponding field. As shown in Figure 1 , a total of 11,839 papers were retrieved on DR from 2010 to 2019, and the annual research output increased with time. In the analysis of the most productive countries shown in Table 1 , the United States accounted for 27.7% of publications and ranked first in the number of publications. This indicates that the United States is the international scientific center of DR research.

Through the analysis of the distribution of research organizations, the most productive organizations and cooperation within the groups in a certain field can be identified. As shown in Table 2 , the most productive research institution was the University of Melbourne (252 documents), followed by Shanghai Jiao Tong University (216 documents) and Johns Hopkins University (208 documents), indicating that these research organizations are at the core of the entire research network. In terms of the number of links, the National University of Singapore presented the highest number (406 links), followed by the University of Melbourne (386 links), which indicated that these organizations are key nodes in the collaboration network (shown in Fig. 3 ).

The establishment of a co-authorship network knowledge map can provide valuable information to individual researchers seeking collaboration opportunities. The co-authorship groups are shown in Figure 4 : the red-colored group has Professor Klein as the center; the green-colored group has Professor Kowluru as the center; the blue-colored group has Professor Aiello as the center, and the yellow-colored group has Professor Spaide as the center.

A distribution analysis of academic journals helps determine the core journals in a certain research field. To this end, Investigative Ophthalmology & Visual Science, which has published the highest number of articles, is the most prolific journal on DR research.

4.2 Intellectual base

Based on the premise that high-quality research will be extensively cited, citation parameters were used to describe related topics within the selected articles. As shown in Table 5 , “A survey on deep learning in medical image analysis” ranked first in both citations and link strength. Through co-citation analysis, a large number of cited references can effectively show the background of a study. Therefore, we conducted a cluster analysis to explore the main topics in DR research. As shown in Table 5 , the 4 co-cited references list various clinical trials that mainly investigated the use of anti-VEGF medications in the management of DR. The publications entitled “A survey on deep learning in medical image analysis” and “Global Prevalence and Major Risk Factors of Diabetic Retinopathy” ranked in the top 2 in both frequency count and link weight, respectively, and are thus considered the core position of the whole knowledge map.

4.3 Research frontiers

The co-occurrence analysis of keywords is a common bibliometric research method in which the assigned keywords are considered to represent the search theme. Thus, the internal structure of the related literature and the frontier discipline can be revealed. As shown in Table 6 , DR topics mainly formed 6 clusters, and keywords in the same cluster showed greater similarity to a specific research topic than keywords in different clusters. Combined with the characteristics and current status of DR research, the 6 clusters are described as follows:

Cluster #1 (red) represents keywords mainly related to the pathogenesis of DR. The extracted co-occurrence keywords include “inflammation”, “angiogenesis”, “oxidative stress”, “apoptosis”, “hypoxia”, and “neurodegeneration”. Chronic hyperglycemia leads to increased inflammation and oxidative stress in the retina, which seems causally related to the development of at least diabetes-induced leakage and the degeneration of retinal capillaries. The incubation of retinal cells in high glucose causes the upregulation of proinflammatory factors, such as Inducible nitric oxide synthase, cyclooxygenase-2, and leukotrienes. [12–16] Inflammatory processes play an important role in the development of early and possibly later stages of DR, and the inflammatory pathogenesis of DR is based on the molecular characteristics of inflammation, as opposed to the classical cellular definition of inflammation. [17] Diabetes-induced oxidative stress plays a role in the development of inflammatory processes in the retina. [18,19] Two months of diabetes in rats significantly increased retinal levels of interleukin-1β and nuclear factor kappa-B, and antioxidants inhibited those increases. [20] Other research has shown that inhibition of interleukin -6 trans-signaling significantly reduces diabetes-induced oxidative damage in the retina. [21] Early proinflammatory changes, such as the appearance of microglia, the formation of advanced glycation endproducts, and the overproduction of VEGF, can directly cause hypoxia in the retina and not necessarily via reactive oxygen species. [22] VEGF is known to be a proinflammatory molecule whose vitreal levels are highly correlated with retinal neovascularization and edema. Many studies have evaluated the association of enzymes or gene polymorphisms with DR; for example, the nitric oxide synthase 3 gene rs869109213 polymorphism alone or in combination with the endothelin receptor B gene rs10507875 polymorphism may be associated with DR in Slovenian patients with type 2 diabetes mellitus, [23] and the methylenetetrahydrofolate reductase C677T polymorphism may contribute to DR development in multiethnic groups. [24] Researching the pathogenesis of DR could provide new therapeutic targets for inhibiting or preventing retinopathy.

Cluster #2 (green) represents keywords related to the epidemiology of and risk factors for DR. Age-standardized to the 2010 population, there are approximately 93 million people with DR, 17 million with proliferative DR, and 21 million with diabetic macular edema (DME). The overall prevalence is 34.6% for any DR, 6.96% for proliferative DR, 6.81% for DME, and 10.2% for vision-threatening DR. [25] The most common risk factors for DR are longer diabetes duration and poorer glycemic and blood pressure control. [26,27] Moreover, the overall prevalence is higher in people with type 1 diabetes than in those with type 2 diabetes. [25] In China in 2010, the pooled prevalence rates of any DR, nonproliferative DR, and proliferative DR were 1.14%, 0.90%, and 0.07% in the general population and 18.45%, 15.06%, and 0.99% in people with diabetes, respectively. A total of 13.16 million Chinese individuals aged 45 years and above live with DR, and the risk factors include residing in rural China, insulin treatment, elevated fasting blood glucose levels, and higher glyeosylated hemoglobin concentrations. [2] Other risk factors for DR include poor blood pressure and lipid control, high body mass index, puberty, pregnancy, and cataract surgery. There are weaker associations with some genetic and inflammatory markers. DR has become a serious global public health problem. [28] Diabetic nephropathy is another major public health problem with social and economic burdens. The prevalence of nephropathy among individuals with retinopathy is 35.6%, and there is a significant association between nephropathy and the development of retinopathy. Two abso ute risk factors for DR are nephropathy and hypertension. [29]

Cluster #3 (brown) represents keywords related to treatments for DR, such as “intravitreal injections” of “anti-VEGF”, “vitrectomy” and “panretinal photocoagulation” (PRP). DME is very common in proliferative diabetic retinopathy (PDR) and is characterized by metamorphopsia and loss of visual acuity (VA). Anti-VEGF intravitreal injections can benefit most patients. Aflibercept, bevacizumab, and ranibizumab are 3 commonly used anti-VEGF agents whose molecular structure and properties differ. [30] Many clinical trials have been conducted to determine the optimal anti-VEGF drug among the 3 listed above, as well as to elucidate their efficacy and guide their administration frequency for patients with DME. The Diabetic Retinopathy Clinical Research Network conducted a comparative effectiveness study for center-involved DME for all 3 drugs at a 2-year follow-up visit. Among eyes with better VA at baseline, no difference was identified in vision outcomes through the 2-year follow-up. For the eyes with worse VA at baseline, the advantage of aflibercept over bevacizumab for mean VA gain persisted through the 2 years, although the difference at 2 years was diminished. The VA difference between aflibercept and ranibizumab for eyes with worse VA at baseline that was noted at 1 year had decreased at 2 years. [31,32] The disadvantages are frequent injection, high medical costs, and poor results in some patients after multiple injections. Compared with anti-VEGF drugs, dexamethasone implants significantly improve anatomical outcomes. However, this does not translate to improve VA, which may be due to the progression of cataracts. Therefore, the dexamethasone implant may be recommended as the first choice for select cases, such as for pseudophakic eyes, anti-VEGF-resistant eyes, or patients reluctant to receive frequent intravitreal injections. [33] PDR is the worst stage of DR. For decades, PRP and vitrectomy have been the standard of care for the treatment of PDR. Recently, anti-VEGF has provided a new standard of care in PDR. [34] Tractional macular detachment occurs in 10% of eyes after anti-VEGF agent pretreatment before vitrectomy for complicated PDR. The main risk factors are days between anti-VEGF injection and vitrectomy, vitreous hemorrhage, and age. [35] However, preoperative intravitreal injections of anti-VEGF agents are effective and safe for complicated PDR. [36,37]

Cluster #4 (yellow) represents keywords related to the screening of DR. DR results in vision loss if not treated early. However, the interpretation of retinal images requires specialized knowledge and expertise in DR, and capital- and labor-intensive screening programs are difficult to rapidly scale up and expand to meet the needs of this growing global epidemic. Many artificial intelligence and deep learning-based methods have been developed from large image datasets in the assessment of retinal photographs for the detection and screening of DR as well as in the segmentation and assessment of optic coherence tomography (OCT) images for the diagnosis and screening of DME. [38] A hospital-based cross-sectional study showed that non-mydriatic funduscopic screening photography was practical and useful for the detection of DR in patients with type 1 and type 2 diabetes. [39] Telemedicine services facilitate the evaluation, diagnosis, and management of remote patients. In ophthalmology, telemedicine is in its infancy, particularly in its application to DR, as current models are largely performed via “store and forward” methods, but remote monitoring and interactive modalities exist. Telemedicine has the potential to improve access to care, decrease the cost of care, and improve adherence to evidence-based protocols. [40]

Cluster #5 (purple) represents keywords related to the histopathology of DR. The blood-retinal barrier (BRB) is a particularly tight and restrictive physiological barrier that regulates ion, protein, and water flux into and out of the retina. It consists of inner and outer components, the inner BRB being formed of tight junctions between retinal capillary endothelial cells and the outer BRB of tight junctions between retinal pigment epithelial cells. OCT is widely used to evaluate the BRB. DR and age-related macular degeneration (AMD) are the 2 most frequent and relevant retinal diseases that are directly associated with alterations of the BRB. [41] DR is a result of retinal neovascularization, and wet AMD is initiated by choroidal neovascularization. [41] It is well known that intravitreal injections of anti-VEGF agents are effective and safe in minimizing neovascularization. [42] Aflibercept and ranibizumab can both reduce macular edema and improve the VA of patients; these drugs are more commonly used in DME. [32] Swept-source OCT demonstrates a significant reduction in choroidal thickness of eyes with PDR compared with that of controls. In the foveal region, the choroid appears to be thinner in DR eyes than in diabetic eyes without retinopathy. [43]

Cluster #6 (blue) represents diagnostic methods for DR, such as “OCT”, “optical coherence tomography angiography (OCTA)”, and “fluorescein angiography (FA)”. FA is a dye-based ocular angiography and has been used in clinical practice for over 50 years. Only the superficial vascular plexus can be seen with it, and it provides limited information about the choroidal circulation. Despite these limitations, FA imaging offers many other advantages, including dynamic information regarding the transit of blood as well as identification of dye leakage from disruptions of BRB by disease. OCT constitutes one of the greatest advances in ophthalmic imaging and is capable of showing structural images of the retina and choroid. Building on this platform, OCTA provides depth-resolved images of blood flow in the retina and choroid with levels of detail far exceeding those obtained with FA. OCTA can generate high contrast, well-defined images of the microvasculature. Besides, OCTA images can be viewed in cross-section to confirm the depth location of vascular pathologies. Finally, OCTA can be performed much more rapidly than FA or indocyanine green angiography, streamlining the clinical workflow. At the same time, OCTA has important limitations. First, the fields of view that can be imaged by OCTA are smaller than those imaged by FA. Second, OCTA signals have a limited dynamic range. [44] Many studies have been conducted to compare OCTA with FA in patients with DR. [45–47]

These studies have found good agreement for the size of the foveal avascular zone (FAZ) and weak agreement regarding the number of microaneurysms (MAs) in both imaging modalities. It is better to assess the FAZ with OCTA and MAs with FA. Complementary use of FA and OCTA is the best diagnostic approach. [46] Diabetic macular ischemia grade, the size of the FAZ on OCTA, ellipsoid zone disruption, and disorganization of the retinal inner layers with OCT are associated with VA. The use of OCTA and OCT can predict VA in DR. [48] With the development of newer, wide-angle imaging technologies, wide-angle OCT, wide-angle OCTA, ultrawide-field FA, longitudinal wide-field swept-source OCTA, and en face OCTA have gradually been used in DR. [49–51] The multimodal imaging approach keeps the findings of any one modality in perspective by integrating that information with potentially useful data obtained by other imaging methods, which allows clinicians to gain the most information from each modality and thereby optimize patient care. [52]

5 Conclusions

We constructed a series of science maps of the annual number of publications, the distribution of countries, international collaborations, author productivity, source journals, cited references, and keywords in DR research. The results of this study may be helpful for ophthalmologists in choosing appropriate journals for publication and organizations or authors for collaborations. The extracted keywords enable researchers to identify new topics and assist them in predicting research directions. However, some limitations should be considered. First, the publications were extracted from the WoSCC between 2010 and 2019, which may not sufficiently represent all of the topics in DR research. Second, the primary data were extracted from WoSCC, which is a database more suited for performing citation analysis. Third, because most publications in the WoSCC were in English, a linguistic bias may exist. Last but not least, the collaboration network analysis successfully displayed the co-occurrence (distance between the 2 nodes/items) and the co-authorship of the institutions (the strength of the links). However, the strength of each pair of linked items is not shown in the final exported file, and VOSviewer was unable to generate a geographical map of co-authorship. Thus, a visualization of geographical location and co-authorship cannot be generated, and an understanding of the relationship thereof cannot be determined.

6 Future studies

Future studies may consider exploring a specific aspect of DR research, such as medicine and surgery. We will use different document databases and other bibliometric methods (such as bibliographic coupling analysis) to study other aspects of DR. In addition, Altmetrics, a new and comprehensive bibliometric method for evaluating the academic and social influences of research outputs, can also be applied in combination with scientometric analysis to better understand the trends and new areas of research in the field. It can be predicted that there will be an increasing number of papers in the coming year. In particular, studies about medicine and imaging will be the next popular hotspots and should receive more attention in the future.

Acknowledgments

The authors would like to thank all reviewers for their valuable comments.

Author contributions

Conceptualization: Yi Dong, Shixin Qi.

Data curation: Yi Dong, Huijuan Liu.

Formal analysis: Yi Dong.

Investigation: Yi Dong, Jianguo Yu.

Methodology: Yi Dong, Yanli Liu, Shixin Qi.

Project administration: Yi Dong, Yanli Liu, Jianguo Yu, Huijuan Liu.

Resources: Yi Dong, Yanli Liu.

Software: Yi Dong.

Supervision: Yi Dong, Jianguo Yu.

Validation: Yi Dong, Shixin Qi.

Visualization: Yi Dong.

Writing – original draft: Yanli Liu.

Writing – review & editing: Yi Dong, Yanli Liu.

• Cited Here |
• Google Scholar

bibliometric analysis; citespace; diabetic retinopathy; vosviewer

• + Favorites
• View in Gallery

Information

• Author Services

Initiatives

You are accessing a machine-readable page. In order to be human-readable, please install an RSS reader.

All articles published by MDPI are made immediately available worldwide under an open access license. No special permission is required to reuse all or part of the article published by MDPI, including figures and tables. For articles published under an open access Creative Common CC BY license, any part of the article may be reused without permission provided that the original article is clearly cited. For more information, please refer to https://www.mdpi.com/openaccess .

Feature papers represent the most advanced research with significant potential for high impact in the field. A Feature Paper should be a substantial original Article that involves several techniques or approaches, provides an outlook for future research directions and describes possible research applications.

Feature papers are submitted upon individual invitation or recommendation by the scientific editors and must receive positive feedback from the reviewers.

Editor’s Choice articles are based on recommendations by the scientific editors of MDPI journals from around the world. Editors select a small number of articles recently published in the journal that they believe will be particularly interesting to readers, or important in the respective research area. The aim is to provide a snapshot of some of the most exciting work published in the various research areas of the journal.

Original Submission Date Received: .

• Active Journals
• Find a Journal
• Proceedings Series
• For Authors
• For Reviewers
• For Editors
• For Librarians
• For Publishers
• For Societies
• For Conference Organizers
• Open Access Policy
• Institutional Open Access Program
• Special Issues Guidelines
• Editorial Process
• Research and Publication Ethics
• Article Processing Charges
• Testimonials
• Preprints.org
• SciProfiles
• Encyclopedia

Article Menu

• Subscribe SciFeed
• Recommended Articles
• Google Scholar
• on Google Scholar
• Table of Contents

Find support for a specific problem in the support section of our website.

Please let us know what you think of our products and services.

Visit our dedicated information section to learn more about MDPI.

JSmol Viewer

A systematic literature review on diabetic retinopathy using an artificial intelligence approach.

1. Introduction

1.1. applications of ai in retina images.

• Classification: Categorization cases are commonly used in binary or multi-class retinal image analysis, such as automated screenings or detecting of the stage of disease or type. ML and DL methods are applicable here based on the level of understandability required or the quantity of the dataset provided.
• Segmentation: The fundamental goal of segmentation-based approaches is to subdivide the objects in a picture. The primary purpose of all these techniques is to investigate morphological features or retrieve a meaningful pattern or feature of relevance from a snapshot, such as borders in 2D or 3D imaging. Segmentation of pigment epithelial detachment (PED) is used to diagnose chorioretinal diseases.
• Prediction: Most predicted situations are alarmed with illness development, future treatment outcomes based on an image, etc. The prediction approach can also be used to depict the local retention region.

1.2. Diabetic Retinopathy (DR)

• Hemorrhages (HM) appear as patches on the retina which can be 125 μm in diameter with an uneven edge. Its two categories are flames (superficial HM) and blot (deep HM) [ 23 ].
• Hard exudates: Hard exudates, which typically can be seen as bright yellow areas on the eye, are caused by hemolysis. These were also found in the eye’s coastal parts and had clear boundaries.
• Soft exudates: White spots on the eye generated from nerve fiber swelling are called soft exudates (cotton wool). These are ovular or circular. Soft or hard secretions constitute white lesions, whereas MA and HM were red growths (EX). A sample image of various stages of DR is provided in Figure 4 . DR is classified as non-proliferative DR (NPDR) and proliferative DR (PDR). Further, NPDR is classified as mild, moderate, and severe, as shown in Figure 5 .

1.3. Evolution of DR Using AI

1.4. prior research.

• Datastores in the discipline of DR detection are accessible online, as well as the existence of DR datasets.
• An exhaustive survey of widely used ML and DL methodologies for DR detection is discussed.
• Feature extraction and classification techniques used in DR are discussed.
• Future research concepts such as domain adaptation, multitask learning, and explainable AI in DR detection are discussed

1.5. Motivation

1.6. research goals, 1.7. contribution of the study.

• To exploring available data sets which have been used for detecting DR.
• To investigate artificial intelligence strategies that have been employed in the literature for DR detection.
• To explore feature extraction and classification.
• To study multiple assessment metrics to analyze DR detection and categorization.
• To highlight the scope of future research, concepts such as domain adaptation, multitask learning, and explainable AI in DR detection techniques used in DR.

2. Research Mechanism of Study

Paradigms for inclusion and exclusion.

• RC1: Recommendations and results must be included in research articles.
• RC2: Scientific data must be included in scientific papers to support their conclusions.
• RC3: The aims and findings of the research must be expressed.
• RC4: For scientific studies, citations must be proper and adequate.

3. RQ1 Artificial Intelligence for DR Detection

3.1. machine learning techniques in dr detection, 3.2. deep learning in dr screening, 3.3. transfer learning in dr, 4. rq2 feature extraction techniques for dr, 4.1. explicit or traditional feature extraction methods, 4.2. direct methods, 5. rq3 datasets available for dr, 6. rq4 evaluation measures used for dr detection, 6.1. false positive rate (fpr), 6.2. false negative rate (fnr), 6.3. accuracy [ 89 ], 6.4. specificity, 6.5. sensitivity/recall rate, 6.6. f-score, 6.8. positive predictive value (ppv), 6.9. negative predictive value (npv), 6.10. false discovery rate (fdr), 6.11. confusion matrix.

• True Negative: when the model’s predicted and the actual value is No.
• True Positive: when the model’s predicted and the actual value is Yes.
• False Negative: when the model’s predicted value is Yes, and the actual value is No. It is also known as a Type-II mistake.
• False Positive: when the model’s predicted value is No, and the actual value is Yes. A type-I mistake is another name for it.

6.12. Kappa Value

7. emr and biomarkers in dr.

• Genetics: The investigation of genes associated with the development of advanced DR, vascular endothelial growth factor (VEGF), lipoproteins, and inflammation. There have been genome-wide association studies and single nucleotide polymorphisms (SNPs) linked to an enhanced danger of sight-threatening retinopathy [ 164 ].
• Epigenetics: It is the study of how environmental variables interact with genes. DNA methylation, histone modification, and microRNAs are among the biomarkers being investigated [ 165 , 166 ].
• Proteomics: It is the study of protein structure and function research in cultured cells and tissues. A current study shows that diabetic patients have higher levels of transport proteins (vitamin D binding protein), arginine N-methyltransferase 5, and inflammatory proteins (leucine-rich alpha-2-glycoprotein) [ 167 , 168 ].
• Metabolomics: The study of chemical traces left by biological activities. Data on increased metabolite cytidine, cytosine, and thymidine found in DR patients using mass spectrometry is included in the studies. These nucleotide concentrations may be relevant in monitoring DR progression and evaluating therapy [ 169 ].

8. RQ5 Challenges and Future Research Directions

9. discussion, 10. conclusions, author contributions, institutional review board statement, informed consent statement, data availability statement, conflicts of interest.

• Khatri, M. Diabetes Complications. Available online: https://www.webmd.com/diabetes/diabetes-complications (accessed on 18 May 2022).
• Chakrabarti, R.; Harper, C.A.; Keeffe, J.E. Diabetic Retinopathy Management Guidelines. Expert Rev. Ophthalmol. 2012 , 7 , 417–439. [ Google Scholar ] [ CrossRef ]
• Early Treatment Diabetic Retinopathy Study Research Group. Grading diabetic retinopathy from stereoscopic color fundus photographs- an extension of the modified Airlie House classification. Ophthalmology 2020 , 127 , S99–S119. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Scanlon, P.H.; Wilkinson, C.P.; Aldington, S.J.; Matthews, D.R. A Practical Manual of Diabetic Retinopathy Management , 1st ed.; Wiley-Blackwell: Hoboken, NJ, USA, 2009; pp. 1–214. [ Google Scholar ] [ CrossRef ]
• Ravelo, J.L. Aging and Population Growth, Challenges for Vision Care: WHO Report. 2019. Available online: https://www.devex.com/news/aging-and-population-growth-challenges-for-vision-care-who-report-95763 (accessed on 3 January 2022).
• WHO. World Report on Vision, 2019. Available online: https://www.who.int/publications/i/item/9789241516570 (accessed on 3 January 2022).
• Kumar, R.; Pal, R. India achieves WHO recommended doctor population ratio: A call for a paradigm shift in public health discourse! J. Fam. Med. Prim. Care 2018 , 7 , 841–844. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• WHO. Global Data on Visual Impairment. 2010. Available online: http://www.who.int/blindness/GLOBALDATAFINALforweb.pdf (accessed on 5 May 2022).
• Centers for Disease Control and Prevention. Common Eye Disorders and Diseases. 2020. Available online: https://www.cdc.gov/visionhealth/basics/ced/index.html (accessed on 10 May 2022).
• Malik, U. Most Common Eye Problems—Signs, Symptoms and Treatment Options. 2021. Available online: https://irisvision.com/most-common-eye-problems-signs-symptoms-and-treatment/ (accessed on 3 April 2022).
• Stoitsis, J.; Valavanis, I.; Mougiakakou, S.G.; Golemati, S.; Nikita, A.; Nikita, K.S. Computer aided diagnosis based on medical image processing and artificial intelligence methods. Nucl. Instrum. Methods Phys. Res. Sect. A 2006 , 569 , 591–595. [ Google Scholar ] [ CrossRef ]
• Mushtaq, G.; Siddiqui, F. Detection of diabetic retinopathy using deep learning methodology. IOP Conf. Ser. Mater. Sci. Eng. 2021 , 1070 , 012049. [ Google Scholar ] [ CrossRef ]
• Taylor, R.; Batey, D. Handbook of Retinal Screening in Diabetes: Diagnosis and Management. In Handbook of Retinal Screening in Diabetes: Diagnosis and Management , 2nd ed.; Wiley-Blackwell: Chichester, UK, 2012; pp. 89–103. [ Google Scholar ]
• Gupta, A.; Chhikara, R. Diabetic Retinopathy: Present and Past. Procedia Comput. Sci. 2018 , 132 , 1432–1440. [ Google Scholar ] [ CrossRef ]
• Ishtiaq, U.; Kareem, S.A.; Abdullah, E.R.M.F.; Mujtaba, G.; Jahangir, R.; Ghafoor, H.Y. Diabetic retinopathy detection through artificial intelligent techniques: A review and open issues. Multimedia Tools Appl. 2019 , 79 , 15209–15252. [ Google Scholar ] [ CrossRef ]
• Lin, J.; Yu, L.; Weng, Q.; Zheng, X. Retinal image quality assessment for diabetic retinopathy screening: A survey. Multimedia Tools Appl. 2020 , 79 , 16173–16199. [ Google Scholar ] [ CrossRef ]
• Qureshi, I. Glaucoma Detection in Retinal Images Using Image Processing Techniques: A Survey. Int. J. Adv. Netw. Appl. 2015 , 7 , 2705–2718. Available online: http://www.ijana.in/papers/V7I2-10.pdf (accessed on 5 April 2022).
• Wang, Z.; Yin, Y.; Shi, J.; Fang, W.; Li, H.; Wang, X. Zoom-in-Net: Deep Mining Lesions for Diabetic Retinopathy Detection. In Proceedings of the International Conference on Medical Image Computing and Computer-Assisted Intervention, Quebec City, QC, Canada, 11–13 September 2017; pp. 267–275. [ Google Scholar ]
• Scotland, G.S.; McNamee, P.; Fleming, A.D.; Goatman, K.A.; Philip, S.; Prescott, G.J.; Sharp, P.F.; Williams, G.J.; Wykes, W.; Leese, G.P.; et al. Costs and consequences of automated algorithms versus manual grading for the detection of referable diabetic retinopathy. Br. J. Ophthalmol. 2010 , 94 , 712–719. [ Google Scholar ] [ CrossRef ]
• Difference between Normal Vision and DR Vision. Available online: https://www.researchgate.net/publication/350930649_DRISTI_a_hybrid_deep_neural_network_for_diabetic_retinopathy_diagnosis/figures?lo=1 (accessed on 20 May 2022).
• Li, T.; Gao, Y.; Wang, K.; Guo, S.; Liu, H.; Kang, H. Diagnostic assessment of deep learning algorithms for diabetic retinopathy screening. Inf. Sci. 2019 , 501 , 511–522. [ Google Scholar ] [ CrossRef ]
• Pratt, H.; Coenen, F.; Broadbent, D.M.; Harding, S.P.; Zheng, Y. Convolutional neural networks for diabetic retinopathy. Procedia Comput. Sci. 2016 , 90 , 200–205. [ Google Scholar ] [ CrossRef ] [ Green Version ]
• Alyoubi, W.L.; Shalash, W.M.; Abulkhair, M.F. Diabetic retinopathy detection through deep learning techniques: A review. Inform. Med. Unlocked 2020 , 20 , 100377. [ Google Scholar ] [ CrossRef ]
• Arrigo, A.; Teussink, M.; Aragona, E.; Bandello, F.; Parodi, M.B. MultiColor imaging to detect different subtypes of retinal microaneurysms in diabetic retinopathy. Eye 2020 , 35 , 277–281. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Yasin, S.; Iqbal, N.; Ali, T.; Draz, U.; Alqahtani, A.; Irfan, M.; Rehman, A.; Glowacz, A.; Alqhtani, S.; Proniewska, K.; et al. Severity Grading and Early Retinopathy Lesion Detection through Hybrid Inception-ResNet Architecture. Sensors 2021 , 21 , 6933. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Guo, S.; Li, T.; Kang, H.; Li, N.; Zhang, Y.; Wang, K. An end-to-end unified framework for multi-lesion segmentation offundus images. Neurocomput 2019 , 349 , 52–63. [ Google Scholar ] [ CrossRef ]
• Li, B.; Li, H.K. Automated Analysis of Diabetic Retinopathy Images: Principles, Recent Developments, and Emerging Trends. Curr. Diabetes Rep. 2013 , 13 , 453–459. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Mishra, A.; Singh, L.; Pandey, M. Short Survey on machine learning techniques used for diabetic retinopathy detection. In Proceedings of the IEEE 2021 International Conference on Computing, Communication, and Intelligent Systems (ICCCIS), Greater Noida, India, 19–20 February 2021; pp. 601–606. [ Google Scholar ] [ CrossRef ]
• Oh, K.; Kang, H.M.; Leem, D.; Lee, H.; Seo, K.Y.; Yoon, S. Early detection of diabetic retinopathy based on deep learning and ultra-wide-field fundus images. Sci. Rep. 2021 , 11 , 1–9. [ Google Scholar ] [ CrossRef ]
• Kermany, D.S.; Goldbaum, M.; Cai, W.; Valentim, C.C.S.; Liang, H.; Baxter, S.L.; McKeown, A.; Yang, G.; Wu, X.; Yan, F.; et al. Identifying Medical Diagnoses and Treatable Diseases by Image-Based Deep Learning. Cell 2018 , 172 , 1122–1131.e9. [ Google Scholar ] [ CrossRef ]
• Mansour, R.F. Deep-learning-based automatic computer-aided diagnosis system for diabetic retinopathy. Biomed. Eng. Lett. 2017 , 8 , 41–57. [ Google Scholar ] [ CrossRef ]
• Pepose, J.S. A prospective randomized clinical evaluation of 3 presbyopia-correcting intraocular lenses after cataract extraction. Am. J. Ophthalmol. 2014 , 3–9. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Boudry, C.; Denion, E.; Mortemousque, B.; Mouriaux, F. Trends and topics in eye disease research in PubMed from 2010 to 2014. PeerJ 2016 , 4 , e1557. [ Google Scholar ] [ CrossRef ] [ PubMed ] [ Green Version ]
• Gardner, G.G.; Keating, D.; Williamson, T.H.; Elliott, A.T. ORIGINAL ARTICLES-Clinical science Automatic detection of diabetic retinopathy using an artificial neural network: A screening tool. Br. J. Ophthalmol. 1996 , 80 , 940–944. [ Google Scholar ] [ CrossRef ] [ Green Version ]
• Shi, C.; Lee, J.; Wang, G.; Dou, X.; Yuan, F.; Zee, B. Assessment of image quality on color fundus retinal images using the automatic retinal image analysis. Sci. Rep. 2022 , 12 , 1–11. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Franklin, S.W.; Rajan, S.E. An automated retinal imaging method for the early diagnosis of diabetic retinopathy. Technol. Health Care 2013 , 21 , 557–569. [ Google Scholar ] [ CrossRef ]
• Li, S.; Zhao, R.; Zou, H. Artificial intelligence for diabetic retinopathy. Chin. Med. J. 2021 , 135 , 253–260. [ Google Scholar ] [ CrossRef ]
• Pragathi, P.; Rao, A.N. An effective integrated machine learning approach for detecting diabetic retinopathy. Open Comput. Sci. 2022 , 12 , 83–91. [ Google Scholar ] [ CrossRef ]
• Zhang, W.; Zhong, J.; Yang, S.; Gao, Z.; Hu, J.; Chen, Y.; Yi, Z. Automated identification and grading system of diabetic retinopathy using deep neural networks. Knowl. Based Syst. 2019 , 175 , 12–25. [ Google Scholar ] [ CrossRef ]
• Asiri, N.; Hussain, M.; Al Adel, F.; Alzaidi, N. Deep learning based computer-aided diagnosis systems for diabetic retinopathy: A survey. Artif. Intell. Med. 2019 , 99 , 101701. [ Google Scholar ] [ CrossRef ]
• Kamal, M.M.; Shanto, M.H.I.; Mirza Mahmud Hossan, M.; Hasnat, A.; Sultana, S.; Biswas, M. A Comprehensive Review on the Diabetic Retinopathy, Glaucoma and Strabismus Detection Techniques Based on Machine Learning and Deep Learning. Eur. J. Med. Health Sci. 2022 , 24–40. [ Google Scholar ] [ CrossRef ]
• Aoun, S.M.; Breen, L.J.; Oliver, D.; Henderson, R.D.; Edis, R.; O’Connor, M.; Howting, D.; Harris, R.; Birks, C. Family carers’ experiences of receiving the news of a diagnosis of Motor Neurone Disease: A national survey. J. Neurol. Sci. 2017 , 372 , 144–151. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Khade, S.; Ahirrao, S.; Phansalkar, S.; Kotecha, K.; Gite, S.; Thepade, S.D. Iris Liveness Detection for Biometric Authentication: A Systematic Literature Review and Future Directions. Inventions 2021 , 6 , 65. [ Google Scholar ] [ CrossRef ]
• Grzybowski, A.; Brona, P.; Lim, G.; Ruamviboonsuk, P.; Tan, G.S.W.; Abramoff, M.; Ting, D.S.W. Artificial intelligence for diabetic retinopathy screening: A review. Eye 2020 , 34 , 451–460. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Ribeiro, L.; Oliveira, C.M.; Neves, C.; Ramos, J.D.; Ferreira, H.; Cunha-Vaz, J. Screening for Diabetic Retinopathy in the Central Region of Portugal. Added Value of Automated ‘Disease/No Disease’ Grading. Ophthalmologica 2015 , 233 , 96–103. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Ipp, E.; Liljenquist, D.; Bode, B.; Shah, V.N.; Silverstein, S.; Regillo, C.D.; Lim, J.I.; Sadda, S.; Domalpally, A.; Gray, G.; et al. Pivotal Evaluation of an Artificial Intelligence System for Autonomous Detection of Referrable and Vision-Threatening Diabetic Retinopathy. JAMA Netw. Open 2021 , 4 , e2134254. [ Google Scholar ] [ CrossRef ]
• Larsen, N.; Godt, J.; Grunkin, M.; Lund-Andersen, H.; Larsen, M. Automated Detection of Diabetic Retinopathy in a Fundus Photographic Screening Population. Investig. Opthalmology Vis. Sci. 2003 , 44 , 767–771. [ Google Scholar ] [ CrossRef ] [ Green Version ]
• Ting, D.S.W.; Cheung, C.Y.-L.; Lim, G.; Tan, G.S.W.; Quang, N.D.; Gan, A.; Hamzah, H.; Garcia-Franco, R.; Yeo, I.Y.S.; Lee, S.Y.; et al. Development and Validation of a Deep Learning System for Diabetic Retinopathy and Related Eye Diseases Using Retinal Images from Multiethnic Populations With Diabetes. JAMA 2017 , 318 , 2211–2223. [ Google Scholar ] [ CrossRef ]
• Gulshan, V.; Peng, L.; Coram, M.; Stumpe, M.C.; Wu, D.; Narayanaswamy, A.; Venugopalan, S.; Widner, K.; Madams, T.; Cuadros, J.; et al. Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Retinal Fundus Photographs. JAMA 2016 , 316 , 2402–2410. [ Google Scholar ] [ CrossRef ]
• Abràmoff, M.D.; Lavin, P.T.; Birch, M.; Shah, N.; Folk, J.C. Pivotal trial of an autonomous AI-based diagnostic system for detection of diabetic retinopathy in primary care offices. NPJ Digit. Med. 2018 , 39 , 1–8. [ Google Scholar ] [ CrossRef ]
• Dong, X.; Du, S.; Zheng, W.; Cai, C.; Liu, H.; Zou, J. Evaluation of an Artificial Intelligence System for the Detection of Diabetic Retinopathy in Chinese Community Healthcare Centers. Front. Med. 2022 , 9 , 840024. [ Google Scholar ] [ CrossRef ]
• Wu, J.; Xin, J.; Hong, L.; You, J.; Zheng, N. New hierarchical approach for microaneurysms detection with matched filter and machine learning. In Proceedings of the 2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Milan, Italy, 25–29 August 2015; IEEE: Milan, Italy, 2015; pp. 4322–4325. [ Google Scholar ] [ CrossRef ] [ Green Version ]
• Biyani, R.S.; Patre, B.M. A clustering approach for exudates detection in screening of diabetic retinopathy. In Proceedings of the 2016 International Conference on Signal and Information Processing (IConSIP), Nanded, India, 6–8 October 2016; pp. 1–5. [ Google Scholar ] [ CrossRef ]
• Naqvi, S.A.G.; Zafar, M.F.; Haq, I.U. Referral system for hard exudates in eye fundus. Comput. Biol. Med. 2015 , 64 , 217–235. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Rahimy, E. Deep learning applications in ophthalmology. Curr. Opin. Ophthalmol. 2018 , 29 , 254–260. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Sisodia, D.S.; Nair, S.; Khobragade, P. Diabetic Retinal Fundus Images: Preprocessing and Feature Extraction for Early Detection of Diabetic Retinopathy. Biomed. Pharmacol. J. 2017 , 10 , 615–626. [ Google Scholar ] [ CrossRef ]
• Xiao, Z.; Zhang, X.; Geng, L.; Zhang, F.; Wu, J.; Tong, J.; Ogunbona, P.O.; Shan, C. Automatic non-proliferative diabetic retinopathy screening system based on color fundus image. Biomed. Eng. Online 2017 , 16 , 122. [ Google Scholar ] [ CrossRef ] [ PubMed ] [ Green Version ]
• Srivastava, R.; Duan, L.; Wong, D.W.; Liu, J.; Wong, T.Y. Detecting retinal microaneurysms and hemorrhages with robustness to the presence of blood vessels. Comput. Methods Programs Biomed. 2017 , 138 , 83–91. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Vanithamani, R.R.C.R.; Renee Christina, R. Exudates in detection and classification of diabetic retinopathy. In International Conference on Soft Computing and Pattern Recognition ; Springer: Cham, Germany, 2016; pp. 252–261. Available online: https://books.google.co.in/books?id=hFNuDwAAQBAJ&pg=PA108&lpg=PA108&dq=Vanithamani+R,+Renee+Christina+R+(2018)+Exudates+in+detection+and+classification+of+diabetic+retinopathy:+252–261&source=bl&ots=CWbiXEy9bP&sig=ACfU3U1vfBvwrh06MvSJbSKzMp8Sl2Cm4w&hl=en& (accessed on 5 April 2022).
• Wang, S.; Tang, H.L.; Al Turk, L.I.; Hu, Y.; Sanei, S.; Saleh, G.M.; Peto, T. Localizing Microaneurysms in Fundus Images Through Singular Spectrum Analysis. IEEE Trans. Biomed. Eng. 2016 , 64 , 990–1002. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Nijalingappa, P.; Sandeep, B. Machine learning approach for the identification of diabetes retinopathy and its stages. In Proceedings of the 2015 International Conference on Applied and Theoretical Computing and Communication Technology (iCATccT), Davangere, India, 29–31 October 2016; pp. 653–658. [ Google Scholar ] [ CrossRef ]
• Xiao, D.; Yu, S.; Vignarajan, J.; An, D.; Tay-Kearney, M.-L.; Kanagasingam, Y. Retinal hemorrhage detection by rule-based and machine learning approach. In Proceedings of the 2017 39th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Jeju Island, Republic of Korea, 11–15 July 2017; pp. 660–663. [ Google Scholar ] [ CrossRef ]
• Almotiri, J.; Elleithy, K.; Elleithy, A. Retinal Vessels Segmentation Techniques and Algorithms: A Survey. Appl. Sci. 2018 , 8 , 155. [ Google Scholar ] [ CrossRef ] [ Green Version ]
• Bui, T.; Maneerat, N.; Watchareeruetai, U. Detection of cotton wool for diabetic retinopathy analysis using neural network. In Proceedings of the IEEE 10th International Workshop on Computational Intelligence and Applications, Hiroshima, Japan, 11–12 November 2017; pp. 203–206. [ Google Scholar ] [ CrossRef ]
• Franklin, S.W.; Rajan, S.E. Computerized screening of diabetic retinopathy employing blood vessel segmentation in retinal images. Biocybern. Biomed. Eng. 2014 , 34 , 117–124. [ Google Scholar ] [ CrossRef ]
• Hanđsková, V.; Pavlovičova, J.; Oravec, M.; Blaško, R. Diabetic Rethinopathy Screening by Bright Lesions Extraction from Fundus Images. J. Electr. Eng. 2013 , 64 , 311–316. [ Google Scholar ] [ CrossRef ] [ Green Version ]
• Kavitha, M.; Palani, S. Hierarchical classifier for soft and hard exudates detection of retinal fundus images. J. Intell. Fuzzy Syst. 2014 , 27 , 2511–2528. [ Google Scholar ] [ CrossRef ]
• Paing, M.P.; Choomchuay, S.; Yodprom, M.D.R. Detection of lesions and classification of diabetic retinopathy using fundus images. In Proceedings of the 2016 9th Biomedical engineering international conference (BMEiCON), Laung Prabang, Laos, 7–9 December 2016; pp. 1–5. [ Google Scholar ] [ CrossRef ]
• Zhou, W.; Wu, C.; Chen, D.; Yi, Y.; Du, W. Automatic Microaneurysm Detection Using the Sparse Principal Component Analysis-Based Unsupervised Classification Method. IEEE Access 2017 , 5 , 2563–2572. [ Google Scholar ] [ CrossRef ]
• Kusakunniran, W.; Wu, Q.; Ritthipravat, P.; Zhang, J. Hard exudates segmentation based on learned initial seeds and iterative graph cut. Comput. Methods Programs Biomed. 2018 , 158 , 173–183. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Khade, S.; Gite, S.; Thepade, S.D.; Pradhan, B.; Alamri, A. Detection of Iris Presentation Attacks Using Feature Fusion of Thepade’s Sorted Block Truncation Coding with Gray-Level Co-Occurrence Matrix Features. Sensors 2021 , 21 , 7408. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Wen, L. Algorithms: A Comparative Study of Bagging, Boosting and Stacking Techniques. Remote Sens 2020 , 12 , 1683. [ Google Scholar ] [ CrossRef ]
• Brownlee, J. Stacking Ensemble Machine Learning with Python. In Machine Learning Mastery ; Machine Learning Mastery: San Francisco, CA, USA, 2020; Available online: https://machinelearningmastery.com/stacking-ensemble-machine-learning-with-python/ (accessed on 21 May 2022).
• Mane, V.M.; Jadhav, D.V.; Shirbahadurkar, S.D. Hybrid classifier and region-dependent integrated features for detection of diabetic retinopathy. J. Intell. Fuzzy Syst. 2017 , 32 , 2837–2844. [ Google Scholar ] [ CrossRef ]
• Fraz, M.M.; Jahangir, W.; Zahid, S.; Hamayun, M.M.; Barman, S.A. Multiscale segmentation of exudates in retinal images using contextual cues and ensemble classification. Biomed. Signal Process. Control. 2017 , 35 , 50–62. [ Google Scholar ] [ CrossRef ] [ Green Version ]
• Prentašić, P.; Lončarić, S. Detection of exudates in fundus photographs using deep neural networks and anatomical landmark detection fusion. Comput. Methods Programs Biomed. 2016 , 137 , 281–292. [ Google Scholar ] [ CrossRef ]
• Bala, M.P.; Vijayachitra, S. Early detection and classification of microaneurysms in retinal fundus images using sequential learning methods. Int. J. Biomed. Eng. Technol. 2014 , 15 , 128. [ Google Scholar ] [ CrossRef ]
• Sopharak, A.; Uyyanonvara, B.; Barman, S.; Williamson, T. Comparative Analysis of Automatic Exudate Detection between Machine Learning and Traditional Approaches. IEICE Trans. Inf. Syst. 2009 , 92 , 2264–2271. [ Google Scholar ] [ CrossRef ] [ Green Version ]
• Srinivasan, R.; Surya, J.; Ruamviboonsuk, P.; Chotcomwongse, P.; Raman, R. Influence of Different Types of Retinal Cameras on the Performance of Deep Learning Algorithms in Diabetic Retinopathy Screening. Life 2022 , 12 , 1610. [ Google Scholar ] [ CrossRef ]
• Valarmathi, S.; Vijayabhanu, R. A Survey on Diabetic Retinopathy Disease Detection and Classification using Deep Learning Techniques. In Proceedings of the 2021 IEEE 7th International Conference on Bio Signals, Images and Instrumentation, ICBSII, Chennai, India, 25–27 March 2021; pp. 1–4. [ Google Scholar ] [ CrossRef ]
• Wang, X.; Lu, Y.; Wang, Y.; Chen, W.-B. Diabetic Retinopathy Stage Classification Using Convolutional Neural Networks. In Proceedings of the 2018 IEEE 19th International Conference on Information Reuse and Integration for Data Science, IRI, Salt Lake City, UT, USA, 7–9 July 2018; pp. 465–471. [ Google Scholar ] [ CrossRef ]
• Chudzik, P.; Majumdar, S.; Calivá, F.; Al-Diri, B.; Hunter, A. Microaneurysm detection using fully convolutional neural networks. Comput. Methods Programs Biomed. 2018 , 158 , 185–192. [ Google Scholar ] [ CrossRef ]
• Yan, Y.; Gong, J.; Liu, Y. A Novel Deep Learning Method for Red Lesions Detection Using Hybrid Feature. In Proceedings of the 31st Chinese Control and Decision Conference, CCDC, Nanchang, China, 3–5 June 2019; pp. 2287–2292. [ Google Scholar ] [ CrossRef ]
• Guo, Y.; Liu, Y.; Oerlemans, A.; Lao, S.; Wu, S.; Lew, M.S. Deep learning for visual understanding: A review. Neurocomputing 2016 , 187 , 27–48. [ Google Scholar ] [ CrossRef ]
• Abbas, Q.; Ibrahim, M.E.A.; Jaffar, M.A. Video scene analysis: An overview and challenges on deep learning algorithms. Multimed. Tools Appl. 2017 , 77 , 20415–20453. [ Google Scholar ] [ CrossRef ]
• Gurcan, O.F.; Beyca, O.F.; Dogan, O. A Comprehensive Study of Machine Learning Methods on Diabetic Retinopathy Classification. Int. J. Comput. Intell. Syst. 2021 , 14 , 1132–1141. [ Google Scholar ] [ CrossRef ]
• Khade, S.; Gite, S.; Pradhan, B. Iris Liveness Detection Using Multiple Deep Convolution Networks. Big Data Cogn. Comput. 2022 , 6 , 67. [ Google Scholar ] [ CrossRef ]
• Ketkar, N.; Moolayil, J. Deep Learning with Python ; Manning Publications: New York, NY, USA, 2021. [ Google Scholar ] [ CrossRef ]
• Olivas, E.S.; Guerrero, J.D.M.; Martinez-Sober, M.; Magdalena-Benedito, J.R.; Serrano, L. Magdalena-Benedito, Handbook of Research on Machine Learning Applications and Trends: Algorithms, Methods, and Techniques ; IGI Global: Pennsylvania, PA, USA, 2010. [ Google Scholar ]
• Masood, S.; Luthra, T.; Sundriyal, H.; Ahmed, M. Identification of diabetic retinopathy in eye images using transfer learning. In Proceedings of the International Conference on Computing, Communication and Automation (ICCCA), Greater Noida, India, 5–6 May 2017; pp. 1183–1187. [ Google Scholar ] [ CrossRef ]
• Xu, X.; Lin, J.; Tao, Y.; Wang, X. An Improved DenseNet Method Based on Transfer Learning for Fundus Medical Images. In Proceedings of the 2018 7th international conference on digital home (ICDH), Guilin, China, 30 November–1 December 2018; pp. 137–140. [ Google Scholar ] [ CrossRef ]
• Lian, C.; Liang, Y.; Kang, R.; Xiang, Y. Deep Convolutional Neural Networks for Diabetic Retinopathy Classification. ACM Int. Conf. Proceeding Ser. 2018 , 72 , 68–72. [ Google Scholar ] [ CrossRef ]
• Blakely, M. ‘The Importance of Sight and Vision,’ Marvel Optics. 2015. Available online: https://www.marveloptics.com/blog/the-importance-of-sight-and-vision-molly-blakely/ (accessed on 7 April 2022).
• Quellec, G.; Charrière, K.; Boudi, Y.; Cochener, B.; Lamard, M. Deep image mining for diabetic retinopathy screening. Med. Image Anal. 2017 , 39 , 178–193. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Oliveira, A.; Pereira, S.; Silva, C. Retinal vessel segmentation based on Fully Convolutional Neural Networks. Expert Syst. Appl. 2018 , 112 , 229–242. [ Google Scholar ] [ CrossRef ] [ Green Version ]
• Orlando, J.I.; Prokofyeva, E.; del Fresno, M.; Blaschko, M.B. An ensemble deep learning based approach for red lesion detection in fundus images. Comput. Methods Programs Biomed. 2018 , 153 , 115–127. [ Google Scholar ] [ CrossRef ] [ Green Version ]
• Mahendran, G.; Dhanasekaran, R. Investigation of the severity level of diabetic retinopathy using supervised classifier algorithms. Comput. Electr. Eng. 2015 , 45 , 312–323. [ Google Scholar ] [ CrossRef ]
• Wu, D.; Zhang, M.; Liu, J.-C.; Bauman, W. On the Adaptive Detection of Blood Vessels in Retinal Images. IEEE Trans. Biomed. Eng. 2006 , 53 , 341–343. [ Google Scholar ] [ CrossRef ]
• Sánchez, C.I.; García, M.; Mayo, A.; Lopez, M.I.; Hornero, R. Retinal image analysis based on mixture models to detect hard exudates. Med. Image Anal. 2009 , 13 , 650–658. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• García, M.; Sánchez, C.I.; López, M.I.; Abásolo, D.; Hornero, R. Neural network based detection of hard exudates in retinal images. Comput. Methods Programs Biomed. 2009 , 93 , 9–19. [ Google Scholar ] [ CrossRef ] [ PubMed ] [ Green Version ]
• Sánchez, C.I.; Hornero, R.; López, M.I.; Aboy, M.; Poza, J.; Abásolo, D. A novel automatic image processing algorithm for detection of hard exudates based on retinal image analysis. Med. Eng. Phys. 2008 , 30 , 350–357. [ Google Scholar ] [ CrossRef ] [ PubMed ] [ Green Version ]
• Quellec, G.; Lamard, M.; Abramoff, M.; Decencière, E.; Lay, B.; Erginay, A.; Cochener, B.; Cazuguel, G. A multiple-instance learning framework for diabetic retinopathy screening. Med. Image Anal. 2012 , 16 , 1228–1240. [ Google Scholar ] [ CrossRef ]
• Köse, C.; Şevik, U.; Ikibaş, C.; Erdöl, H. Simple methods for segmentation and measurement of diabetic retinopathy lesions in retinal fundus images. Comput. Methods Programs Biomed. 2012 , 107 , 274–293. [ Google Scholar ] [ CrossRef ]
• Giancardo, L.; Meriaudeau, F.; Karnowski, T.P.; Li, Y.; Garg, S.; Tobin, K.W.; Chaum, E. Exudate-based diabetic macular edema detection in fundus images using publicly available datasets. Med. Image Anal. 2012 , 16 , 216–226. [ Google Scholar ] [ CrossRef ]
• Zhang, B.; Karray, F.; Li, Q.; Zhang, L. Sparse Representation Classifier for microaneurysm detection and retinal blood vessel extraction. Inf. Sci. 2012 , 200 , 78–90. [ Google Scholar ] [ CrossRef ]
• Qureshi, R.J.; Kovacs, L.; Harangi, B.; Nagy, B.; Peto, T.; Hajdu, A. Combining algorithms for automatic detection of optic disc and macula in fundus images. Comput. Vis. Image Underst. 2012 , 116 , 138–145. [ Google Scholar ] [ CrossRef ]
• Noronha, K.; Acharya, U.R.; Nayak, K.P.; Kamath, S.; Bhandary, S.V. Decision support system for diabetic retinopathy using discrete wavelet transform. Proc. Inst. Mech. Eng. Part H J. Eng. Med. 2012 , 227 , 251–261. [ Google Scholar ] [ CrossRef ]
• Gharaibeh, N.; Al-Hazaimeh, O.M.; Abu-Ein, A.; Nahar, K.M. A Hybrid SVM NAÏVE-BAYES Classifier for Bright Lesions Recognition in Eye Fundus Images. Int. J. Electr. Eng. Inform. 2021 , 13 , 530–545. [ Google Scholar ] [ CrossRef ]
• Al Hazaimeh, O.M.; Nahar, K.M.; Al Naami, B.; Gharaibeh, N. An effective image processing method for detection of diabetic retinopathy diseases from retinal fundus images. Int. J. Signal Imaging Syst. Eng. 2018 , 11 , 206. [ Google Scholar ] [ CrossRef ]
• Akram, M.U.; Khalid, S.; Khan, S.A. Identification and classification of microaneurysms for early detection of diabetic retinopathy. Pattern Recognit. 2013 , 46 , 107–116. [ Google Scholar ] [ CrossRef ]
• Harini, R.; Sheela, N. Feature extraction and classification of retinal images for automated detection of Diabetic Retinopathy. In Proceedings of the 2016 Second International Conference on Cognitive Computing and Information Processing (CCIP), Mysuru, India, 12–13 August 2016; Nagabhusan, T.N., Sundararajan, N., Suresh, S., Eds.; Sri Jayachamarajendra College of Engineering, JSS TI Campus: Mysuru, India, 2016; p. 570006. [ Google Scholar ]
• Umapathy, A.; Sreenivasan, A.; Nairy, D.S.; Natarajan, S.; Rao, B.N. Image Processing, Textural Feature Extraction and Transfer Learning based detection of Diabetic Retinopathy. In Proceedings of the 2019 9th International Conference on Bioscience, Biochemistry and Bioinformatics, Singapore, 7–9 January 2019; pp. 17–21. [ Google Scholar ] [ CrossRef ]
• Zago, G.T.; Andreão, R.V.; Dorizzi, B.; Salles, E.O.T. Diabetic retinopathy detection using red lesion localization and convolutional neural networks. Comput. Biol. Med. 2019 , 116 , 103537. [ Google Scholar ] [ CrossRef ]
• Jiang, H.; Yang, K.; Gao, M.; Zhang, D.; Ma, H.; Qian, W. An Interpretable Ensemble Deep Learning Model for Diabetic Retinopathy Disease Classification. In Proceedings of the 2019 41st Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Berlin, Germany, 23–27 July 2019; pp. 2045–2048. [ Google Scholar ] [ CrossRef ]
• Abràmoff, M.D.; Garvin, M.K.; Sonka, M. Retinal Imaging and Image Analysis. IEEE Rev. Biomed. Eng. 2010 , 3 , 169–208. [ Google Scholar ] [ CrossRef ]
• Gargeya, R.; Leng, T. Automated Identification of Diabetic Retinopathy Using Deep Learning. Ophthalmology 2017 , 124 , 962–969. [ Google Scholar ] [ CrossRef ]
• Doshi, D.; Shenoy, A.; Sidhpura, D.; Gharpure, P. Diabetic retinopathy detection using deep convolutional neural networks. In Proceedings of the 2016 International Conference on Computing, Analytics and Security Trends (CAST), Pune, India, 11 July 2016; pp. 261–266. [ Google Scholar ]
• Ghosh, R.; Ghosh, K.; Maitra, S. Automatic detection and classification of diabetic retinopathy stages using CNN. In Proceedings of the 2017 4th International Conference on Signal Processing and Integrated Networks (SPIN), Delhi, India, 26–27 August 2017; pp. 550–554. [ Google Scholar ] [ CrossRef ]
• Gondal, W.M.; Kohler, J.M.; Grzeszick, R.; Fink, G.A.; Hirsch, M. Weakly-supervised localization of diabetic retinopathy lesions in retinal fundus images. In Proceedings of the 2017 IEEE international conference on image processing (ICIP), Beijing, China, 17–20 September 2017; pp. 2069–2073. [ Google Scholar ] [ CrossRef ] [ Green Version ]
• Jiang, Y.; Wu, H.; Dong, J. Automatic Screening of Diabetic Retinopathy Images with Convolution Neural Network Based on Caffe Framework. In Proceedings of the 1st International Conference on Medical and Health Informatics 2017, Taichung city, Taiwan, 20–22 May 2017; pp. 90–94. [ Google Scholar ] [ CrossRef ]
• Prentasic, P.; Loncaric, S. Weighted ensemble based automatic detection of exudates in fundus photographs. IEEE 2014 , 2014 , 138–141. [ Google Scholar ] [ CrossRef ]
• Roy, P.; Tennakoon, R.; Cao, K.; Sedai, S.; Mahapatra, D.; Maetschke, S.; Garnavi, R. A novel hybrid approach for severity assessment of Diabetic Retinopathy in colour fundus images. In Proceedings of the 2017 IEEE 14th International Symposium on Biomedical Imaging (ISBI 2017), Melbourne, Australia, 18–21 April 2017; pp. 1078–1082. [ Google Scholar ] [ CrossRef ]
• Xu, K.; Feng, D.; Mi, H. Deep Convolutional Neural Network-Based Early Automated Detection of Diabetic Retinopathy Using Fundus Image. Molecules 2017 , 22 , 2054. [ Google Scholar ] [ CrossRef ] [ Green Version ]
• Yang, Y.; Li, T.; Li, W.; Wu, H.; Fan, W.; Zhang, W. Lesion Detection and Grading of Diabetic Retinopathy via Two-Stages Deep Convolutional Neural Networks. In International Conference on Medical Image Computing and Computer-Assisted Intervention ; Springer: Cham, Germany, 2017; pp. 533–540. [ Google Scholar ] [ CrossRef ] [ Green Version ]
• van Grinsven, M.J.J.P.; van Ginneken, B.; Hoyng, C.B.; Theelen, T.; Sanchez, C.I. Fast Convolutional Neural Network Training Using Selective Data Sampling: Application to Hemorrhage Detection in Color Fundus Images. IEEE Trans. Med. Imaging 2016 , 35 , 1273–1284. [ Google Scholar ] [ CrossRef ]
• Budak, U.; Şengür, A.; Guo, Y.; Akbulut, Y. A novel microaneurysms detection approach based on convolutional neural networks with reinforcement sample learning algorithm. Health Inf. Sci. Syst. 2017 , 5 , 14. [ Google Scholar ] [ CrossRef ]
• Zhou, W.; Wu, C.; Chen, D.; Wang, Z.; Yi, Y.; Du, W. Automatic Microaneurysms Detection Based on Multifeature Fusion Dictionary Learning. Comput. Math. Methods Med. 2017 , 2017 , 1–11. [ Google Scholar ] [ CrossRef ] [ PubMed ] [ Green Version ]
• Barkana, B.D.; Sariçiçek, I.; Yildirim, B. Performance analysis of descriptive statistical features in retinal vessel segmentation via fuzzy logic, ANN, SVM, and classifier fusion. Knowl. Based Syst. 2017 , 118 , 165–176. [ Google Scholar ] [ CrossRef ]
• Dasgupta, A.; Singh, S. A fully convolutional neural network based structured prediction approach towards the retinal vessel segmentation. In Proceedings of the 14th International Symposium on Biomedical Imaging (ISBI), Melbourne, VIC, Australia, 18–21 April 2017; pp. 248–251. [ Google Scholar ]
• Mo, J.; Zhang, L. Multi-level deep supervised networks for retinal vessel segmentation. Int. J. Comput. Assist. Radiol. Surg. 2017 , 12 , 2181–2193. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Tan, J.H.; Acharya, U.R.; Bhandary, S.V.; Chua, K.C.; Sivaprasad, S. Segmentation of optic disc, fovea, and retinal vasculature using a single convolutional neural network. J. Comput. Sci. 2017 , 20 , 70–79. [ Google Scholar ] [ CrossRef ] [ Green Version ]
• Vega, R.; Sanchez-Ante, G.; Falcon-Morales, L.E.; Sossa, H.; Guevara, E. Retinal vessel extraction using lattice neural networks with dendritic processing. Comput. Biol. Med. 2015 , 58 , 20–30. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Wang, S.; Yin, Y.; Cao, G.; Wei, B.; Zheng, Y.; Yang, G. Hierarchical retinal blood vessel segmentation based on feature and ensemble learning. Neurocomputing. Neurocomputing 2015 , 149 , 708–717. [ Google Scholar ] [ CrossRef ]
• Choi, J.Y.; Yoo, T.K.; Seo, J.G.; Kwak, J.; Um, T.T.; Rim, T.H. Multi-categorical deep learning neural network to classify retinal images: A pilot study employing small database. PLoS ONE 2017 , 12 , 16. [ Google Scholar ] [ CrossRef ] [ PubMed ] [ Green Version ]
• Mumtaz, R.; Hussain, M.; Sarwar, S.; Khan, K.; Mumtaz, S.; Mumtaz, M. Automatic detection of retinal hemorrhages by exploiting image processing techniques for screening retinal diseases in diabetic patients. Int. J. Diabetes Dev. Ctries. 2017 , 38 , 80–87. [ Google Scholar ] [ CrossRef ]
• Santhi, D.; Manimegalai, D.; Parvathi, S.; Karkuzhali, S. Segmentation and classification of bright lesions to diagnose diabetic retinopathy in retinal images. Biomed. Eng. Biomed. Tech. 2016 , 61 , 443–453. [ Google Scholar ] [ CrossRef ]
• Li, G.; Zheng, S.; Li, X. Exudate Detection in Fundus Images via Convolutional Neural Network. In International Forum on Digital TV and Wireless Multimedia Communications ; Springer: Singapore, 2018; pp. 193–202. [ Google Scholar ] [ CrossRef ]
• Bala, P.; Vijayachitra, S. A Sequential learning method for detection and classification of exudates in retinal images to assess diabetic retinopathy. J. Biol. Syst. 2014 , 22 , 413–428. [ Google Scholar ] [ CrossRef ]
• Rahim, S.S.; Jayne, C.; Palade, V.; Shuttleworth, J. Automatic detection of microaneurysms in colour fundus images for diabetic retinopathy screening. Neural Comput. Appl. 2015 , 27 , 1149–1164. [ Google Scholar ] [ CrossRef ]
• Omar, M.; Khelifi, F.; Tahir, M.A. Detection and classification of retinal fundus images exudates using region based multiscale LBP texture approach. In Proceedings of the 2016 International Conference on Control, Decision and Information Technologies (CoDIT), Saint Julian’s, Malta, 6–8 April 2016; pp. 227–232. [ Google Scholar ] [ CrossRef ]
• Abbas, Q.; Fondon, I.; Sarmiento, A.; Jiménez, S.; Alemany, P. Automatic recognition of severity level for diagnosis of diabetic retinopathy using deep visual features. Med. Biol. Eng. Comput. 2017 , 55 , 1959–1974. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Ouyang, W.; Luo, P.; Zeng, X.; Qiu, S.; Tian, Y.; Li, H.; Yang, S.; Wang, Z.; Xiong, Y.; Qian, C.; et al. Deepid-net: Multi-stage and deformable deep convolutional neural networks for object detection. arXiv 2014 , arXiv:1409.3505. [ Google Scholar ]
• Shan, J.; Li, L. A deep learning method for microaneurysm detection in fundus images. In Proceedings of the IEEE First International Conference on Connected Health: Applications, Systems, and Engineering Technologies, Washington, DC, USA, 27–29 June 2016; pp. 357–358. [ Google Scholar ]
• Shirbahadurkar, S.D.; Mane, V.M.; Jadhav, D.V. Early Stage Detection of Diabetic Retinopathy Using an Optimal Feature Set. In Advances in Intelligent Systems and Computing ; Springer: Berlin/Heidelberg, Germany, 2017; Volume 678, pp. 15–23. [ Google Scholar ] [ CrossRef ]
• SK, S. A Machine Learning Ensemble Classifier for Early Prediction of Diabetic Retinopathy. J. Med. Syst. 2017 , 41 , 201. [ Google Scholar ] [ CrossRef ]
• Abràmoff, M.D.; Lou, Y.; Erginay, A.; Clarida, W.; Amelon, R.; Folk, J.C.; Niemeijer, M. Improved Automated Detection of Diabetic Retinopathy on a Publicly Available Dataset Through Integration of Deep Learning. Investig. Ophthalmol. Vis. Sci. 2016 , 57 , 5200–5206. [ Google Scholar ] [ CrossRef ] [ PubMed ] [ Green Version ]
• Antal, B.; Hajdu, A. An ensemble-based system for automatic screening of diabetic retinopathy. Knowl. Based Syst. 2014 , 60 , 20–27. [ Google Scholar ] [ CrossRef ] [ Green Version ]
• Carrera, E.V.; Gonzalez, A.; Carrera, R. Automated detection of diabetic retinopathy using SVM. In Proceedings of the IEEE XXIV International Conference on Electronics, Electrical Engineering and Computing (INTERCON), Cusco, Peru, 15–18 August 2017; pp. 1–4. [ Google Scholar ] [ CrossRef ]
• Gegundez-Arias, M.E.; Marin, D.; Ponte, B.; Alvarez, F.; Garrido, J.; Ortega, C.; Vasallo, M.J.; Bravo, J.M. A tool for automated diabetic retinopathy pre-screening based on retinal image computer analysis. Comput. Biol. Med. 2017 , 88 , 100–109. [ Google Scholar ] [ CrossRef ]
• Li, X.; Pang, T.; Xiong, B.; Liu, W.; Liang, P.; Wang, T. Convolutional neural networks based transfer learning for diabetic retinopathy fundus image classification. In Proceedings of the 10th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI), Shanghai, China, 14–16 October 2017. [ Google Scholar ]
• Arunkumar, R.; Karthigaikumar, P. Multi-retinal disease classification by reduced deep learning features. Neural Comput. Applic. 2017 , 28 , 329–334. [ Google Scholar ] [ CrossRef ]
• Tan, J.H.; Fujita, H.; Sivaprasad, S.; Bhandary, S.V.; Rao, A.K.; Chua, K.C.; Acharya, U.R. Automated segmentation of exudates, hemorrhages, and microaneurysms using a single convolutional neural network. Inf. Sci. 2017 , 420 , 66–76. [ Google Scholar ] [ CrossRef ]
• Takahashi, H.; Tampo, H.; Arai, Y.; Inoue, Y.; Kawashima, H. Applying artificial intelligence to disease staging: Deep learning for the improved staging of diabetic retinopathy. PLoS ONE 2017 , 12 , e0179790. [ Google Scholar ] [ CrossRef ] [ Green Version ]
• Hemanth, D.J.; Anitha, J.; Indumathy, A. Diabetic Retinopathy Diagnosis in Retinal Images Using Hopfield Neural Network. IETE J. Res. 2016 , 62 , 893–900. [ Google Scholar ] [ CrossRef ]
• Sharma, S.; Singh, G.; Sharma, M. A comprehensive review and analysis of supervised-learning and soft computing techniques for stress diagnosis in humans. Comput. Biol. Med. 2021 , 134 , 104450. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Lakshminarayanan, V.; Kheradfallah, H.; Sarkar, A.; Balaji, J.J. Automated Detection and Diagnosis of Diabetic Retinopathy: A Comprehensive Survey. J. Imaging 2021 , 7 , 165. [ Google Scholar ] [ CrossRef ]
• Shultz, T.R.; Fahlman, S.E.; Craw, S.; Andritsos, P.; Tsaparas, P.; Silva, R.; Drummond, C.; Lanzi, P.L.; Gama, J.; Wiegand, R.P. Confusion Matrix. Encycl. Mach. Learn. 2011 , 61 , 209. [ Google Scholar ]
• Wikipedia, F. Cohen Kappa. 2006. Available online: https://thenewstack.io/cohens-kappa-what-it-is-when-to-use-it-and-how-to-avoid-its-pitfalls (accessed on 12 March 2022).
• Hernández, C.; Porta, M.; Bandello, F.; Grauslund, J.; Harding, S.P.; Aldington, S.J.; Egan, C.; Frydkjaer-Olsen, U.; García-Arumí, J.; Gibson, J.; et al. The Usefulness of Serum Biomarkers in the Early Stages of Diabetic Retinopathy: Results of the EUROCONDOR Clinical Trial. J. Clin. Med. 2020 , 9 , 1233. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Jacoba, C.M.P.; Celi, L.A.; Silva, P.S. Biomarkers for Progression in Diabetic Retinopathy: Expanding Personalized Medicine through Integration of AI with Electronic Health Records. Semin. Ophthalmol. 2021 , 36 , 250–257. [ Google Scholar ] [ CrossRef ]
• Records, H. HHS Public Access. Biomarkers 2022 , 36 , 250–257. [ Google Scholar ]
• Control, D.; Trial, C. Progression of Retinopathy with Intensive versus Conventional Treatment in the Diabetes Control and Complications Trial. Ophthalmology 1995 , 102 , 647–661. [ Google Scholar ] [ CrossRef ]
• Group, A.S.; Group, A.E.S.; Chew, E.Y.; Ambrosius, W.T.; Davis, M.D.; Danis, R.P.; Gangaputra, S.; Greven, C.M.; Hubbard, L.; Esser, B.A.; et al. Effects of Medical Therapies on Retinopathy Progression in Type 2 Diabetes. N. Engl. J. Med. 2010 , 363 , 233–244. [ Google Scholar ] [ CrossRef ]
• Kuo, J.Z.; Wong, T.Y.; Rotter, J.I. Challenges in elucidating the genetics of diabetic retinopathy. JAMA Ophthalmol. 2014 , 132 , 96–107. [ Google Scholar ] [ CrossRef ] [ Green Version ]
• Mastropasqua, R.; Toto, L.; Cipollone, F.; Santovito, D.; Carpineto, P.; Mastropasqua, L. Role of microRNAs in the modulation of diabetic retinopathy. Prog. Retin. Eye Res. 2014 , 43 , 92–107. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Cooper, M.E.; El-Osta, A. Epigenetics. Circ. Res. 2010 , 107 , 1403–1413. [ Google Scholar ] [ CrossRef ] [ PubMed ] [ Green Version ]
• Torok, Z.; Peto, T.; Csosz, E.; Tukacs, E.; Molnar, A.M.; Berta, A.; Tozser, J.; Hajdu, A.; Nagy, V.; Domokos, B.; et al. Combined Methods for Diabetic Retinopathy Screening, Using Retina Photographs and Tear Fluid Proteomics Biomarkers. J. Diabetes Res. 2015 , 2015 , 1–8. [ Google Scholar ] [ CrossRef ] [ PubMed ] [ Green Version ]
• Lu, C.-H.; Lin, S.-T.; Chou, H.-C.; Lee, Y.-R.; Chan, H.-L. Proteomic analysis of retinopathy-related plasma biomarkers in diabetic patients. Arch. Biochem. Biophys. 2013 , 529 , 146–156. [ Google Scholar ] [ CrossRef ]
• Xia, J.-F.; Wang, Z.-H.; Liang, Q.-L.; Wang, Y.-M.; Li, P.; Luo, G.-A. Correlations of six related pyrimidine metabolites and diabetic retinopathy in Chinese type 2 diabetic patients. Clin. Chim. Acta 2011 , 412 , 940–945. [ Google Scholar ] [ CrossRef ] [ PubMed ]
• Hussain, F.; Hussain, R.; Hossain, E. Explainable Artificial Intelligence (XAI): An Engineering Perspective. arXiv 2021 , arXiv:2101.03613. Available online: http://arxiv.org/abs/2101.03613 (accessed on 21 March 2022).
• Jang, S.I.; Girard, M.J.; Thiéry, A.H. Thiery, Explainable diabetic retinopathy classification based on neural-symbolic learning. CEUR Workshop Proc. 2021 , 2986 , 104–113. [ Google Scholar ]
• Deshpande, N.M.; Gite, S.; Pradhan, B.; Assiri, M.E. Explainable Artificial Intelligence–A New Step towards the Trust in Medical Diagnosis with AI Frameworks: A Review. Comput. Model. Eng. Sci. 2022 , 133 , 1–30. [ Google Scholar ] [ CrossRef ]
• Leopold, H.A.; Singh, A.; Sengupta, S.; Zelek, J.S.; Lakshminarayanan, V. Recent advances in deep learning applications for retinal diagnosis using OCT. 2020. In State of the Art in Neural Networks ; Elsevier: New York, NY, USA, 2020. [ Google Scholar ]
• Liu, R.; Li, Q.; Xu, F.; Wang, S.; He, J.; Cao, Y.; Shi, F.; Chen, X.; Chen, J. Application of artificial intelligence-based dual-modality analysis combining fundus photography and optical coherence tomography in diabetic retinopathy screening in a community hospital. Biomed. Eng. Online 2022 , 21 , 1–11. [ Google Scholar ] [ CrossRef ]
• Nguyen, D.M.H.; Mai, T.T.N.; Than, N.T.T.; Prange, A.; Sonntag, D. Self-supervised Domain Adaptation for Diabetic Retinopathy Grading Using Vessel Image Reconstruction. In German Conference on Artificial Intelligence (Künstliche Intelligenz) ; Springer: Cham, Germany, 2021; pp. 349–361. [ Google Scholar ] [ CrossRef ]
• Song, R.; Cao, P.; Yang, J.; Zhao, D.; Zaiane, O.R. A Domain Adaptation Multi-instance Learning for Diabetic Retinopathy Grading on Retinal Images. In Proceedings of the 2020 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), Seoul, Republic of Korea, 16–19 December 2020; pp. 743–750. [ Google Scholar ] [ CrossRef ]
• Crawshaw, M. Multi-task learning with deep neural networks: A survey. arXiv 2020 , arXiv:2009.09796. [ Google Scholar ]
• Foo, A.; Hsu, W.; Lee, M.L.; Lim, G.; Wong, T.Y. Multi-Task Learning for Diabetic Retinopathy Grading and Lesion Segmentation. Proc. Conf. AAAI Artif. Intell. 2020 , 34 , 13267–13272. [ Google Scholar ] [ CrossRef ]

Share and Cite

Bidwai, P.; Gite, S.; Pahuja, K.; Kotecha, K. A Systematic Literature Review on Diabetic Retinopathy Using an Artificial Intelligence Approach. Big Data Cogn. Comput. 2022 , 6 , 152. https://doi.org/10.3390/bdcc6040152

Bidwai P, Gite S, Pahuja K, Kotecha K. A Systematic Literature Review on Diabetic Retinopathy Using an Artificial Intelligence Approach. Big Data and Cognitive Computing . 2022; 6(4):152. https://doi.org/10.3390/bdcc6040152

Bidwai, Pooja, Shilpa Gite, Kishore Pahuja, and Ketan Kotecha. 2022. "A Systematic Literature Review on Diabetic Retinopathy Using an Artificial Intelligence Approach" Big Data and Cognitive Computing 6, no. 4: 152. https://doi.org/10.3390/bdcc6040152

Article Metrics

Article access statistics, further information, mdpi initiatives, follow mdpi.

Subscribe to receive issue release notifications and newsletters from MDPI journals

• Open access
• Published: 03 February 2024

Present and future screening programs for diabetic retinopathy: a narrative review

• Andreas Abou Taha 1 ,
• Sebastian Dinesen 1 , 2 , 3 ,
• Anna Stage Vergmann 1 , 2 &
• Jakob Grauslund 1 , 2 , 3  

International Journal of Retina and Vitreous volume  10 , Article number:  14 ( 2024 ) Cite this article

1298 Accesses

2 Citations

1 Altmetric

Metrics details

Diabetes is a prevalent global concern, with an estimated 12% of the global adult population affected by 2045. Diabetic retinopathy (DR), a sight-threatening complication, has spurred diverse screening approaches worldwide due to advances in DR knowledge, rapid technological developments in retinal imaging and variations in healthcare resources.

Many high income countries have fully implemented or are on the verge of completing a national Diabetic Eye Screening Programme (DESP). Although there have been some improvements in DR screening in Africa, Asia, and American countries further progress is needed. In low-income countries, only one out of 29, partially implemented a DESP, while 21 out of 50 lower-middle-income countries have started the DR policy cycle. Among upper-middle-income countries, a third of 59 nations have advanced in DR agenda-setting, with five having a comprehensive national DESP and 11 in the early stages of implementation.

Many nations use 2–4 fields fundus images, proven effective with 80–98% sensitivity and 86–100% specificity compared to the traditional seven-field evaluation for DR. A cell phone based screening with a hand held retinal camera presents a potential low-cost alternative as imaging device. While this method in low-resource settings may not entirely match the sensitivity and specificity of seven-field stereoscopic photography, positive outcomes are observed.

Individualized DR screening intervals are the standard in many high-resource nations. In countries that lacks a national DESP and resources, screening are more sporadic, i.e. screening intervals are not evidence-based and often less frequently, which can lead to late recognition of treatment required DR.

The rising global prevalence of DR poses an economic challenge to nationwide screening programs AI-algorithms have showed high sensitivity and specificity for detection of DR and could provide a promising solution for the future screening burden.

In summary, this narrative review enlightens on the epidemiology of DR and the necessity for effective DR screening programs. Worldwide evolution in existing approaches for DR screening has showed promising results but has also revealed limitations. Technological advancements, such as handheld imaging devices, tele ophthalmology and artificial intelligence enhance cost-effectiveness, but also the accessibility of DR screening in countries with low resources or where distance to or a shortage of ophthalmologists exists.

Diabetic retinopathy (DR) is a recognized sight-threatening complication to diabetes and is recommended for screening by the World Health Organization [ 1 , 2 , 3 ]. Various studies confirm the cost-effectiveness of DR screening despite variations in approaches and the availability of diverse imaging technologies across different countries [ 1 , 4 ].

The imperative for DR screening is expected to escalate concomitantly with the rising prevalence of diabetes [ 5 ]. As a consequence, it is crucial to elucidate the current status of DR screening at a global scale. This narrative review is necessary to assess the prevailing levels of awareness, accessibility and implementation of DR screening programs worldwide. Understanding the existing landscape will provide insight into the adequacy of the current screening measures and highlight areas that may require enhancement to effectively address the growing prevalence of diabetes and its associated ocular complications.

This narrative review aims to provide a global perspective of DR epidemiology and screening while exploring new approaches alongside development of artificial intelligence (AI) technology.

Data sources

This narrative review aims to be as comprehensive as possible in identifying data. The sources used for identification of literature were MEDLINE, Embase and The Cochrane Database of Systematic reviews. We used the search terms “diabetic retinopathy”, “screening of diabetic retinopathy”, “prevalence of diabetic retinopathy”, “incidence of diabetic retinopathy”, “artificial intelligence”, “deep learning” and obtained information on ongoing DR screening programs, not published in scientific journals, from the official pages of the World Health Organization and the International Diabetes Federation.

Inclusion criteria

Studies and reports focusing on epidemiology and screening programs for DR were considered for inclusion. A comprehensive approach involved the inclusion of both cross-sectional and longitudinal studies to investigate the prevalence of DR, but we considered only longitudinal studies for exploration of DR and PDR incidences. Inclusion criteria encompassed studies that (1) examined national or subnational DR screening programs regardless of economic status, (2) investigated epidemiology of DR and (3) were written in English.

Global epidemiology of DR

It is estimated that diabetes affects 783 million people aged 20–79 years worldwide by 2045, which equals 12.2% of the global adult population [ 5 ]. As life expectancy continues to rise and prevalence of diabetes increases, the prevalence of DR is expected to rise alongside [ 6 ]. Figure  1 displays the global prevalence of DR and vision threatening DR in both type 1 and 2 diabetes according to data from two comprehensive systematic reviews [ 6 , 7 ].

The figure illustrates the global prevalence of diabetic retinopathy, including vision-threatening cases, based on populations-based systematic reviews [ 6 , 7 ]. The proportion of vision-threatening cases are highlighted above the overall prevalence of diabetic retinopathy. DR diabetic retinopathy, VTDR vision threatening diabetic retinopathy. NAC North America and Caribbean, SACA South and Central America, EUR Europe, MENA Middle East and Northern Africa, AFR Africa, SEA South East Asia, WP Western Pacific

Incidence of DR in type 1 diabetes

Most studies investigating the incidence rates of DR in patients with type 1 diabetes are of older date [ 8 , 9 , 10 ]. These include a study by Klein et al. from 1989 that reported a 4 year incidence of DR in type 1 diabetes of 59.0% [ 8 ], a European study from 1986 that reported a 5 year incidence of 47.0% [ 9 ] and a Swedish study from 2003 that found a 10 year incidence of 39% [ 10 ].

Incidence of DR in type 2 diabetes

A Danish cohort study of patients with type 2 diabetes who attended the Danish screening program for DR showed a 5 year incidence of 3.8% for DR [ 11 ] compared to 4% in a study from United Kingdom (UK) [ 12 ]. A group from India reported a 4 year incidence of 9.2% [ 13 ], while studies of older date from Hong Kong [ 14 ], Australia [ 15 ] and USA [ 16 ] reported substantially higher 5 year incidences of 15.2, 22.2 and 38.6%, respectively.

Incidence of PDR

We identified 11 population-based studies that investigated the incidence of PDR over various follow-up periods of four [ 8 , 16 , 17 , 18 ], five [ 11 , 19 , 20 , 21 , 22 ], nine [ 23 ], ten [ 10 ] and 25 years [ 24 ]. Among these studies, three focused on patients with type 1 diabetes [ 8 , 10 , 24 ], five on patients with type 2 diabetes [ 11 , 16 , 18 , 20 , 23 ], while four studies encompassed populations that comprised patients with both type 1 and 2 diabetes [ 17 , 19 , 21 , 22 ]. The incidences of PDR are displayed in Fig.  2 arranged chronologically based on the year marking the baseline date. Starting with the earliest study and progressing to the latest, the line of tendency shows that the incidence of PDR has significantly declined over the 32 year period.

Trends in the incidence of proliferative diabetic retinopathy in population-based studies of type 1 α or 2 diabetes β and some including both types of diabetes γ . The year marks the baseline date of each follow-up period and the uppercase number is the reference number. PDR proliferative diabetic retinopathy

Current DR screening recommendations

Countries like Iceland, UK, Ireland, and Denmark have national diabetic eye screening programmes (DESP) [ 1 , 4 ]. Many nations are making considerable progress in developing regional screening and treatment services including Norway, Sweden, the Netherlands, Czech Republic, Italy, Poland, Serbia, Hungary, Turkey and others that can be studied in Table  1 [ 4 ].

A consultative group of the International Agency for the Prevention of Blindness (IAPB) categorized 10 South-East Asia countries (SEAC) into low (Myanmar and Timor-Leste), middle (Bhutan, Indonesia, Maldives, Myanmar, Nepal and Sri Lanka) and high resource (Thailand and India) level and made recommendations of DR management [ 25 ]. Even though only four of these countries (India, Nepal, Sri Lanka and Thailand) have developed national DR guidelines, the middle resource countries have made improvements in DR screening due to the increasing prevalence of diabetes and its complication [ 25 , 26 ].

However, DR screening in SEACs, like many low-, lower-middle- and upper-middle-income countries that lacks a national DESP, is sporadic and the methods are either screening camps, telemedicine vans, opportunistic screening, or physician-led screenings utilizing direct ophthalmoscopy, where only a small number of patients undergoes mydriatic imaging [ 26 , 27 ]. A review authored by Vujosevic et al. [ 4 ], observed advancements in DR screening in African and Asian countries, which included Botswana, China, Singapore, Indonesia, and Bangladesh.

In examining the management of DR in various Middle- and South American countries, a disparity in resource allocation and healthcare provision become evident [ 28 , 29 , 30 , 31 , 32 ].

Peru has established a comprehensive DR referral network in La Libertad, but still faces challenges in in impeding widespread screening and treatment, due to limited resources, thus categorizing it as a low-resource nation [ 28 ].

The middle-resource countries Argentina [ 33 ], Brazil [ 34 ], Costa Rica [ 29 ], and Mexico [ 30 , 31 ] demonstrates different stages of progress in DR screening and management. However, they share common struggles in achieving uniformity and comprehensive access in DR care [ 32 ]. Argentina's particular challenge lies in its highly distorted economy with persistently high inflation and a massive fiscal deficit, which had led to economic constraints and had impacted the national healthcare strategies for diabetes and DR [ 33 ]. On the other hand, Mexico [ 30 , 31 ] and Costa Rica [ 29 ] both underscore the need for improved healthcare strategies and effective disease management.

A study from Costa Rica [ 29 ] showed that 23.5% of individuals with diabetes had retinopathy and/or maculopathy, with 6.2% having Vision-Threatening DR. The study urges the need for improvement, especially among the older population, in DR screening methods and management e.g. the authors suggests that conventional screening methods like direct ophthalmoscopy have low sensitivity and may not be as effective.

Another study found that the coverage of DR screening among diabetic patients in Brazil [ 34 ] has increased from 12.1% in 2014 to 21.2% in 2019. Nevertheless, it was concluded that further progress is required in these regions, due to the fact that screenings of DR often are private insurance-based health care, decentralized health care screening, or has not been expanded from localized regions to encompass the entire nation, and therefore with significant regional differences [ 34 ].

The studies illustrate the broader challenge in the Latin American context, where the growth in DR screening is yet to meet the needs dictated by the prevalence and complexity of the condition.

SEAC, Middle- and South American countries need more widespread access to trained staff in order to have a fully functional DESP [ 4 , 35 ] e.g. Thailand face challenges due to the long distance to specialized medical practices, while China is challenged with only 20 ophthalmologists per one million people in contrast to 49 in the UK and 59 in the USA. [ 36 ].

The USA also faces challenges due to the fact that DR screening differs across different states and are insurance-based. Many health insurance plans, including the national Medicare, typically cover annual diabetic eye exams, but not all insurances cover more intensive follow-up or economic loss due to less work that particular day, which can lead to the screening being influenced by personal economics instead of evidence based recommendations [ 37 , 38 , 39 , 40 ].

Curran et al. [ 41 ] found that of 29 identified low-income countries, only four had data available on DR policy planning, and just one had a partially rolled out DESP. Among the 50 lower-middle-income countries, 21 had begun a DR policy cycle, with a single nation having a national DESP and 18 with DESPs in the early implementation phase. For Upper-Middle-Income Countries, 22 out of 59 countries had advanced in DR agenda-setting. Only five of these Upper-Middle-Income Countries had a comprehensive national DESP in place with 11 more in the partially implemented stages of DESP.

Practical approaches

Visual acuity.

Several DESPs incorporate the assessment of visual acuity as a part of the screening routine [ 1 , 42 ], but is not sufficiently sensitive to stand alone [ 43 ]. This limitation arises from the fact that a significant number of patients may remain asymptomatic until vision threatening DR manifests, often precipitated by vitreous bleeding or clinically significant diabetic macula edema.

Classification scales

Several classification systems of DR exist e.g. English National Screening Programme -, Wisconsin Diabetic Retinopathy—and the Scottish Diabetic retinopathy grading system. The gold standard in classification of DR in clinical trials has traditionally been the Early Treatment of Diabetic Retinopathy Study (ETDRS) classification scale [ 2 , 44 , 45 ]. This is an evidence based approach to screening and have demonstrated its effectiveness in predicting the risk of progression to proliferative DR and vision loss. However, the ETDRS scale use in a clinical settings is limited due to its complexity and many levels of DR classification. A simplified version of the ETDRS system, the International Clinical Diabetic Retinopathy (ICDR) scale [ 46 ], is recommended by various international clinical guidelines, which includes the guidelines established by the International Council of Ophthalmology (ICO) for everyday clinical practice [ 2 ]. Therefore, several nations use the ICDR severity scale in DR screening worldwide. The ICDR severity scale categorise DR into following levels accordingly to the severity of DR.

Level 0 is the absence of DR. Level 1 is mild non-proliferative DR (NPDR) characterized exclusively by microaneurysms and/or dot haemorrhages. Level 2 representing moderate NPDR, which is defined as more severe than level 1 but less than level 3. Level 3 indicates severe NPDR, where there’s observed more than 20 intraretinal haemorrhages in each of the four quadrant, or definite venous beading in at least 2 quadrants or prominent intraretinal microvascular abnormalities in at least 1 quadrant, but no proliferative DR. Level 4: signifies proliferative DR [ 46 ].

Standard fundus images

The gold standard for evaluating DR in clinical trials has traditionally been ETDRS seven-standard fields, which are a compilation of seven stereoscopic 30-degree fundus images [ 44 , 45 ]. A review from 2020 [ 4 ] found that a limited number of fundus images (typically two to four) exhibit a sensitivity ranging from 80 to 98% and a specificity between 86 and 100% when compared to the results obtained from ETDRS seven-fields in detecting DR. Conversely, a single central field was found to have lower sensitivity (ranging from 54 to 78%) and specificity (between 88 and 89%) when compared to the results of the ETDRS seven-standard fields.

The use of limited single-field fundus photos [ 4 ] is found to be effective, especially considering the difficulties, expenses, and time constraints associated with performing the ETDRS seven-standard fields, making it impractical for routine screening [ 1 ]. As a result, most Western nations rely on the simplicity and efficiency of limited single-field fundus photos, covering around 30% of the retinal surface [ 42 ].

The Danish and UK guidelines recommend a minimum two-field mydriatic fundus photos for DR screening. The retinal images should encompass a minimum horizontal field of view of 45° in the UK and 70–80° in Denmark. The vertical coverage should be at least 40° in the UK and 45° in Denmark [ 1 , 42 ].

The recommendation from the IAPB in SEAC is that low resource SEACs uses a minimum of four-field non-mydriatic fundus photos with a 30° camera, whereas middle- and high resource SEACs uses minimum two-field non-mydriatic fundus photos with wide-field (50°or more) camera [ 25 ].

Alternative image modalities

A recent alternative technique has emerged, known as the cell phone-based approach. This method involves utilizing a handheld condensing lens in combination with a smartphone camera to capture retinal images [ 47 , 48 , 49 ].

A review [ 4 ] conducted in Western Australia, where handheld retinal cameras were introduced for community-based clinical assessments of DR in low-resource settings, demonstrated positive result and thereby showed a potential for such systems to expand eye care services to underserved areas and remote locations.

No handheld devices have yet matched the sensitivity and specificity of seven-field stereoscopic photography in detecting sight-threatening DR. Rajalakshmi et al. [ 49 ] conducted a comparative study in which they evaluated the performance of a smartphone-based retinal camera against seven-field digital retinal photography. It was found that these methods produced identical results in 92.7% of patients, with a substantial kappa statistic of 0.90. Jacoba et al. [ 50 ] discovered that depending on the referral threshold, up to 37.0% of individual eyes with PDR might remain undetected when utilizing handheld photos.

This aligns with the result of several studies [ 51 , 52 ] which found high agreement in DR classification and image quality between handheld fundus cameras with standard tabletop fundus cameras for DR. However, disagreements in microaneurysms, small hemorrhages, and intraretinal microvascular abnormalities, contributed to the higher discordance within non-proliferative DR due to the decreased resolution in the retinal microvasculature. Moreover, the studies found that for referable DR and vision-threatening DR the agreement was 85% with only a substantial kappa statistic of 0.7. Consequently, they recommended lowering the referral thresholds to an eye-centre when utilizing handheld devices.

On the other hand, DR detection in smartphone-based fundus photography using AI [ 52 , 53 , 54 ] showed high sensitivity and specificity in detecting DR and sight-threatening DR, suggesting that AI-based smartphone retinal imaging could be a valuable tool for mass retinal screening in diabetes.

Screening intervals

The ICO recommends that the interval between DR screenings varies from 1 month to 2 years depending on the patients DR severity according to the ICDR Severity Scale. In recent years national guidelines for DR screening in several Western countries have shifted from fixed or sporadic screening intervals to a more individualized approach. These updated guidelines seek to optimize healthcare resources by adjusting screening intervals according to individualized risks of DR progression. Factors like glycaemic control and blood pressure are taken into account, leading to shorter or longer screening intervals, even when patients fall within the same severity group on the ICDR scale [ 1 , 42 , 55 ].

In SEACs patients with mild non-proliferative diabetic retinopathy (NPDR), the recommended screening interval according to IAPB is one year. This is the same for moderate NPDR [ 27 ]. Another study [ 32 ] recommend screening interval of 2 years in case of no- and mild NPDR in low- and middle income countries, and 1 year interval for moderate NPDR. In the case of proliferative DR, once the condition is stabilized, the recommended interval is six months [ 26 , 27 , 32 ].

Automated retinal image analysis

The availability of resources for nationwide screening programs remains limited in many countries. Nevertheless, advancements in technology have emerged as game-changers in screening strategies, enhancing cost-effectiveness. Technologies like scanning confocal ophthalmology with UWF, handheld mobile devices, tele ophthalmology for remote grading, and AI for automated detection and classification of DR have transformed screening approaches. These innovations are not only improving the efficiency of screening, but also contributing to better cost-effectiveness outcomes [ 4 ].

Four reviews [ 36 , 56 , 57 , 58 ] refers to numerous studies where deep learning (DL) algorithms have consistently demonstrated remarkable high sensitivity, specificity, and area under the receiver operating characteristic curve for detection of DR, Diabetic macula edema and other eye conditions like glaucoma as well as age-related macular degeneration (AMD).

Cheung et al. [ 56 ] suggests that the use of DL technology could reduce costs and improve access while enhancing patient outcomes through early detection and treatment of DR. Ballemo et al. [ 36 ] highlights that studies on DL have been conducted across various countries, showing promising results. The obstacles related to current DR screening and the motivations for implementing AI differ across nations.

The motivation for adopting AI in countries with well-established healthcare systems like the UK and Singapore is to sustain a high-quality healthcare service for patients while optimizing available resources [ 36 ].

AI could be an instrument capable of enhancing screening availability to people in lower-middle-income countries, but also to countries with long distances to specialized medical practices, and to countries with low numbers of ophthalmologists per million people [ 36 , 57 ]. Studies have looked at AI for photo analysis of fundus photo taken by handheld smartphone devices at patients with dilated pupils [ 54 , 59 , 60 ]. Using a four-fundus image [ 59 ], a two-image [ 54 ], and a single-image [ 60 ] approach showed good results with automatic AI screening, even though Penha et al. [ 60 ] emphasize that it has been established that a single image protocol loses diagnostic accuracy in comparison to a two-image protocol when it comes to expert human reading. However, with automatic reading using AI systems, the performance of a single image protocol was considered satisfactory for screening.

Cheung et al. [ 56 ] and Bellemo et al. [ 36 ] propose a model where retinal images will be firstly analysed by the DL systems. If the system does not find the need for further intervention, the patients will be rescanned accordingly to the screening program. However, if the system finds the need for intervention, two possible options are proposed: a semi-automated model where the images will be read by doctors or trained graders before referring the patients to an eye-centre or a fully-automated model where patients will be referred to an eye-centre without further investigations. Both models will lower the number of images doctors or trained graders should analyse, but could also increase the amount of patients that undergoes screening per day.

The progress made in imaging, treatment and understanding of DR has disrupted the existing DR screening guidelines, giving rise to diverse practical approaches to DR screening in various countries.

A recent review from 2021 [ 32 ] underscores the global variations in DR screening practices. It stated that developed countries focus on an effort to enhance the effectiveness and accuracy of DESPs, like optimized screening intervals and adoption of new imaging technologies. In contrast, for the vast majority of world countries, especially those with limited resources, the primary challenge lies in establishing basic DR screening infrastructures. The focus in these countries is therefore making basic DR screening accessible and efficient.

In numerous low-, lower- and middle-income nations, the primary challenge hindering the effectiveness of screening programs is the absence of DESP, insufficient resources, and limited access to skilled healthcare professionals [ 27 ]. Furthermore, the individual financial situations can influence the screening process, screening frequency, and access to treatment.

Screening is a vital step in disease management, and can potentially reduce the disease burden according to The World Health Organization [ 61 ]. Particularly in low-, lower- and middle-income nations this is more complex, because screening may improve disease detection but doesn’t always lead to adequate treatment. Thus, for screening programs to be truly impactful, they must be integrated with effective treatment strategies to ensure a meaningful decrease in the burden of diseases and to prevent serious consequences [ 61 ].

For countries that have fully implemented or are nearing the completion of a national DESP, opportunities for further enhancements still exist. Denmark has implemented a nationwide DR screening program in 2013 and the guidelines for this program have been grounded in comprehensive evidence-based practices since 2018 [ 1 ]. The screening in Denmark [ 2 , 42 ] is conducted by private practicing ophthalmologists and hospitals ensuring comprehensive coverage. A recent Danish study [ 62 ] revealed an impressive overall agreement of 93% between inter-grader reliability within the Danish screening program for DR. These results highlight the reliability and consistency of grading outcomes within the Danish screening program for DR.

Another potential area of improvement, as demonstrated by the Danish screening program, is the aspect where the screening results are documented in the Danish Registry of Diabetic Retinopathy (DiaBase), which is a national clinical quality database [ 63 ]. This feature enables the program to make informed decisions and adapt its screening strategies based on the data collected. For instance, a cohort study conducted in 2022 [ 64 ] examined participants in the Danish screening program from 2013 to 2018. The study determined the characteristics of patients who experienced delays in attending screenings and evaluated the impact of these delays on the progression of DR.

Shortcomings in these evidence-based guidelines does on the other hand exist. These guidelines focus solely on the vascular aspect of the disease and disregard evaluation of neural retina and diabetic retinal neurodegeneration. Scientific evidence has suggested that early neural degeneration may precede or coexist with vascular lesions and impact visual function [ 47 ]. The existing guidelines fails to account for the regression or resolution of retinal neovascularization as well as the influence of retinal areas with ischaemia or hypo perfusion.

There have been expression of concerns that the screening relies on limited standard field photographs (seven-field or less) [ 2 , 46 ]. The issue with fundus photos covering a “limited part” of the retinal surface is the risk of exclusion of peripheral retinal lesions, which could have prognostic implications that can enhance outcome prediction. New imaging technologies like UWF can capture approximately 82% of the retinal surface. Studies using UWF imaging also showed that around 50% of neovascularization cases are predominantly peripheral [ 47 , 48 ]. However, Kernt et al. [ 55 ] and Silva et al. [ 65 ] showed that UWF imaging rarely resulted in better detection of peripheral lesions in eyes with PDR that would not have been detected otherwise with ETDRS seven-standard fields. With the growing prevalence of diabetes [ 5 ] the demand for cost-effective DR screenings is substantial. Inadequate resources allocated to advanced imaging technologies, trained professionals or specialized facilities, might hinder accessibility and timely diagnosis, leading to increased long-term healthcare expenses due to untreated DR complications [ 1 , 3 , 4 , 56 , 57 ]. The implementation of AI in screening for DR can potentially enhance the cost-effectiveness, however this raises ethical concerns such as data privacy, algorithmic bias, and patient consent, which need to be addressed. Addressing issue like transparent AI algorithms, safeguarding patient data, and maintaining human oversight in decision-making are necessary steps to overcome these concerns [ 56 ]. The black box phenomenon is a returning concern because clinicians and patients always seek a reason for conclusions. This lack of understanding of how the algorithm comes to its conclusions, makes it impossible for physicians to detect potential biases. The issue of responsibility in the event of adverse patient outcomes due to AI-based technology is also a critical concern of many physicians [ 36 , 56 ].

DL algorithms demand substantial data to achieve acceptable performance levels. However, countries with lower income are often underrepresented in these datasets, which can lead to less accurate DL algorithms for these populations. Moreover “real-world" experiments are vital for validating DL algorithms and minimize bias [ 36 , 56 ]. Lower income countries may lack the resources required for developing, implementing, and retraining, which can decrease the benefits of AI based DR screening [ 57 ].

Many healthcare systems are not able to share data due to the protection of sensitive personal information, and the availability for diverse datasets from various settings and populations are limited [ 56 ]. In the realm of AI applied to DR screening, it is imperative for researchers to diligently pursue collaborative initiatives centred on data sharing. This concerted effort is essential to strengthen generalizability of AI models, thereby fortifying the safety and efficacy of forthcoming advancements in this research domain.

This review sheds light on the need for effective DR screening programs, especially with the projected rise in diabetes cases worldwide. While some countries have well-established national screening programs, many struggle with decentralized or private-based healthcare systems and the need of broader access to trained professionals, which all impact screening effectiveness.

Screening approaches, including practical classification scales, individualize screening intervals and new imaging modalities shows promising results, but also limitations. The emergence of AI technology raises hope for countries to either continues high standards DR screening without enormous economic cost, or enhancing screening accessibility to their inhabitants. However ethical concerns and the need for robust validation of AI persist.

Availability of data and materials

Not applicable. Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.

Abbreviations

• Diabetic retinopathy

Proliferative diabetic retinopathy

• Artificial intelligence

United Kingdom

Diabetes eye screening programs

International Agency for the Prevention of Blindness

South-East Asia countries

Early Treatment of Diabetic Retinopathy Study

International Clinical Diabetic Retinopathy Scale

International Council of Ophthalmology

Non-proliferative Diabetic Retinopathy

Grauslund J, Andersen N, Andresen J, Flesner P, Haamann P, Heegaard S, et al. Evidence-based Danish guidelines for screening of diabetic retinopathy. Acta Ophthalmol. 2018;96(8):763–9.

Article   CAS   PubMed   Google Scholar  

Wong TY, Sun J, Kawasaki R, Ruamviboonsuk P, Gupta N, Lansingh VC, et al. Guidelines on diabetic eye care: the international council of ophthalmology recommendations for screening, follow-up, referral, and treatment based on resource settings. Ophthalmology. 2018;125(10):1608–22.

Article   PubMed   Google Scholar  

Lanzetta P, Sarao V, Scanlon PH, Barratt J, Porta M, Bandello F, et al. Fundamental principles of an effective diabetic retinopathy screening program. Acta Diabetol. 2020;57(7):785–98.

Article   PubMed   PubMed Central   Google Scholar  

Vujosevic S, Aldington SJ, Silva P, Hernández C, Scanlon P, Peto T, et al. Screening for diabetic retinopathy: new perspectives and challenges. Lancet Diabetes Endocrinol. 2020;8(4):337–47.

Sun H, Saeedi P, Karuranga S, Pinkepank M, Ogurtsova K, Duncan BB, et al. IDF diabetes atlas: global, regional and country-level diabetes prevalence estimates for 2021 and projections for 2045. Diabetes Res Clin Pract. 2022;183:109119.

Teo ZL, Tham YC, Yu M, Chee ML, Rim TH, Cheung N, et al. Global prevalence of diabetic retinopathy and projection of burden through 2045: systematic review and meta-analysis. Ophthalmology. 2021;128(11):1580–91.

Hashemi H, Rezvan F, Pakzad R, Ansaripour A, Heydarian S, Yekta A, et al. Global and regional prevalence of diabetic retinopathy; a comprehensive systematic review and meta-analysis. Semin Ophthalmol. 2022;37(3):291–306.

Klein R, Klein BE, Moss SE, Davis MD, DeMets DL. The wisconsin epidemiologic study of diabetic retinopathy. IX. Four-year incidence and progression of diabetic retinopathy when age at diagnosis is less than 30 years. Arch Ophthalmol. 1989;107(2):237–43.

Burger W, Hövener G, Düsterhus R, Hartmann R, Weber B. Prevalence and development of retinopathy in children and adolescents with type 1 (insulin-dependent) diabetes mellitus. A longitudinal study. Diabetologia. 1986;29(1):17–22.

Henricsson M, Nyström L, Blohmé G, Ostman J, Kullberg C, Svensson M, et al. The incidence of retinopathy 10 years after diagnosis in young adult people with diabetes: results from the nationwide population-based diabetes incidence study in Sweden (DISS). Diabetes Care. 2003;26(2):349–54.

Grauslund J, Pedersen FN, Andersen N, Andresen J, Bek T, Dinesen S, et al. Presence and development of diabetic retinopathy in 153 238 patients with type 2 diabetes in the Danish registry of diabetic retinopathy. Acta Ophthalmol. 2023;101(2):207–14.

Jones CD, Greenwood RH, Misra A, Bachmann MO. Incidence and progression of diabetic retinopathy during 17 years of a population-based screening program in England. Diabetes Care. 2012;35(3):592–6.

Raman R, Ganesan S, Pal SS, Gella L, Kulothungan V, Sharma T. Incidence and progression of diabetic retinopathy in Urban India: Sankara Nethralaya-diabetic retinopathy epidemiology and molecular genetics study (SN-DREAMS II), report 1. Ophthalmic Epidemiol. 2017;24(5):294–302.

Song H, Liu L, Sum R, Fung M, Yap MK. Incidence of diabetic retinopathy in a Hong Kong Chinese population. Clin Exp Optom. 2011;94(6):563–7.

Cikamatana L, Mitchell P, Rochtchina E, Foran S, Wang JJ. Five-year incidence and progression of diabetic retinopathy in a defined older population: the blue mountains eye study. Eye. 2007;21(4):465–71.

Klein R, Klein BE, Moss SE, Davis MD, DeMets DL. The wisconsin epidemiologic study of diabetic retinopathy. X. Four-year incidence and progression of diabetic retinopathy when age at diagnosis is 30 years or more. Arch Ophthalmol. 1989;107(2):244–9.

Varma R, Choudhury F, Klein R, Chung J, Torres M, Azen SP. Four-year incidence and progression of diabetic retinopathy and macular edema: the Los Angeles Latino eye study. Am J Ophthalmol. 2010;149(5):752–61.

Leske MC, Wu SY, Hennis A, Nemesure B, Hyman L, Schachat A. Incidence of diabetic retinopathy in the Barbados eye studies. Ophthalmology. 2003;110(5):941–7.

Dinesen S, Stokholm L, Subhi Y, Peto T, Savarimuthu TR, Andersen N, et al. Five-year incidence of proliferative diabetic retinopathy and associated risk factors in a nationwide cohort of 201 945 Danish patients with diabetes. Ophthalmol Sci. 2023;3(3):100291.

Gange WS, Lopez J, Xu BY, Lung K, Seabury SA, Toy BC. Incidence of proliferative diabetic retinopathy and other neovascular sequelae at 5 years following diagnosis of type 2 diabetes. Diabetes Care. 2021;44(11):2518–26.

Article   CAS   PubMed   PubMed Central   Google Scholar  

Janghorbani M, Jones RB, Allison SP. Incidence of and risk factors for proliferative retinopathy and its association with blindness among diabetes clinic attenders. Ophthalmic Epidemiol. 2000;7(4):225–41.

McCarty DJ, Fu CL, Harper CA, Taylor HR, McCarty CA. Five-year incidence of diabetic retinopathy in the melbourne visual impairment project. Clin Exp Ophthalmol. 2003;31(5):397–402.

Leske MC, Wu SY, Hennis A, Nemesure B, Schachat AP, Hyman L, et al. Nine-year incidence of diabetic retinopathy in the Barbados eye studies. Arch Ophthalmol. 2006;124(2):250–5.

Grauslund J, Green A, Sjølie AK. Prevalence and 25 year incidence of proliferative retinopathy among Danish type 1 diabetic patients. Diabetologia. 2009;52(9):1829–35.

Takkar B, Das T, Thamarangsi T, Rani PK, Thapa R, Nayar PD, et al. Development of diabetic retinopathy screening guidelines in South-East Asia region using the context, challenges, and future technology. Semin Ophthalmol. 2022;37(1):97–104.

Muqit MMK, Kourgialis N, Jackson-deGraffenried M, Talukder Z, Khetran ER, Rahman A, et al. Trends in diabetic retinopathy, visual acuity, and treatment outcomes for patients living with diabetes in a fundus photograph-based diabetic retinopathy screening program in Bangladesh. JAMA Netw Open. 2019;2(11):e1916285.

Ramasamy K, Raman R, Tandon M. Current state of care for diabetic retinopathy in India. Curr Diab Rep. 2013;13(4):460–8.

Salamanca O, Geary A, Suárez N, Benavent S, Gonzalez M. Implementation of a diabetic retinopathy referral network. Peru Bull World Health Organ. 2018;96(10):674–81.

Acevedo Castellón RI, Carranza Vargas E, Cortés Chavarría RE, Rodríguez Vargas GA. Rapid assessment of avoidable blindness and diabetic retinopathy in individuals aged 50 years or older in Costa Rica. PLoS ONE. 2019;14(2):e0212660.

Polack S, Yorston D, López-Ramos A, Lepe-Orta S, Baia RM, Alves L, et al. Rapid assessment of avoidable blindness and diabetic retinopathy in Chiapas, Mexico. Ophthalmology. 2012;119(5):1033–40.

Barquera S, Campos-Nonato I, Aguilar-Salinas C, Lopez-Ridaura R, Arredondo A, Rivera-Dommarco J. Diabetes in Mexico: cost and management of diabetes and its complications and challenges for health policy. Global Health. 2013;9:3.

Das T, Takkar B, Sivaprasad S, Thanksphon T, Taylor H, Wiedemann P, et al. Recently updated global diabetic retinopathy screening guidelines: commonalities, differences, and future possibilities. Eye. 2021;35(10):2685–98.

Caporale JE, Elgart JF, Gagliardino JJ. Diabetes in argentina: cost and management of diabetes and its complications and challenges for health policy. Global Health. 2013;9:54.

Fernandes AG, Ferraz AN, Brant R, Malerbi FK. Diabetic retinopathy screening and treatment through the Brazilian National Health Insurance. Sci Rep. 2022;12(1):13941.

Article   CAS   PubMed   PubMed Central   ADS   Google Scholar  

Loncarek K, Petricek I, Brajac I, Filipović T, Vatavuk Z, Stalekar H. Quality of screening for diabetic retinopathy in the Rijeka region of Croatia. Coll Antropol. 2005;29(Suppl 1):107–10.

PubMed   Google Scholar  

Bellemo V, Lim G, Rim TH, Tan GSW, Cheung CY, Sadda S, et al. Artificial intelligence screening for diabetic retinopathy: the real-world emerging application. Curr Diab Rep. 2019;19(9):72.

Silva PS, Cavallerano JD, Tolls D, Omar A, Thakore K, Patel B, et al. Potential efficiency benefits of nonmydriatic ultrawide field retinal imaging in an ocular telehealth diabetic retinopathy program. Diabetes Care. 2014;37(1):50–5.

Fransen SR, Leonard-Martin TC, Feuer WJ, Hildebrand PL. Clinical evaluation of patients with diabetic retinopathy: accuracy of the Inoveon diabetic retinopathy-3DT system. Ophthalmology. 2002;109(3):595–601.

Cavallerano JD, Silva PS, Tolson AM, Francis T, Tolls D, Patel B, et al. Imager evaluation of diabetic retinopathy at the time of imaging in a telemedicine program. Diabetes Care. 2012;35(3):482–4.

Kirkizlar E, Serban N, Sisson JA, Swann JL, Barnes CS, Williams MD. Evaluation of telemedicine for screening of diabetic retinopathy in the Veterans Health Administration. Ophthalmology. 2013;120(12):2604–10.

Curran K, Piyasena P, Congdon N, Duke L, Malanda B, Peto T. Inclusion of diabetic retinopathy screening strategies in national-level diabetes care planning in low- and middle-income countries: a scoping review. Health Res Policy Syst. 2023;21(1):2.

Scanlon PH. The english national screening programme for diabetic retinopathy 2003–2016. Acta Diabetol. 2017;54(6):515–25.

Scanlon PH, Foy C, Chen FK. Visual acuity measurement and ocular co-morbidity in diabetic retinopathy screening. Br J Ophthalmol. 2008;92(6):775–8.

Early photocoagulation for diabetic retinopathy. ETDRS report number 9. Early treatment diabetic retinopathy study research group. Ophthalmology. 1991;98(5 Suppl):766–85.

Google Scholar  

Fundus photographic risk factors for progression of diabetic retinopathy. ETDRS report number 12. Early treatment diabetic retinopathy study research group. Ophthalmology. 1991;98(5 Suppl):823–33.

Wilkinson CP, Ferris FL 3rd, Klein RE, Lee PP, Agardh CD, Davis M, et al. Proposed international clinical diabetic retinopathy and diabetic macular edema disease severity scales. Ophthalmology. 2003;110(9):1677–82.

Yang Z, Tan TE, Shao Y, Wong TY, Li X. Classification of diabetic retinopathy: past, present and future. Front Endocrinol. 2022;13:1079217.

Article   Google Scholar  

Gangwani RA, Lian JX, McGhee SM, Wong D, Li KK. Diabetic retinopathy screening: global and local perspective. Hong Kong Med J. 2016;22(5):486–95.

CAS   PubMed   Google Scholar  

Rajalakshmi R, Arulmalar S, Usha M, Prathiba V, Kareemuddin KS, Anjana RM, et al. Validation of smartphone based retinal photography for diabetic retinopathy screening. PLoS ONE. 2015;10(9):e0138285.

Jacoba CMP, Salongcay RP, Aquino LAC, Salva CMG, Saunar AV, Alog GP, et al. Comparisons of handheld retinal imaging devices with ultrawide field images for determining diabetic retinopathy severity. Acta Ophthalmol. 2023;101(6):670–8.

de Oliveira JAE, Nakayama LF, Zago Ribeiro L, de Oliveira TVF, Choi S, Neto EM, et al. Clinical validation of a smartphone-based retinal camera for diabetic retinopathy screening. Acta Diabetol. 2023;60(8):1075–81.

Rajalakshmi R, Prathiba V, Arulmalar S, Usha M. Review of retinal cameras for global coverage of diabetic retinopathy screening. Eye. 2021;35(1):162–72.

Karakaya M, Hacisoftaoglu RE. Comparison of smartphone-based retinal imaging systems for diabetic retinopathy detection using deep learning. BMC Bioinformatics. 2020;21(Suppl 4):259.

Malerbi FK, Andrade RE, Morales PH, Stuchi JA, Lencione D, de Paulo JV, et al. Diabetic retinopathy screening using artificial intelligence and handheld smartphone-based retinal camera. J Diabetes Sci Technol. 2022;16(3):716–23.

Kernt M, Hadi I, Pinter F, Seidensticker F, Hirneiss C, Haritoglou C, et al. Assessment of diabetic retinopathy using nonmydriatic ultra-widefield scanning laser ophthalmoscopy (Optomap) compared with ETDRS 7-field stereo photography. Diabetes Care. 2012;35(12):2459–63.

Cheung CY, Tang F, Ting DSW, Tan GSW, Wong TY. Artificial intelligence in diabetic eye disease screening. Asia Pac J Ophthalmol. 2019. https://doi.org/10.22608/APO.201976 .

Cleland CR, Rwiza J, Evans JR, Gordon I, MacLeod D, Burton MJ, et al. Artificial intelligence for diabetic retinopathy in low-income and middle-income countries: a scoping review. BMJ Open Diabetes Res Care. 2023. https://doi.org/10.1136/bmjdrc-2023-003424 .

Nielsen KB, Lautrup ML, Andersen JKH, Savarimuthu TR, Grauslund J. Deep learning-based algorithms in screening of diabetic retinopathy: a systematic review of diagnostic performance. Ophthalmol Retina. 2019;3(4):294–304.

Rajalakshmi R, Subashini R, Anjana RM, Mohan V. Automated diabetic retinopathy detection in smartphone-based fundus photography using artificial intelligence. Eye. 2018;32(6):1138–44.

Penha FM, Priotto BM, Hennig F, Przysiezny B, Wiethorn BA, Orsi J, et al. Single retinal image for diabetic retinopathy screening: performance of a handheld device with embedded artificial intelligence. Int J Retina Vitreous. 2023;9(1):41.

Dans LF, Silvestre MA, Dans AL. Trade-off between benefit and harm is crucial in health screening recommendations. Part I: general principles. J Clin Epidemiol. 2011;64(3):231–9.

Thykjaer AS, Andresen J, Andersen N, Bek T, Heegaard S, Hajari J, et al. Inter-grader reliability in the Danish screening programme for diabetic retinopathy. Acta Ophthalmol. 2023;101(7):783–8.

Andersen N, Hjortdal J, Schielke KC, Bek T, Grauslund J, Laugesen CS, et al. The Danish registry of diabetic retinopathy. Clin Epidemiol. 2016;8:613–9.

Thykjær AS, Andersen N, Bek T, Heegaard S, Hajari J, Laugesen CS, et al. Attendance in a national screening program for diabetic retinopathy: a population-based study of 205,970 patients. Acta Diabetol. 2022;59(11):1493–503.

Silva PS, Cavallerano JD, Sun JK, Soliman AZ, Aiello LM, Aiello LP. Peripheral lesions identified by mydriatic ultrawide field imaging: distribution and potential impact on diabetic retinopathy severity. Ophthalmology. 2013;120(12):2587–95.

Murthy GVS, Gilbert C, Shukla R, Bala V, Anirudh GG, Mukpalkar S, et al. Overview and project highlights of an initiative to integrate diabetic retinopathy screening and management in the public health system in India. Indian J Ophthalmol. 2020;68(Suppl 1):S12–5.

Gaiha SM, Shukla R, Gilbert CE, Anchala R, Gudlavalleti MV. Is India’s policy framework geared for effective action on avoidable blindness from diabetes? Indian J Endocrinol Metab. 2016;20(Suppl 1):S42-50.

PubMed   PubMed Central   Google Scholar  

Mwangi N, Ng’ang’a M, Gakuo E, Gichuhi S, Macleod D, Moorman C, et al. Effectiveness of peer support to increase uptake of retinal examination for diabetic retinopathy: study protocol for the DURE pragmatic cluster randomized clinical trial in Kirinyaga, Kenya. BMC Public Health. 2018;18(1):871.

Bhatta S, Pant N, Pant SR. Prevalence of diabetes and diabetic retinopathy in far-western province of Nepal. J Nepal Health Res Counc. 2023;20(4):875–80.

Bechange S, Roca A, Schmidt E, Gillani M, Ahmed L, Iqbal R, et al. Diabetic retinopathy service delivery and integration into the health system in Pakistan-findings from a multicentre qualitative study. PLoS ONE. 2021;16(12):e0260936.

Piyasena M, Murthy GVS, Yip JLY, Gilbert C, Peto T, Premarathna M, et al. A qualitative study on barriers and enablers to uptake of diabetic retinopathy screening by people with diabetes in the Western Province of Sri Lanka. Trop Med Health. 2019;47:34.

Hall CE, Hall AB, Mallya J, Courtright P, Kok G. Establishing a screening programme for diabetic retinopathy in Kilimanjaro Region, Tanzania using intervention mapping. Eye. 2022;36(Suppl 1):17–24.

Curran K, Lohfeld L, Congdon N, Peto T, Hoang TT, Nguyen HT, et al. Ophthalmologists’ and patients’ perspectives on treatments for diabetic retinopathy and maculopathy in Vietnam: a descriptive qualitative study. BMJ Open. 2022;12(7):e055061.

Yahya T, Nazzal Z, Abdul-Hadi AR, Belkebir S, Hamarshih M, Fuqaha A, et al. Diabetic retinopathy screening barriers among Palestinian primary health care patients: a qualitative study. J Diabetes Metab Disord. 2020;19(2):875–81.

Lewis AD, Hogg RE, Chandran M, Musonda L, North L, Chakravarthy U, et al. Prevalence of diabetic retinopathy and visual impairment in patients with diabetes mellitus in Zambia through the implementation of a mobile diabetic retinopathy screening project in the Copperbelt province: a cross-sectional study. Eye. 2018;32(7):1201–8.

Pidro A, Ahmedbegovic-Pjano M, Grisevic S, Sofic-Drino V, Gabric K, Biscevic A. Epidemiology of diabetic retinopathy at eye clinic Svjetlost Sarajevo: two years retrospective single center study. Mater Sociomed. 2019;31(4):290–3.

Blake A, Munby HN, Katlego PM, Sebuyuyu PM, Nkomozana O, Bangure R, Kerr-Muir M, Ngondi J. Characteristics of patients with diabetic retinopathy in Gaborone. Botswana Tanzania Health Res Bullet. 2015;17(1):1–7.

You QS, Xu L, Wang YX, Liang QF, Cui TT, Yang XH, et al. Prevalence of diabetic retinopathy as cause for visual impairment: the Beijing public health care project. Clin Exp Ophthalmol. 2013;41(6):608–9.

Ram S, Mohammadnezhad M, Ram K, Prasad K, Pal M, Dalmia P. Increasing and sustaining diabetic retinopathy screening in Fiji by leveraging community health workers (CHWs) services: a qualitative study. Heliyon. 2022;8(11):e11379.

Chelala E, Saleh N, Dirani A, Fadlallah A, Baz P, Slim E, et al. Screening of diabetic retinopathy and maculopathy in Lebanese population using retinography and SD-OCT: The role of telemedicine. J Med Liban. 2015;63(2):59–65.

Ngah NF, Muhamad NA, Asnir ZZ, Abdul Aziz RA, Mhad Kassim Z, Sahar SA, et al. Descriptive assessment on diabetic retinopathy screening in an awareness programme in Malaysia. Int J Ophthalmol. 2020;13(11):1808–13.

Abdool Z, Naidoo K, Visser L. Development of a diabetic retinopathy screening model for a district health system in Limpopo Province, South Africa. Afr Vis Eye Health. 2022. https://doi.org/10.4102/aveh.v81i1.568 .

Ruamviboonsuk P, Tiwari R, Sayres R, Nganthavee V, Hemarat K, Kongprayoon A, et al. Real-time diabetic retinopathy screening by deep learning in a multisite national screening programme: a prospective interventional cohort study. Lancet Digit Health. 2022;4(4):e235–44.

Murray RB, Metcalf SM, Lewis PM, Mein JK, McAllister IL. Sustaining remote-area programs: retinal camera use by Aboriginal health workers and nurses in a Kimberley partnership. Med J Aust. 2005;182(10):520–3.

Al Alawi E, Ahmed AA. Screening for diabetic retinopathy: the first telemedicine approach in a primary care setting in Bahrain. Middle East Afr J Ophthalmol. 2012;19(3):295–8.

Maberley D, Walker H, Koushik A, Cruess A. Screening for diabetic retinopathy in James Bay, Ontario: a cost-effectiveness analysis. CMAJ. 2003;168(2):160–4.

Arenas-Cavalli JT, Abarca I, Rojas-Contreras M, Bernuy F, Donoso R. Clinical validation of an artificial intelligence-based diabetic retinopathy screening tool for a national health system. Eye. 2022;36(1):78–85.

Krieger B, Hallik R, Kala K, Ülper K, Polonski M. Validation of mobile-based funduscope for diabetic retinopathy screening in Estonia. Eur J Ophthalmol. 2022;32(1):508–13.

Laitinen A, Laatikainen L, Härkänen T, Koskinen S, Reunanen A, Aromaa A. Prevalence of major eye diseases and causes of visual impairment in the adult Finnish population: a nationwide population-based survey. Acta Ophthalmol. 2010;88(4):463–71.

Massin P, Chabouis A, Erginay A, Viens-Bitker C, Lecleire-Collet A, Meas T, et al. OPHDIAT: a telemedical network screening system for diabetic retinopathy in the Ile-de-France. Diabetes Metab. 2008;34(3):227–34.

Kreft D, McGuinness MB, Doblhammer G, Finger RP. Diabetic retinopathy screening in incident diabetes mellitus type 2 in Germany between 2004 and 2013—a prospective cohort study based on health claims data. PLoS ONE. 2018;13(4):e0195426.

Kristinsson JK, Hauksdóttir H, Stefánsson E, Jónasson F, Gíslason I. Active prevention in diabetic eye disease. A 4-year follow-up. Acta Ophthalmol Scand. 1997;75(3):249–54.

Diabetic Retina Screen: The National Diabetic Retinal Screening Programme. https://www.diabeticretinascreen.ie/home/default.aspx?id=2&amprcss=print .

Scarpa G, Urban F, Vujosevic S, Tessarin M, Gallo G, Visentin A, et al. The nonmydriatic fundus camera in diabetic retinopathy screening: a cost-effective study with evaluation for future large-scale application. J Ophthalmol. 2016;2016:4625096.

Diabetes: A National Public Health Priority. A National Strategy for Diabetes. 2016.

Zealand HN. Diabetic Retinal Screening, Grading, Monitoring and Referral Guidance. 2016.

Carlsen S, Skrivarhaug T, Thue G, Cooper JG, Gøransson L, Løvaas K, et al. Glycemic control and complications in patients with type 1 diabetes—a registry-based longitudinal study of adolescents and young adults. Pediatr Diabetes. 2017;18(3):188–95.

Jamison C, Peto T, Quinn N, D’Aloisio R, Cushley LN, Johnston PC. Screening attendance, prevalence and severity of diabetic retinopathy (DR) in a cohort of patients with diabetes mellitus secondary to chronic pancreatitis (DMsCP) in Northern Ireland. BMJ Open Diabetes Res Care. 2021. https://doi.org/10.1136/bmjdrc-2021-002267 .

Pereira AMP, da Silva Laureano RM, de Lima Neto FB. Five regions, five retinopathy screening programmes: a systematic review of how Portugal addresses the challenge. BMC Health Serv Res. 2021;21(1):756.

Looker HC, Nyangoma SO, Cromie DT, Olson JA, Leese GP, Black MW, et al. Rates of referable eye disease in the Scottish national diabetic retinopathy screening programme. Br J Ophthalmol. 2014;98(6):790–5.

Tan TE, Wong TY. Diabetic retinopathy: looking forward to 2030. Front Endocrinol. 2022;13:1077669.

Gračner T. Screening for diabetic retinopathy—a twelve-month review. Acta Clin Croat. 2020;59(3):424–30.

Chung YR, Ha KH, Lee K, Kim DJ. Diabetic retinopathy and related clinical practice for people with diabetes in Korea: a 10-year trend analysis. Diabetes Metab J. 2020;44(6):928–32.

Romero-Aroca P, López-Galvez M, Martinez-Brocca MA, Pareja-Ríos A, Artola S, Franch-Nadal J, et al. Changes in the epidemiology of diabetic retinopathy in Spain: a systematic review and meta-analysis. Healthcare. 2022. https://doi.org/10.3390/healthcare10071318 .

Agardh E, Agardh CD, Hansson-Lundblad C. The five-year incidence of blindness after introducing a screening programme for early detection of treatable diabetic retinopathy. Diabet Med. 1993;10(6):555–9.

Quality and Efficiency of Diabetes Care in Sweden: National Performance Assessment 2011. https://www.socialstyrelsen.se/globalassets/sharepoint-dokument/artikelkatalog/statistik/2014-3-18.pdf .

van Eijk KN, Blom JW, Gussekloo J, Polak BC, Groeneveld Y. Diabetic retinopathy screening in patients with diabetes mellitus in primary care: Incentives and barriers to screening attendance. Diabetes Res Clin Pract. 2012;96(1):10–6.

Gillow JT, Gray JA. The national screening committee review of diabetic retinopathy screening. Eye. 2001;15(Pt 1):1–2.

Download references

Acknowledgements

Publication fees were covered by an educational grant from F. Hoffmann-La Roche Ltd. The company had no influence in the research, in the writing or in the manuscript subject selection, which were at the discretion of the editors.

No other funding to apply.

Author information

Authors and affiliations.

Department of Ophthalmology, Odense University Hospital, Sdr. Boulevard 29, 5000, Odense, Denmark

Andreas Abou Taha, Sebastian Dinesen, Anna Stage Vergmann & Jakob Grauslund

Department of Clinical Research, University of Southern Denmark, Odense, Denmark

Sebastian Dinesen, Anna Stage Vergmann & Jakob Grauslund

Steno Diabetes Center Odense, Odense University Hospital, Odense, Denmark

Sebastian Dinesen & Jakob Grauslund

You can also search for this author in PubMed   Google Scholar

Contributions

ASV and JG contributed significantly to the design of the study. Additionally, ASV and JG was involved as a reviewer for the study conducted by AAT and SD, providing valuable insights and critical analysis. With AAT as the first author and SD as the second author, both AAT and SD were responsible for constructing the manuscript, consolidating the research findings, and organizing the content into a coherent form for publication. AAT’s and SD’s contributions were instrumental in shaping the final document. All authors have approved the submitted work.

Corresponding author

Correspondence to Andreas Abou Taha .

Ethics declarations

Ethics approval and consent to participate.

Not applicable.

Consent for publication

Competing interests.

The authors declare that they have no competing interests.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Abou Taha, A., Dinesen, S., Vergmann, A.S. et al. Present and future screening programs for diabetic retinopathy: a narrative review. Int J Retin Vitr 10 , 14 (2024). https://doi.org/10.1186/s40942-024-00534-8

Download citation

Received : 22 December 2023

Accepted : 19 January 2024

Published : 03 February 2024

DOI : https://doi.org/10.1186/s40942-024-00534-8

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Screening program
• Epidemiology

International Journal of Retina and Vitreous

ISSN: 2056-9920

• Submission enquiries: Access here and click Contact Us
• General enquiries: [email protected]

Log in using your username and password

• Search More Search for this keyword Advanced search
• Latest content
• BMJ Journals

You are here

• Volume 12, Issue 3
• Prevalence and risk factors for diabetic retinopathy at diagnosis of type 2 diabetes: an observational study of 77 681 patients from the Swedish National Diabetes Registry
• Article Text
• Article info
• Citation Tools
• Rapid Responses
• Article metrics

• http://orcid.org/0000-0003-3701-5065 Sheyda Sofizadeh 1 , 2 ,
• http://orcid.org/0000-0002-3376-4707 Katarina Eeg-Olofsson 2 , 3 , 4 ,
• Marcus Lind 1 , 2 , 4
• 1 Department of Medicine , NU-Hospital Group , Uddevalla , Sweden
• 2 Department of Molecular and Clinical Medicine , University of Gothenburg , Gothenburg , Sweden
• 3 National Diabetes Register , Centre of Registers , Gothenburg , Sweden
• 4 Department of Medicine , Sahlgrenska University Hospital , Gothenburg , Sweden
• Correspondence to Dr Marcus Lind; marcus.lind{at}gu.se

Introduction To assess the prevalence of diabetic retinopathy (DR) in persons with newly diagnosed type 2 diabetes (T2D) to understand the potential need for intensified screening for early detection of T2D.

Research design and methods Individuals from the Swedish National Diabetes Registry with a retinal photo <2 years after diagnosis of T2D were included. The proportion of patients with retinopathy (simplex or worse) was assessed. Patient characteristics and risk factors at diagnosis were analyzed in relation to DR with logistic regression.

Results In total, 77 681 individuals with newly diagnosed T2D, mean age 62.6 years, 41.1% females were included. Of these, 13 329 (17.2%) had DR.

DR was more common in older persons (adjusted OR 1.03 per 10-year increase, 95% CI 1.01 to 1.05) and men compared with women, OR 1.10 (1.05 to 1.14). Other variables associated with DR were OR (95% CI): lower education 1.08 (1.02 to 1.14); previous stroke 1.18 (1.07 to 1.30); chronic kidney disease 1.29 (1.07 to 1.56); treatment with acetylsalicylic acid 1.14 (1.07 to 1.21); ACE inhibitors 1.12 (1.05 to 1.19); and alpha blockers 1.41 (1.15 to 1.73). DR was more common in individuals born in Asia (OR 1.16, 95% CI 1.08 to 1.25) and European countries other than those born in Sweden (OR 1.11, 95% CI 1.05 to 1.18).

Conclusions Intensified focus on screening of T2D may be needed in Sweden in clinical practice since nearly one-fifth of persons have retinopathy at diagnosis of T2D. The prevalence of DR was higher in men, birthplace outside of Sweden, and those with a history of stroke, kidney disease, and hypertension.

• Diabetes Mellitus, Type 2
• Diabetic Retinopathy
• Prediabetic State
• Glycated Hemoglobin A

Data availability statement

Data are available upon reasonable request. Data can be accessed after a written research proposal and support from investigators and upon request and after legal procedures have taken place for making data transfer possible.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:  http://creativecommons.org/licenses/by-nc/4.0/ .

https://doi.org/10.1136/bmjdrc-2023-003976

Statistics from Altmetric.com

Request permissions.

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

WHAT IS ALREADY KNOWN ON THIS TOPIC

Diabetic retinopathy at diagnosis of type 2 diabetes (T2D) is used as a surrogate marker to indicate late detected T2D, but contemporary and population-based studies are sparse.

WHAT THIS STUDY ADDS

The study reveals that a significant proportion (17.2%) of individuals newly diagnosed with T2D in Sweden already have DR at diagnosis, indicating that a significant proportion of patients have had long-term hyperglycemia before diagnosis.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE, OR POLICY

Attention is needed in clinical practice in Sweden regarding screening for T2D in persons with a risk profile and further research is urgently needed regarding potential benefits of structured screening in the population.

Introduction

Diabetic retinopathy (DR) is the most common microvascular complication of diabetes. 1 High blood glucose levels are a critical risk factor for DR, and the risk and severity of DR are directly related to glycated hemoglobin A1c (HbA1c) level over time in both type 1 diabetes and type 2 diabetes (T2D). 2–5 Since DR typically develops over several years, individuals with DR at diagnosis of T2D generally have elevated blood glucose levels long before diagnosis. 6 Hypertension in conjunction with hyperglycemia is also a well-established risk factor for DR progression. 7 Other risk factors that have been associated with retinopathy in persons with T2D are Body Mass Index (BMI), dyslipidemia, insulin treatment, and nephropathy. 8–11

The Swedish National Diabetes Registry (NDR) includes the majority of persons with T2D within the country. 12 Diabetes care in Sweden has significantly improved over time and more patients are reaching glucose control targets. Given intensive treatment in patients with newly diagnosed T2D, undetected hyperglycemia before diagnosis of T2D may be at least as harmful or more so than after diagnosis of T2D. When T2D is undetected individuals may unknowingly have glycemic levels clearly above targets, while after diagnosis modern diabetes care enables patients in many instances to achieve HbA1c targets associated with low risk of diabetes complications. 12 Early hyperglycemia can also be detrimental over time by virtue of legacy effects, and before diagnosis patients do not receive the same level of attention in terms of screening and treatment for complications. 13 14

Prevalence of retinopathy at diagnosis of T2D has been used as a surrogate marker for late detected T2D in several other studies. 15 The aim of the current study was to evaluate to what extent DR exists in persons with newly diagnosed T2D in Sweden and to investigate factors related to increased risk of DR among patients included in the NDR, which includes the absolute majority of persons with T2D in Sweden.

Research design and methods

The study was approved by the Swedish Ethical Review Authority (Dnr 977-17).

Data sources

We conducted a registry-based study using data from the NDR. After patients provide verbal informed consent, data are reported directly to the NDR from clinical visits to primary care clinics and hospital diabetes clinics 12 16 and include risk factors, medications, and complications for individuals with diabetes. Data for the current population of persons with T2D were linked with data from the Swedish Cause of Death Registry, the National Inpatient and Outpatient Registries, the Prescribed Drug Registry, and the Longitudinal Integration Database for Health Insurance and Labour Market Studies. 16–18

Study population

Individuals diagnosed with T2D from January 1, 2015 to December 31, 2019 with data about DR less than 2 years after diagnosis of T2D were included. Retinal screening is recommended to be performed soon after diagnosis of T2D. Retinal screening is performed by an ophthalmologist or a nurse specialized in ophthalmology. If more severe stages of retinopathy exist, an ophthalmologist is consulted. Information on retinopathy is recorded in the NDR by nurses and physicians working in primary care and outpatient diabetes clinics at hospital. Retinopathy is recorded as non, simplex, non-proliferative, or proliferative retinopathy. However, the variable with best coverage only includes information on whether any retinopathy exists. This variable was used in the current study since the prevalence of DR at diagnosis of T2D was estimated. The procedure for retinal screening has been described in greater detail in earlier studies. 19

T2D diagnosis required a clinical diagnosis of T2D and fulfilling the following epidemiologic definition: treatment with either diet or non-insulin antihyperglycemic agents only or diagnosis at 40 years of age or older receiving insulin therapy or insulin and oral antihyperglycemic agents. 16 20 Persons with a diagnosis of type 1 diabetes or less than 18 years of age at index date were excluded.

There were 138 888 adults in the NDR with newly diagnosed T2D. Of these, 61 207 (44%) did not have data about DR less than 2 years after diagnosis ( figure 1 ). A total of 77 681 persons with T2D remained and were included in the cohort.

• Download figure
• Open in new tab
• Download powerpoint

Flow chart of participants in the current study recruited from the National Diabetes Registry (NDR). BMI, Body Mass Index; DR, diabetic retinopathy; HbA1c, glycated hemoglobin A1c; T2D, type 2 diabetes.

Study procedures

The number and proportion of patients with DR at diagnosis of T2D were calculated. The following variables were evaluated if they were related to the presence of DR: age, sex, smoking, HbA1c level, BMI, blood pressure, level of education, geographic area of birth, diabetes-related medications, renal complications, and cardiovascular comorbidities.

To be representative of the time point of T2D diagnosis (first entry date in the NDR at year of diagnosis), HbA1c and blood pressure measurements had to exist less than 4 weeks after diagnosis of diabetes to be included in the analyses. BMI and smoking data had to exist within 6 weeks and 6 months after diagnosis, respectively.

HbA1c was reported according to the International Federation of Clinical Chemistry standard, measured in mmol/mol, and converted to percent units according to the National Glycohemoglobin Standardization Program for dual reporting criteria. 21 Laboratory methods at participating care units for analyzing HbA1c were regularly checked with central reference samples of HbA1c to ensure high accuracy. 22 HbA1c categories included commonly used targets of HbA1c as well as cut-offs used for very poor glucose control in the NDR of 70 mmol/mol (8.6%). 12 23 HbA1c was categorized as <48 mmol/mol (6.5%), 48–52 mmol/mol (6.5%–6.9%), 53–57 mmol/mol (7.0%–7.4%), 58–70 mmol/mol (7.5%–8.6%), and >70 mmol/mol (>8.6%). Blood pressure was defined as the mean value of two supine readings with a cuff of appropriate size and after at least 5 min of rest. Systolic blood pressure (SBP) was categorized as <110 mm Hg with increments of 10 mm Hg with the highest category ≥140 mm Hg, diastolic blood pressure (DBP) as <60, 60–<70, 70–<80, 80–<85, and ≥ 85 mm Hg. Commonly used levels of BMI for classifying underweight normal weight, obesity, and severe obesity were used when evaluating BMI 23 24 : <18.5, 18.5–<25, 25–<30, 30–<35, and ≥35 kg/m 2 . Smoking was categorized as “No” (never smokers and previous smokers) versus “Yes” (current smokers), education level as up to 9 years, 10–12 years, or college/university; and geographic area of birth as Africa, Asia, Europe (excluding Sweden), Oceania, North America, South America, and Sweden.

International classification of diseases (ICD-10) codes were used to define study comorbidities (ICD codes are described in online supplemental material ). Comorbidities were investigated back until year 1997 when ICD-10 was introduced. If a diagnosis of a certain comorbidity existed during the time period from 1997 until diagnosis of T2D, it was regarded as prevalent. The following comorbidities were evaluated: coronary heart disease (CHD), stroke, atrial fibrillation, heart failure, coronary artery bypass graft, peripheral arterial disease and chronic kidney disease (CKD).

Supplemental material

Anatomical therapeutic classifications based on the prescribed drug registry were used for evaluation of drugs. The following classes of drugs were evaluated: acetylsalicylic acid, antihypertensive, beta-blockers, ACE inhibitors, angiotensin II receptor blockers, calcium channel blockers, alpha-blockers, and diuretics. Presence of hyperlipidemia and hypertension before diagnosis of T2D was both defined by use of prescribed drugs before diagnosis of T2D.

Statistical analyses

Descriptive statistics are presented for patients with T2D who had a registration regarding retinopathy in NDR <2 years after diagnosis of T2D (Main cohort) and patients with T2D without registration about retinopathy examinations in the same time window (Excluded group). The groups were compared with comparative tests (t-test and standard mean difference) to describe any differences. 25–27

In patients with newly diagnosed T2D with a registration of retinopathy examination (Main cohort), characteristics of the proportion of patients with DR were compared with those without DR and descriptive statistics are presented and the groups were compared with t-test and χ 2 -test. Multiple logistic regression was used to evaluate variables associated with DR at diagnosis of T2D. Results are presented as adjusted ORs with 95% CIs.

Variables associated with DR were analyzed with logistic regression in three different subcohorts to include as many individuals as possible for a certain variable. Missing variables were handled in the regression analysis using complete cases. Each analysis included individuals with data on all variables. Subcohort 1 (n=73 350) included all patients with data on age, sex, educational level, comorbidities, diabetes-related drugs, and HbA1c. Subcohort 2 (n=56 764) included patients who had blood pressure and BMI data in addition to subcohort 1. Subcohort 3 (n=52 697) included patients who had smoking data in addition to subcohort 2. Two-tailed tests were used and a significance level of 0.05 was applied. The results are reported as OR with their 95% CI. SAS V.9.4 was used for statistical analyses.

Role of the funding source

The funders had no role in study design, data collection and analysis, preparation of the manuscript, or decision to submit for publication.

Prevalence of DR

A total of 77 681 individuals with newly diagnosed T2D were included in the current study (main cohort). Overall, patient characteristics were numerically similar to the 61 207 patients without information on retinopathy less than 2 years after diagnosis of T2D ( table 1 ). Mean age in the main cohort compared with excluded patients was 62.6 and 62.8 years and 41.1% and 41.8% were females, respectively. Geographic area of birth was also similar with 76.2% and 74.2% born in Sweden, 11.6% and 11.7% in other European countries, and 8.5% vs 9.7% in Asia, respectively. Education level 10–12 years was 48.1% vs 46.1% and college/university was 22.9% vs 22.5%, respectively.

• View inline

Patient characteristics at diagnosis of type 2 diabetes (T2D) shown both for the main cohort and excluded persons without information of retinopathy screening within 2 years of their diagnosis

The frequency of comorbidities was numerically similar with a prevalence of CHD of 12.0% vs 13.5% and stroke 3.8% vs 4.5%, respectively. Mean HbA1c level was 58.1 mmol/mol (7.5%) and 54.3 mmol/mol (7.1%). Although significant differences between the groups existed for several variables, the numeric differences were small illustrated by low standard mean differences ( table 1 ). Percentage of patients in the main cohort and excluded patients having missing data around the time of T2D diagnosis of HbA1c, smoking, BMI, and blood pressure were overall similar in the two groups, but a slightly larger proportion of included patients had missing data on blood pressure ( online supplemental table S1 ).

In total, 13 329 (17.2%) had DR at diagnosis of T2D. Patient characteristics for persons with and without DR in the main cohort by diagnosis of DR are presented in table 2 .

Patient characteristics main cohort participants with and without diabetic retinopathy (DR) at diagnosis of type 2 diabetes (T2D)

Risk factors for DR

We evaluated adjusted ORs in subcohorts of individuals with data on the covariates ( figure 1 ). Characteristics were overall similar in subcohorts 1–3 ( online supplemental table S2 ).

In subcohort 1, 73 350 (94%) individuals with data on age, sex, comorbidities, educational level, geographic area of birth, prescribed medications, and glycemic control (HbA1c within 4 weeks after diagnosis) were analyzed with logistic regression ( figure 2 ).

ORs of retinopathy by patient characteristics in persons with newly diagnosed type 2 diabetes from multivariable logistic regression models in subcohort 1 (n=73 350) with information on age, sex, comorbidities, educational level, geographic area of birth, prescribed medications, and glycemic control (HbA1c within 4 weeks after diagnosis). Reference groups for each variable are indicated by the y-axis labels. Points and error bars represent ORs and 95% CIs. HbA1c, glycated hemoglobin A1c.

DR was more common in older persons, by OR 1.03 (95% CI 1.01 to 1.05, p=0.004) per 10 years increase and more common in men compared with women OR 1.10 (95% CI 1.05 to 1.14, p<0.001). Other variables associated with DR were lower education, OR 1.08 for primary versus college/university (95% CI 1.02 to 1.14, p=0.009), previous stroke, OR 1.18 (95% CI 1.07 to 1.30, p=0.001), CKD, OR 1.29 (95% CI 1.07 to 1.56, p=0.008), treatment with acetylsalicylic acid, OR 1.14 (95% CI 1.07 to 1.21, p<0.001), ACE inhibitors, OR 1.12 (95% CI 1.05 to 1.19, p<0.001), and alpha blockers, OR 1.41 (95% CI 1.15 to 1.73, p<0.001). With respect to geographic area, DR was more common in individuals born in Asia, OR 1.16 (95% CI 1.08 to 1.25, p<0.001) and European countries other than Sweden, OR 1.11 (95% CI 1.05 to 1.18, p<0.001) compared with those born in Sweden ( figure 2 ).

In subcohort 2, 56 764 patients (73% of the main cohort) additionally had data on SBP and DBP less than 4 weeks after diagnosis of T2D and BMI less than 6 weeks after diagnosis. The risk of DR increased with higher SBP with an OR of 1.33 (95% CI 1.20 to 1.46, p<0.001) for an SBP ≥140 mm Hg compared with those having an SBP of 110–119 mm Hg. In contrast, the risk of DR decreased with higher BMI with an OR of 0.75 (95% CI 0.69 to 0.81, p<0.001) and 0.72 (95% CI 0.66 to 0.78, p<0.001) for those with BMI 30–34.9 and ≥35 kg/m 2 compared with 18.5–24.9 kg/mg 2 , respectively ( figure 3 ).

ORs of retinopathy by patient characteristics in persons with newly diagnosed type 2 diabetes (T2D) from multivariable logistic regression models in subcohort 2 (n=56 764) with information on age, sex, comorbidities, educational level, geographic area of birth, prescribed medications, and glycemic control (HbA1c within 4 weeks after diagnosis) and also including systolic blood pressure and diastolic blood pressure less than 4 weeks after diagnosis of T2D and BMI less than 6 weeks after diagnosis. Reference groups for each variable are indicated by the y-axis labels. Points and error bars represent ORs and 95% CIs. BMI, Body Mass Index; HbA1c, glycated hemoglobin A1c.

In subcohort 3, 52 697 patients (68% in the main cohort) additionally had data on smoking at less than 6 months after diagnosis of T2D. Smoking showed no association with DR with an OR of 1.05 (95% CI 0.95 to 1.12, p=0.17). ORs for other variables were similar to those from subcohort 1 and 2 (data not shown).

Conclusions

In this nationwide study from Sweden, using DR at diagnosis of T2D as a marker for late detected T2D, almost one-fifth of patients had DR at diagnosis of T2D. DR was more common in men, individuals born in Asia, and those with a history of stroke and kidney disease. High SBP and elevated HbA1c levels were also associated with DR. A higher proportion of patients with normal weight had DR at diagnosis of T2D compared with those who were overweight or obese. DR was less common in individuals with previous CHD.

Prevalence of retinopathy as an indicator for late detected T2D has been used in several earlier studies. 15 However, contemporary population-based studies of the prevalence of DR are overall spars. In a UK-based study examining newly diagnosed persons with T2D until year 2017, the prevalence of DR ranged from 14% to 25% depending on whether pre-diabetes had been recorded as diagnosis or not before diagnosis of T2D. 28 A systematic review and meta-analysis including studies generally performed more than 10 years ago found that the pooled prevalence of DR at diagnosis of T2D was 14.6% (95% CI 11.9% to 17.3%). 15 Some studies have reported that DR is present in up to 15%–20% of patients at the time of diagnosis of T2D, while others have reported that DR is present in around 5%–10%. 6 15 29–33

Hyperglycemia and hypertension are risk factors for DR in persons with established T2D as confirmed in randomized settings. 2 7 Studies have also reported hyperglycemia and hypertension to be more common in patients with DR at diagnosis of T2D. 1 15 34 DR at diagnosis of T2D has also been reported to be more common in persons with renal complications whereas smoking has shown divergent associations. 5 32 In different populations of individuals with DR has been more common in men compared with women. 35

Experience from clinical practice and studies in type 1 diabetes, where the initial hyperglycemia is generally more abrupt, suggest that hyperglycemia generally needs to exist over a long period of time before DR appears. 3 4 Data indicate that diabetes is generally present for at least 5 years before signs of retinopathy appear, and it may be more than 10 years after diagnosis of diabetes before clinical diagnosis of DR. 6 That almost one-fifth of patients in the current study had DR at diagnosis of T2D indicates that long-standing hyperglycemia before diagnosis of T2D is relatively common in Sweden, and hyperglycemia increases risk of complications at diagnosis of T2D. Furthermore, legacy effects of earlier hyperglycemia may worsen prognosis after diagnosis compared with persons with early detection. 13 14 Moreover, many individuals do not receive treatments for preventing diabetes complications before diagnosis of T2D such as lipid-lowering and antihypertensive drugs, lifestyle advice, and screening programs for complications. 23 It is possible that diabetes complications and mortality can be reduced during this high-risk phase if diabetes is detected early, and intensive prevention programs are started. ACE inhibitors and angiotensin-2 receptor blockers are likely beneficial in preventing or slowing the progression of early DR. 36 Further, studies indicate that the use of antiplatelet/anticoagulant medications may reduce the risk of developing non-proliferative DR among patients with T2D while fibrates may benefit diabetic macular edema. 36 37

Diabetes care in Sweden has significantly improved over time with a large proportion of persons with T2D obtaining a target HbA1c level <52 mmol/mol (6.9%). 38 However, that a relatively large proportion of patients have DR at diagnosis of T2D indicates that strategies for detecting T2D at earlier stages need to improve. Although diabetes care for persons with established T2D has substantially improved over time, detecting diabetes at an early stage has not achieved corresponding success. 12 When clearly elevated glucose levels exist before diagnosis, the harm due to legacy effects will likely not be evident until later years. 13 14

Guidelines suggest that overweight and obese individuals should be screened for T2D. 23 39 Other individuals in focus are first-degree relatives of individuals with T2D, that is, having a hereditary component. Specific risk scores exist that can be used for screening for T2D. 40 However, clearly structured programs for screening risk groups are lacking in most countries, while screening is generally random and, in many instances, may be missed. In the ADDITION study, structured screening for T2D was evaluated, but clear benefits on a population level could not be confirmed. 41 More research is needed into implementing structured screening programs for at-risk persons with T2D to detect disease at an early stage. Currently, by greater focus in clinical practice by extended screening of T2D, it may also be possible to detect pre-diabetes and prevent T2D more efficiently through lifestyle interventions. 42

In the current study, most risk factors for DR at diagnosis were expected. However, we did not expect that those with high BMI were less likely to have DR compared with those with normal weight. It is possible that individuals with normal BMI who end up developing T2D may be screened later for T2D after a more long-term hyperglycemia. It was also of interest that individuals born in Asia and then migrating to Sweden had higher risk of DR at diagnosis of T2D compared with those born in Sweden. One possible explanation is that this patient group may be less informed regarding T2D risk factors and need for screening. Another is that disease progression differs since persons born in Asia who are not overweight or obese generally develop T2D more often compared with those born in Western countries. 43 44 Retinopathy progression has shown to be more common in certain ethnic groups in earlier studies including Indian, Pakistani, and South Asian African ethnic groups. 10 45 46

One strength of the current study is the population-based design where the NDR covers the majority of persons with T2D in Sweden. A limitation is that 44% of the newly diagnosed had no data available in the NDR on retinopathy less than 2 years after diagnosis of T2D and were therefore not included in the current analysis. However, patient characteristics were similar overall among included and excluded patients indicating major selection bias is not likely. Although some patient characteristics differed between the included and excluded patients, they were overall numerically small, except for HbA1c where a somewhat greater difference existed at 58.1 mmol/mol (7.5%) vs 54.3 mmol/mol (7.1%). Mean HbA1c was somewhat lower among excluded patients possibly indicating slightly lower prevalence of DR in this population. Nevertheless, even if a lower proportion of excluded patients had DR, the overall proportion of patients having DR would still be relatively high. It is unclear to what extent those patients without data on DR in the NDR lacked a retinal screening or if results of screening had not been recorded. The NDR is dependent on health professionals registering information on retinopathy in the NDR based on clinical eye examinations. The study was limited that a minority of patients had information on albuminuria, creatinine levels, and grading of retinopathy at the time of diagnosis of T2D and these variables were therefore not included in the analyses.

Since a large proportion of persons with T2D in Sweden reach HbA1c targets, indicating high overall quality of diabetes care compared with many other countries, similar challenges in terms of detecting persons with T2D at an early stage of hyperglycemia seem likely in other European countries and parts of the world. It also seems likely that slightly lowering glycemic targets (eg, from 52 mmol/mol to 48 mmol/mol) in patients with established T2D, often intensively debated, may have relatively little influence on prognosis, 16 whereas many individuals with much higher levels remain undetected in turn leading to complications already at diagnosis. Therefore, we view early detection of T2D as a key challenge to resolve in the field of T2D.

In conclusion, intensified screening for T2D in clinical practice is needed in Sweden since almost one-fifth of these persons have retinopathy at diagnosis indicating long-standing hyperglycemia. The prevalence of DR was higher in certain patient groups including men, birthplace outside of Sweden, and those with a history of stroke, kidney, disease, and high SBP. Further research is needed to develop efficient strategies and programs to not only screen for T2D at random in clinical practice but also more structured screening to detect T2D earlier. This is of particular concern since many persons may have hyperglycemia before diagnosis and are not targets of efficient prevention strategies for complications before diagnosis.

Ethics statements

Patient consent for publication.

Not applicable.

Ethics approval

The study was approved by the Swedish Ethical Review Authority (Dnr 977-17). We conducted a registry-based study using data from the NDR. After patients provide verbal informed consent, data are reported directly to the NDR from clinical visits to primary care clinics and hospital diabetes clinics.

Acknowledgments

We thank all clinicians involved in the care of individuals with type 2 diabetes for data collection and we also thank the staff at the NDR. We also want to thank Caddie Zhou, Joakim Ruist, Albin Nydén, and Hanne Carlsen for support with statistical analyses

• Nentwich MM ,
• UK Prospective Diabetes Study (UKPDS) Group
• Diabetes Control and Complications Trial Research Group, ,
• The Diabetes Control and Complications Trial Research Group
• Stratton IM ,
• Kohner EM ,
• Aldington SJ , et al
• Harris MI ,
• Welborn TA , et al
• UK Prospective Diabetes Study Group
• Ren Q , et al
• Magliah SF ,
• Bardisi W ,
• Al Attah M , et al
• Abualrob M , et al
• Agarwal R ,
• Evans JR , et al
• Swedish National Diabetes Register
• Holman RR ,
• Bethel MA , et al
• Coleman RL , et al
• Tancredi M ,
• Rosengren A ,
• Svensson A-M , et al
• Appelros P ,
• Ludvigsson JF ,
• Svedberg P ,
• Olén O , et al
• Hellgren K-J , et al
• Seyed Ahmadi S ,
• Svensson A-M ,
• Pivodic A , et al
• John G , International HBA1c Consensus Committee
• American Diabetes Association
• World Health Organization
• Westreich D ,
• Wiesen C , et al
• Palladino R ,
• Khunti K , et al
• Looker HC ,
• Nyangoma SO ,
• Cromie D , et al
• Mahmood N ,
• Shaikh Z , et al
• Kaucsar E ,
• Oh TK , et al
• Borch‐Johnsen K , et al
• Qin Y , et al
• Bearse MA Jr ,
• Cavallerano JD ,
• Sun JK , et al
• Lin CL , et al
• Nationella Diabetesregistret
• Davidson KW ,
• Mangione CM , et al
• Lindström J ,
• Tuomilehto J
• Simmons RK ,
• Echouffo-Tcheugui JB ,
• Sharp SJ , et al
• Tuomilehto J ,
• Eriksson JG , et al
• Ramachandran A ,
• Snehalatha C ,
• Shetty AS , et al
• Nugawela MD ,
• Gurudas S ,
• Prevost AT , et al
• Olvera-Barrios A ,
• Anderson J , et al

Presented at The abstract has been presented orally at the European Association for the Study of Diabetes (EASD) in Hamburg, Germany, 2023.

Contributors All authors were involved in the design, analyses, interpretation of results, and critical review of the manuscript and approved submission. SS and ML wrote the first draft of the manuscript. ML as the corresponding author had full access to the studied data and takes responsibility for data integrity and accuracy of analyses.

Funding The Swedish state under the agreement between the Swedish government and the county councils, the ALF-agreement [ALFGBG-966173] and The Fyrbodal Research Council.

Competing interests SS has been a consultant for AstraZeneca, Boehringer-Ingelheim, Novo Nordisk, and Sanofi. KE-O has received fees for lecturing and/or for consulting from Sanofi, Novo Nordisk, Eli Lilly, and Abbott Diabetes Care. ML has been a consultant for Boehringer-Ingelheim, Eli Lilly, NordicInfu Care, and Novo Nordisk and received research grants from Eli Lilly and Novo Nordisk.

Provenance and peer review Not commissioned; externally peer reviewed.

Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Read the full text or download the PDF:

Research Suggests TNF Inhibitors Reduce Incidence of Diabetic Retinopathy

Findings support the role of these agents in patients with concomitant diabetes and rheumatic disease..

New data demonstrated that TNF inhibitors have a positive impact on retinal microvasculature and lower the incidence of diabetic retinopathy among rheumatic disease patients with type 2 diabetes.

The study also suggests that glycemic control may be the most critical factor for the development of diabetic complications in rheumatic disease patients, even among those receiving effective anti-inflammatory treatment, according to the investigators.

This cross-sectional analysis, which included 50 diabetic patients with rheumatic diseases (group 1) and an age-, sex-, and HbA1c-matched control cohort (group 2) of diabetic patients, aimed to better understand the impact of anti-TNF (biological) therapies on the incidence and progression of diabetic retinopathy.

Researchers assessed the presence or absence of diabetic retinopathy while also evaluating the following comorbidities as possible confounding factors: hypertension, coronary artery disease and hyperlipidemia.

In each group, 100 eyes of 50 patients was evaluated. Among patients in group 1, data showed that only three had non-proliferative retinopathy. Researchers reported that the median duration of rheumatic disease and diabetes was nine and 11 years, respectively. The mean duration of anti-TNF therapy was four years

When looking at the control group of diabetes-only patients, 13 developed some form newly diagnosed diabetic retinopathy over the last five years. Researchers observed that the calculated retinopathy occurrence between the groups was statistically significant, and they reported an incidence rate ratio for patients receiving anti-TNF treatment 0.4 in this analysis.

“In support of the literature, anti-TNF medication seems to reduce the incidence of diabetic retinopathy independent of its glucose-lowering effect,” the study authors noted in their recent Graefe's Archive for Clinical and Experimental Ophthalmology paper. “Although the outcomes of direct intravitreal administration studies have been controversial, with the increasing and widespread use of anti-TNFs, their anti-inflammatory role in the pathophysiology of diabetes and its complications has become a hot research topic.

“This effect can be attributed to its anti-inflammatory role in diabetes pathogenesis,” they concluded. “Further prospective studies with larger cohorts are necessary to establish the long-term impact of biologics on the development and treatment of DR.”

Related Content

• Vision Screening Programs Needed in Socioeconomically Disadvantaged Communities
• Stroke Associated with Increased Prevalence of Ocular Disease
• Pupil Tracking Spots Early Glaucoma
• Dream Update: Omega-3 Effect May Be Temporary
• Mismanagement Tops List of Glaucoma Lawsuits

Current Issue

Table of Contents

Read digital edition, read pdf edition, subscriptions, related topics.

Copyright © 2024 Jobson Medical Information LLC unless otherwise noted.

All rights reserved. Reproduction in whole or in part without permission is prohibited.

Faculty and Disclosures

As an organization accredited by the ACCME, Medscape, LLC requires everyone who is in a position to control the content of an education activity to disclose all relevant financial relationships with any commercial interest. The ACCME defines "relevant financial relationships" as financial relationships in any amount, occurring within the past 12 months, including financial relationships of a spouse or life partner, that could create a conflict of interest.

Medscape, LLC encourages Authors to identify investigational products or off-label uses of products regulated by the US Food and Drug Administration, at first mention and where appropriate in the content.

Hans-Peter Hammes, MD, PhD

Professor, University of Heidelberg, Heidelberg, Germany; Section Head, Endocrinology, University Hospital, Medical School of Mannheim, Mannheim, Germany

Disclosure: Hans-Peter Hammes, MD, PhD, has disclosed that he has served as an advisor or consultant to Novartis and Lilly.

Topics in Diabetic Retinopathy

The symposium entitled "Diabetic Retinopathy -- Diagnostic and Treatment Novelties [1] " centered on 2 important areas of research: (1) the retina as an additional independent risk indicator of cardiovascular morbidity and mortality and (2) clinical treatments.

Insulin-like Growth Factor-1 Antagonists in the Treatment of Retinopathy

Maria Grant, MD, [2] University of Florida, Gainesville, reported on 2 large, as yet unpublished studies in which patients with preexisting diabetic retinopathy were treated with an insulin-like growth factor (IGF) antagonist. With the report of Poulsen (1953) that patients with spontaneous destruction of the pituitary gland experienced resolution of proliferative diabetic retinopathy, a long series of experimental and clinical studies assessed whether the growth hormone (GH)-IGF-1 axis had a role in the initiation and/or progression of diabetic retinopathy. It was demonstrated by several groups that (1) patients with diabetic retinopathy had elevated serum IGF-1 levels; (2) patients with proliferative diabetic retinopathy had elevated IGF-1 levels in their vitreous; and (3) IGF-1 was acting with the other important retinal growth factor, vascular endothelial growth factor (VEGF), to induce retinal neovascularizations.

Experimental inhibition of the GH-IGF-1 axis in a model of acute proliferative retinopathy prevented new vessel formation, and small clinical trials by Dr. Grant [3] and researchers in Europe indicated that the administration of the somatostatin analog octreotide was beneficial. However, the need for frequent administration of the drug prevented wider therapeutic application. With the development of a long-acting release form of octreotide ( Sandostatin LAR, Novartis), the groundwork was established for larger clinical trials.

Dr. Grant reported the results of two of these trials, known as the 802 study and the 804 study. In the 802 study, 61 centers in 15 European countries recruited 585 patients who had moderate-to-severe nonproliferative to non-high-risk proliferative diabetic retinopathy (Early Treatment Diabetic Retinopathy Study [ETDRS] levels 47-61). The primary endpoints of the study were to assess the effect of 2 doses of octreotide long-acting release injection (20 mg and 30 mg once per month) on the progression of preexisting diabetic retinopathy, ie, prevention of a 3-step progression on the ETDRS retinopathy scale or a more than 2-step progression in an individual eye. Secondary endpoints were the time to development of macular edema and the loss of visual acuity. Safety and tolerability also were assessed.

In the 804 study, 313 patients from sites in the United States, Canada, and Brazil received 30 mg octreotide long-acting release injection once per month; primary endpoints were the same as the 802 study. (Patient characteristics will be described in detail soon in a full paper.) The most important finding was the significant delay in time to progression of retinopathy with the administration of 30 mg of octreotide in the 804 study (hazard ratio [HR], 0.6; P < .043). No effect was observed for visual acuity and progression to macular edema. Safety and tolerability were within margins of previous studies, the most frequent side effects being diarrhea, development of cholelithiasis, and mild hypoglycemia.

The 802 study failed to confirm these results. When comparing serum IGF-1 levels in the 2 studies as parameters of the efficacy of octreotide treatment, there was a difference in the level of suppression, indicating that the 804 trial had better patient compliance and study monitoring than the 802 study. Together, these data are consistent with the concept that the GH-IGF-1 system plays an important role in the propagation of retinal neovascularization, and that treatment with somatostatin analogs can significantly inhibit the clinically relevant progression to more severe stages of diabetic retinopathy.

Lipids and Retinopathy

There is an ongoing debate about whether elevated lipids are important in the pathogenesis of diabetic retinopathy. Several historical anecdotal reports have shown that diabetic patients with elevated lipids were more prone to diabetic macular edema, and that treatment with lipid-lowering drugs resolved these deposits. Because elevated lipids are involved in atherosclerosis and vessel stenosis, including stenosis of the carotid artery, an indirect relationship may exist between hyperlipidemia and diabetic retinopathy, given that moderate carotid artery stenosis protects from diabetic retinopathy, whereas more severe stenosis leads to ischemic retinopathy. The Atherosclerosis Risk in Communities (ARIC) study showed a weak but significant correlation between thickening of the carotid artery intima-media wall and diabetic retinopathy. [4] In that light, Paul Dodson, MBBS, MD, FRCP, FRCOphth, [5] Birmingham Heartlands Hospital, Birmingham, United Kingdom, summarized studies on the effect of statins and fibrates in the treatment of diabetic retinopathy.

In the Collaborative Atorvastatin Diabetes Study (CARDS), [6] approximately 1400 patients with type 2 diabetes received 10 mg of atorvastatin for primary prevention of coronary heart disease, which resulted in a 26% drop in total cholesterol and a 40% drop in low-density lipoprotein (LDL) cholesterol. Treatment duration was 4-4.5 years. At baseline and at annual follow-up, the investigators reported whether any fundal examination record from the previous year showed "no retinopathy," "nonproliferative retinopathy," "preproliferative retinopathy," or "proliferative retinopathy." Whether the patient had received photocoagulation in the past year was also noted, but the type or purpose of any photocoagulation was not recorded. Retinal photographs were not obtained. The investigators used accelerated failure time models with interval censoring to examine whether there was any treatment effect on retinopathy over a median 4-year follow-up.

The study's main problem was that there were considerable data missing, both at baseline and during follow-up. Of 2838 patients enrolled in CARDS, only 65% had retinopathy status recorded at baseline, and 39% of these patients had some retinopathy. There was no effect of treatment on progression of retinopathy severity by at least 1 step (6% lower rate in the atorvastatin group; P = .5). At least 1 follow-up recording of photocoagulation status was available in 2298 (81%) participants. Baseline status was available in just 1729 of these (61% overall). The incidence of photocoagulation was 6.03/100 person-years at risk (95% confidence interval [CI]: 5.28, 6.90) in the placebo group and 5.50/100 person-years at risk (95% CI: 4.81, 6.30) in the atorvastatin group. This 13% reduction in coagulation with atorvastatin treatment ( P = .4) increased to a 21% reduction on adjusting for baseline status but remained nonsignificant ( P = .1).

Firm conclusions about the effect of atorvastatin on retinal outcomes in CARDS are hindered by the lack of photographs and considerable missing data. Although there was no clear evidence of a treatment effect, the results, although nonsignificant, are consistent with some protective effect.

The Eye as a Risk Marker for Cardiovascular Morbidity and Mortality: Yes and No

The retinal circulation has long been considered a window to the systemic circulation. Studies by Gunn (1898) had already demonstrated retinal changes in patients with hypertension. However, ophthalmoscopy as a routine diagnostic procedure had been found too imprecise to allow for assessments of retinal vessel changes as a risk indicator of increased cardiovascular morbidity and mortality. Gabrielle Tikellis, PhD, [7] Center of Eye Research Australia, Melbourne, Australia, summarized data from the 6 studies over the past few years that have demonstrated clearly that vessel changes in the eye serve as a risk marker: ARIC, the Cardiovascular Health Study (CHS), the Beaver Dam Eye Study, the Blue Mountain Eye Study, the Wisconsin Epidemiologic Study of Diabetic Retinopathy Study (WESD), and the Rotterdam study.

Two distinct groups of lesions were analyzed with novel technologies to digitize and quantitatively analyze retinal photographs: (1) focal lesions, including arteriolar narrowing, arteriovenous nicking, microaneurysms/dot hemorrhages, cotton-wool spots, retinal arteriolar wall opacification; and (2) diffuse lesions, such as generalized arteriolar narrowing and changes in the arteriovenous ratio of vessel width (AVR). The overall prevalence of retinal signs of damage was 3% to 14%. A decrease in retinal arteriolar diameter was observed with increasing blood pressure, and a reduced AVR (meaning a progressive generalized narrowing of retinal arterioles) was associated with higher risk of developing metabolic syndrome or overt diabetes. Isolated venular dilatation was associated with developing proteinuria. Cotton-wool spots indicated a 6.4-fold greater risk for stroke, and microaneurysms indicated a 4-fold greater risk for stroke. These lesions were even more predictive for congestive heart failure. In persons with diabetes, a 4.5-fold greater increased risk for congestive heart failure was observed when retinopathy was present. In the Beaver Dam Eye Study, a reduction in AVR was significantly associated with hypertension, increased intima-media thickness, stroke, congestive heart failure, and cardiovascular mortality. With the improved precision of novel methods of analysis, the retina provides independent information on cardiovascular morbidity and mortality.

Providing a contrary view, Manon van Hecke, MD, PhD, [8] VU University Medical Center, Amsterdam, The Netherlands, pointed out that several of the above studies did not consistently find associations between focal and generalized changes in the eye and cardiovascular events, indicating that the issue may be much more complex than when viewed at first glance. Depending on the type of study and the parameters assessed, retinal vessel changes did not always reflect cardiovascular prognosis. For example, in the Hoorn Study, [9] a population-based cohort study of over 600 patients with type 2 diabetes, there was a significant association of retinopathy with all-cause and cardiovascular mortality. However, after controlling for diabetes duration, body mass index, and prior cardiovascular disease, this association was no longer found.

The mechanisms explaining why retinal vessels may not exactly mirror what happens to the macrovascular systems are, however, not clear. The Hoorn study addressed the following question: If microvascular dysfunction affects the risk for atherosclerosis, by which mechanism(s) does this occur? For that purpose, a cohort of 256 patients with type 2 diabetes was studied. Retinal photographs were taken and related to flow-mediated dilatation of forearm blood flow (endothelial-dependent) and to nitroglycerin-dependent vasodilatation (endothelial-independent). No difference was found in flow-mediated vasodilatation, nitroglycerin-dependent vasodilatation, and intima-media thickness for those with and without retinal abnormalities. The only retinal parameter that showed a significant association with intima-media thickness was retinal venular dilatation. Thus, retinal microvascular disease was not associated with large artery endothelial dysfunction in the established type 2 diabetes group.

Ruboxistaurin and Retinopathy

An oral presentation session on diabetic retinopathy was entitled "More than VEGF and PKC"; however, the most important paper in this session actually was about protein kinase C (PKC). Lloyd Paul Aiello, MD, PhD, [10] Joslin Diabetes Center, Harvard Medical School, Boston, Massachusetts, reported on a combined analysis of the trials with the PKC inhibitor ruboxistaurin in diabetic patients. The final analysis was done on 608 patients with the following characteristics: 59 years of age (SD 11); 85% had type 2 diabetes, with an average duration of 15 years; mean glycated hemoglobin (A1C) was 8.2%; 10% had retinopathy ETDRS level < 47, and 60% had ETDRS level of 47. The primary aim was prevention of retinopathy progression; the secondary aim was the prevention of vision loss. Although the primary target was not met, the reduction in sustained moderate vision loss was 4.2% (relative risk reduction, 41%; P = .011). Safety and tolerability were shown to be excellent in light of the need for long-term administration.

Calculating the Risk for Diabetic Retinopathy

Data from the United Kingdom Prospective Diabetes Study (UKPDS) have been used to calculate the risk for fatal and nonfatal macrovascular complications in patients with type 2 diabetes, and a downloadable version (UKPDS Risk Engine 2.0) is available online. [11] At the American Diabetes Association (ADA) meeting, Ruth L. Coleman, [12] Diabetes Trials Unit, University of Oxford, Oxford, United Kingdom, presented a risk calculator for diabetic retinopathy. UKPDS data were used to construct a model on the basis of 1949 patients for whom complete data were available. Their baseline characteristics (mean age, 54; A1C 7.1%; systolic blood pressure, 136 mm Hg; total cholesterol 210 mg/dL; HDL 43 mg/dL; body mass index [BMI] 28) did not differ from the original cohort of 5100 patients. After 12 years, a total of 50% of the patients had developed at least mild nonproliferative diabetic retinopathy. Univariate risk-factor analysis revealed macroalbuminuria, microalbuminuria, prior cardiovascular events, and metabolic control as the most important risk factors. In a multivariate analysis, macroalbuminuria and metabolic control were retained. Increasing age and smoking reduced the risk, whereas diabetes duration and systolic blood pressure contributed stepwise. Risk of developing diabetic retinopathy over t years in patients without diabetic retinopathy is:

Risk ( t ) = 1 - exp(-0.038 x 0.950 AGE-55 x 0.793 CURRENT SMOKER x 1.251 HbA1c-6.8 x 1.111 (SBP - 135.5)/10 x 2.800 MACROALBUMINURIA x 1.046 DIABETES DURATION x Sigmas 0.174

Thus, the likelihood of developing diabetic retinopathy in patients with type 2 diabetes can be calculated from available clinical information with sufficient precision. Internal validation of the system has been performed successfully, whereas external validation awaits.

• Diabetic retinopathy -- diagnostic and treatment novelties. Symposium. Program and abstracts of the American Diabetes Association 66th Scientific Sessions; June 9-13, 2006; Washington, DC.
• Grant M. Treating diabetic retinopathy with IGF-1 antagonists. Diabetic retinopathy -- diagnostic and treatment novelties. Symposium. Program and abstracts of the American Diabetes Association 66th Scientific Sessions; June 9-13, 2006; Washington, DC.
• Grant MD, Mames RN, Fitzgerald C, et al. The efficacy of octreotide in the therapy of severe nonproliferative and early proliferative diabetic retinopathy: a randomized controlled study. Diabetes Care. 2001;24:182-183. Abstract
• Klein R, Sharrett AR, Klein BE, et al. The association of atherosclerosis, vascular risk factors, and retinopathy in adults with diabetes: the atherosclerosis risk in communities study. Ophthalmology. 2002;109:1225-1234. Abstract
• Dodson P. The effect of statins on diabetic retinopathy. Diabetic retinopathy -- diagnostic and treatment novelties. Symposium. Program and abstracts of the American Diabetes Association 66th Scientific Sessions; June 9-13, 2006; Washington, DC.
• Colhoun HM, Betteridge DJ, Durrington PN, et al. Primary prevention of cardiovascular disease with atorvastatin in type 2 diabetes in the Collaborative Atorvastatin Diabetes Study (CARDS): multicentre randomized placebo-controlled trial. Lancet. 2004;364:685-696. Abstract
• Tikellis G. The eye is a risk marker for cardiovascular morbidity and mortality. Diabetic retinopathy -- diagnostic and treatment novelties. Symposium. Program and abstracts of the American Diabetes Association 66th Scientific Sessions; June 9-13, 2006; Washington, DC.
• van Hecke MV. The eye is not a risk marker for cardiovascular morbidity and mortality. Diabetic retinopathy -- diagnostic and treatment novelties. Symposium. Program and abstracts of the American Diabetes Association 66th Scientific Sessions; June 9-13, 2006; Washington, DC.
• van Hecke MV, Dekker JM, Nijpels G, et al. Are retinal microvascular abnormalities associated with large artery endothelial dysfunction and intima-media thickness: the Hoorn study. Clin Sci (Lond). 2006;110:597-604. Abstract
• Aiello LP, Vignati L, Sheetz MJ, et al. Effect of ruboxistaurin (RBX) on diabetic macular edema (DME) and visual loss:a meta-analysis of the PKC-DRS and PKC-DRS2. Program and abstracts of the American Diabetes Association 66th Scientific Sessions; June 9-13, 2006; Washington, DC. 230 OR.
• Diabetes Trials Unit, University of Oxford. UK Prospective Diabetes Study. Available at: http://www.dtu.ox.ac.uk/index.html?maindoc=/ukpds/ Accessed June 29, 2006.
• Coleman RL, Stevens RJ, Aldington SJ, Holman RR. Estimating risk of clinically evident retinopathy in type 2 diabetes. Program and abstracts of the American Diabetes Association 66th Scientific Sessions; June 9-13, 2006; Washington, DC. 229-OR.

Medscape Diabetes © 2006 

Deep Transfer Learning-Based Automated Diabetic Retinopathy Detection Using Retinal Fundus Images in Remote Areas

• Research Article
• Open access
• Published: 29 May 2024
• Volume 17 , article number  135 , ( 2024 )

Cite this article

You have full access to this open access article

• Ayesha Jabbar 1 , 2 ,
• Shahid Naseem 3 ,
• Jianqiang Li 1 , 4 ,
• Tariq Mahmood   ORCID: orcid.org/0000-0002-4299-7756 5 , 6 ,
• Kashif Jabbar 2 ,
• Amjad Rehman 5 &
• Tanzila Saba 5  

346 Accesses

Explore all metrics

Diabetic retinopathy (DR) significantly burdens ophthalmic healthcare due to its wide prevalence and high diagnostic costs. Especially in remote areas with limited medical access, undetected DR cases are on the rise. Our study introduces an advanced deep transfer learning-based system for real-time DR detection using fundus cameras to address this. This research aims to develop an efficient and timely assistance system for DR patients, empowering them to manage their health better. The proposed system leverages fundus imaging to collect retinal images, which are then transmitted to the processing unit for effective disease severity detection and classification. Comprehensive reports guide subsequent medical actions based on the identified stage. The proposed system achieves real-time DR detection by utilizing deep transfer learning algorithms, specifically VGGNet. The system’s performance is rigorously evaluated, comparing its classification accuracy to previous research outcomes. The experimental results demonstrate the robustness of the proposed system, achieving an impressive 97.6% classification accuracy during the detection phase, surpassing the performance of existing approaches. Implementing the automated system in remote areas has transformed healthcare dynamics, enabling early, cost-effective DR diagnosis for millions. The system also streamlines patient prioritization, facilitating timely interventions for early-stage DR cases.

Similar content being viewed by others

Explaining Eye Diseases Detected by Machine Learning Using SHAP: A Case Study of Diabetic Retinopathy and Choroidal Nevus

Deep Learning Approach for Early Detection of Alzheimer’s Disease

A deep learning framework for early diagnosis of Alzheimer’s disease on MRI images

Avoid common mistakes on your manuscript.

1 Introduction

In 2019, the International Diabetes Federation reported that the global number of people suffering from diabetes mellitus (DM) was approximately 463 million, projected to reach around 700 million by 2045. Diabetic retinopathy (DR) is a retinal disorder resulting from impaired glucose tolerance, which leads to the deterioration of retinal capillaries as a consequence of diabetes [ 1 ] It can result in visual issues, making early diagnosis of the severity of DR crucial for effective treatment [ 2 , 3 ]. DR affects about 80% of diabetics, and its early stages often have no symptoms, though some individuals may notice minor eye-light changes. It is classified as an inflammatory neurovascular disorder with neurological impairment, eventually leading to compromised vision and potential blindness. Consequently, DR represents a significant and prevalent disease in remote areas, especially where proper medical facilities are unavailable or limited [ 4 , 5 ]. Diabetic retinopathy is challenging to diagnose and requires multiple clinical tests, such as visual acuity assessment, pupil dilation, and computerized tomography [ 6 ].

1.1 Diabetic Retinopathy

One of the leading causes of visual loss is diabetic retinopathy in both developing and industrialized nations’ working populations. Diabetic people are 25 times more at risk of becoming blind as compared to healthy people [ 7 , 8 ]. Retinopathy is a condition that affects the retina and the blood capillaries due to diabetes. Blood vessels are tree-shaped branching patterns with minor curvature that arise from the optic disc. The vessel’s average diameter is roughly 1/40 of the retina diameter. The bright yellowish disc from which arteries and optic nerve fibers emerge is known as the optic disc. Electrical impulses were sent from the retina to the visual cortex via the optic disc. It has a diameter of 1.5–2 mm [ 9 ]. Diabetic retinopathy in its initial phase is known as non-proliferative diabetic retinopathy (NPDR). At least one microaneurysm is present in this condition, with or without hemorrhages in the retina, hard fluid, cotton wool patches, or venous loops. Microaneurysms are the most common clinical abnormality in the eye [ 1 , 10 ]. They may appear as minute, dark red spots or as tiny hemorrhages inside the light-sensitive retina, either alone or in clusters. Some blood veins that supply the retina get blocked as the disease advances, and this stage is known as moderate NPDR. Severe NPDR is the following stage, where several more capillaries are blocked [ 11 ]. Both types of diabetes, diabetes mellitus and diabetes insipidus, induce diabetic retinopathy. In its early stages, it is asymptomatic, but if left untreated for an extended period, it can result in irreversible eyesight loss. The issue is that patients may not be aware of it until it has progressed to an advanced stage. Diabetes mellitus can cause proliferative diabetic retinopathy (PDR), which is a condition that can cause permanent vision impairments. As the disease progresses, eyesight loss becomes unavoidable. Diabetes-related retinopathy is the primary cause of blindness worldwide, especially in developing nations. Therefore, there is a critical need for an effective and accurate diagnostic system [ 12 , 13 ]. Figure   1 illustrates the various levels of diabetic retinopathy disease.

Different levels of diabetic retinopathy

To prevent vision loss caused by this severe disease, the global scientific community strives to develop more efficient and precise methods for early identification of diabetic retinopathy [ 14 ]. Utilizing automated detection techniques and advanced technologies can significantly save the diagnostic process time, effort, and resources [ 15 ]. Implementing an automated detection system holds great potential for providing high-quality eye-related services in remote areas, emphasizing the importance of addressing this issue and offering an automated solution as a preventive measure. Deep learning (DL) algorithms have emerged as a prominent solution for various medical imaging analysis challenges, surpassing the limitations of traditional machine learning methods [ 16 , 17 ]. Deep learning models excel in swiftly identifying significant features in retinal images without human intervention. By employing multiple processing layers, these models can learn data representations at different levels of abstraction. These advancements have made substantial progress in voice recognition, pattern recognition, object identification, and other domains. DL’s backpropagation approach allows the exploration of complex pattern structures within extensive datasets, enabling the model to update its internal parameters and compute representations for each layer based on the input from the preceding layer.

DL-based has emerged as a focal point in various academic disciplines in recent years, driven by its remarkable ability to directly extract meaningful features from training data [ 18 ]. DL is now considered a promising image/video categorization and detection technology. DL algorithms employ sophisticated processes, including data processing and abstraction construction, to optimize performance. In genomics, machine learning is extensively utilized to unveil intricate relationships within data and generate novel biological insights [ 19 ]. However, the continuous growth of genomics data demands even more powerful DL models to unlock deeper insights. DL’s transformative impact is evident in domains such as machine learning and natural language processing, where it efficiently harnesses vast datasets to achieve significant advancements [ 20 ].

Deep learning has emerged as a potent instrument for improving computer-aided diagnosis (CAD) systems, resulting in enhanced diagnostic accuracy, expanded coverage of disorders, and the implementation of real-time medical image disease detection systems. Notably, Wang et al. [ 21 ] introduced a DL architecture based on the U-net model, which effectively differentiated optic discs in diabetic retinopathy detection. Utilizing convolutional neural networks (CNNs), the DL model independently analyzed dark and shaded retinal fundus images, yielding distinct and accurate segmentation outputs. DL’s reliance on vast datasets can be challenging despite its effectiveness. To address this, transfer learning has emerged as a valuable approach, reducing the demand for extensive data by leveraging previously optimized model parameters over new data distribution domains. This technique reduces training time when adapting the model to related tasks with different datasets [ 22 ]. Often, models pre-trained on complex feature datasets like ImageNet are employed for transfer learning, leading to improved accuracy across various tasks [ 23 ]. With diabetes cases projected to reach 552 million worldwide by 2030, telemedicine programs need to enhance their imaging capabilities and embrace automated image analysis methods to effectively address and manage diabetic retinopathy [ 24 , 25 , 26 ].

1.2 Problem Statement

The increasing prevalence of diabetic retinopathy in remote areas and limited access to medical resources and timely diagnosis present a significant healthcare challenge. The scarcity of doctors has led to more undetected cases, raising the risk of vision impairment and blindness among affected individuals. Addressing this pressing issue is vital to prevent further escalation of the disease’s impact on public health. Harnessing cutting-edge methods and sophisticated technologies is essential to address this healthcare disparity. Advancements in medical imaging and diagnostic techniques offer transformative opportunities, especially for conditions like diabetic retinopathy. Among these technologies, the fundus imaging system stands out for its exceptional ability to capture detailed and comprehensive retinal images. This study introduces a groundbreaking fundus camera-based real-time monitoring system designed to enable early identification of diabetic retinopathy. The system intends to provide accurate and efficient disease detection by employing cutting-edge computer vision algorithms and DL-based techniques. The integration of real-time monitoring capabilities enables timely interventions, optimizing patient outcomes and enhancing disease management. Beyond immediate patient benefits, the proposed automated monitoring system has the potential to revolutionize healthcare in remote regions. Optimizing medical resources and streamlining disease detection can alleviate the burden on healthcare facilities, empowering them to focus on critical cases and adopt a comprehensive approach to diabetic retinopathy management. Patients with diabetic retinopathy stand to derive substantial advantages from this technological breakthrough. The system’s automated nature and remote accessibility empower individuals to manage their health proactively. Regular monitoring and timely interventions are essential in preventing disease progression, ultimately reducing the risk of vision loss and lessening the burden on individuals and healthcare systems.

1.3 Aim of the Research

The proposed model aims to develop a system capable of analyzing images and accurately identifying the level of diabetic retinopathy. The suggested model utilizes a more straightforward and lighter approach, requiring less CPU computing power, making it suitable for smaller devices. Moreover, the suggested method demonstrates improved accuracy in data categorization compared to existing models. This diagnosis system would aid clinicians in evaluating the disease by detecting diabetic retinopathy by capturing retinal images using a fundus camera.

1.4 Objectives of the Research

This study aims to contribute significantly to diabetic retinopathy diagnosis by introducing innovative and effective methods for assessing the disease and providing optimal patient care.

to improve the classification accuracy of diabetic retinopathy stages through preprocessing approaches such as transformation, non-local mean denoising, and image filtering.

To address data imbalance by applying data augmentation specific to each grade.

To employ transfer learning-based models for automatic feature extraction, reducing training overhead and handling insufficient annotated training data.

To develop a healthcare system for diabetic retinopathy using a fundus camera to capture eye images

To focus on accurately classifying the level or stage of diabetic retinopathy disease.

To generate automated reports for further consultations or medication decisions.

The framework of the article is as follows: Sect.  2 provides an exhaustive overview of related work in the disciplines of detection, segmentation, colorization, and classification. Section  3 provides an overview of the dataset, processing, data augmentation, and feature extraction techniques. The Sect.  4 describes the framework, mathematical modeling, and algorithms used in the study, along with an overview of the model. The empirical findings are thoroughly analyzed and discussed in Sect.  5 . Finally, Sect.  6 the conclusion is presented and outlines possible future directions for research.

2 Literature Review

Kimar et al. [ 24 ] proposed a system to aid prediction and diagnosis in cases of DR. The system’s effectiveness and utility for DR applications were demonstrated, showcasing its potential for improved diagnostic capabilities. Iqbal et al. [ 3 ] suggested a computer-assisted approach for rating the severity of different phases of DR. Their method involved extracting unique characteristics from retinal layers, resulting in twelve distinctive features. By employing a deep fusion approach for classification, the authors achieved high accuracy rates, with 93% accuracy for binary classification between normal and retinopathy grades and 98% accuracy for mild/moderate DR grade classification. Ali et al. [ 27 ] suggested a three-classification approach to identify retinopathy lesions. They utilized a filter bank approach to retrieve lesion candidates for each lesion and subsequently categorized DR lesions. The system’s performance was thoroughly validated using various performance measures. Azar et al. [ 5 ] developed a robust segmentation technique for detecting and evaluating DR and maculopathy. After preprocessing, distinct portions of the retinal image, including the fovea, microaneurysms (MAs), exudates (EXs), and hemorrhages (HMs), were segmented. Researchers used concentric circular zones to differentiate maculopathy and non-proliferative diabetic retinopathy (NPDR) using aberrant fundus images and abnormalities patterns. Skouta et al. [ 28 ] proposed an automated classification approach for diabetic retinopathy, emphasizing contrast normalization for better differentiation between noise spots and microaneurysms (MAs) detection.

Sabeena et al. [ 29 ] utilized customized convolutional networks to differentiate retinal impairments, such as hemorrhages, microaneurysms, and neovascularization. They employed a variety of pre-trained networks to surpass the performance of existing systems. Ramachandran et al. [ 30 ] developed a tailored strategy for feature learning and characterization of non-diabetic retinopathy (DR) stages from DR-affected stages using deep convolutional neural networks (CNNs). To enhance the effectiveness of the DR classification system, they integrated image information with the extracted CNN features. Das et al. [ 31 ] applied a heat map optimization approach for CNN training to detect DR disease. Their method involved image and lesion-level classification to detect lesions such as hard or soft exudates and hemorrhages. Khana et al. [ 32 ] introduced a CNN design based on the Inception-V3 model for distinguishing between mild to severe DR, diabetic macular edema, and completely gradable cases. Rachapudi et al. [ 33 ] developed various segmentation methodologies for identifying retinal vessels, hard exudates, and microaneurysms. It utilized PCA for feature extraction and neural networks for two-class categorization. Minija et al. [ 34 ] introduced a textural analysis approach using deep CNNs for identifying blood arteries and hemorrhages in diabetic retinopathy. Their process involved diagnosis and severity level grading of retinal fundus images through a two-step approach. Saranya et al. [ 35 ] designed a three-stage categorization system for classifying fundus images of the retina in diabetic retinopathy using colored fundus images. Saranya et al. [ 36 ] developed a patch-based microaneurysm detection technique for five-stage diabetic retinopathy severity grading, utilizing random forest and neural network classifiers. PCA and random forest were employed for dimensionality reduction and performance improvement [ 37 ].

The advancements in deep learning have profoundly impacted various fields, including machine learning and natural language processing, primarily due to its ability to leverage vast datasets. In genomics modeling tasks, deep learning has become the preferred approach for estimating the influence of genetic variation on gene regulation mechanisms like DNA availability, especially in tasks related to DR. DL-based models, such as deep artificial neural networks (ANNs), have proven effective in capturing complex patterns in data, making them valuable tools for medical image disease detection systems [ 38 ]. The author of this study developed a specialized CNN structure based on the U-net model, designed explicitly for recognizing optical plates in retinal fundus images. By analyzing dark and shaded fundus images independently, the CNN produced distinct outputs, leading to improved detection of disorders and enabling real-time medical image disease detection [ 39 ]. Previous research often relied on end to end training of deep learning systems from raw fundus images to DR grade labels. However, these systems can miss critical lesion characteristics due to the inherent "black-box" nature of deep learning. To overcome this limitation, the study developed a lesion-aware subnetwork that enhances the ability to extract lesion characteristics. By embedding prior knowledge into the DR grading network, the study aimed to enhance the overall efficacy of the grading system and guide the model to make more informed decisions [ 40 ]. This research uses a DeepDR network trained end to end, utilizing features from the lesion-aware sub-network and original images, unlike previous studies using multiple CNNs to identify and classify lesions. This comprehensive approach resulted in improved grading outcomes and better performance for diagnosing various phases of DR in real-world datasets [ 41 ]. In other relevant works, researchers [ 42 ] developed a model to identify optic disc (OD) and optic cup (OC) by accurately measuring their respective areas. They employed watershed change and morphological separation procedures for locating and differentiating OD, achieving high predictive performance scores in their testing. Similarly, in [ 43 ], a method was proposed that successfully localized various shades of retinal fundus images and achieved a mean predictive performance of 92.8%, showcasing the effectiveness of their approach.

In achieving optimum picture classification outcomes, it is common practice to include handmade and non-handcrafted features inside a computer vision system [ 44 ]. Non-handcrafted features, including PCA, CBD, and CNN, were extracted, while handcrafted features were derived from localized phase quantization and completed local features. Parthiban et al. [ 45 ] introduced a diabetic retinopathy (DR) evaluation technique based on Google Inception, outperforming licensed ophthalmologist grading. Pratt et al. [ 46 ] adapted an existing denoising method for classifying a dataset of fundus images of the retina. They designed a convolutional neural network (CNN) capable of distinguishing microaneurysms, hemorrhages, and exudates by identifying intricate characteristics, particularly for the early detection of microaneurysms. Belderrar et al. [ 47 ] developed an autonomous diabetic retinopathy disease detection method. They pre-processed fundus images using a fuzzy histogram-based approach, followed by feature learning and intensity grade classification. Mansour et al. [ 48 ] developed a computer-based diagnostic method for classifying damaged blood vessels and fundus images into distinct categories. Hasan et al. [ 49 ] presented a novel detection method using convolutional neural networks with leveraged residual connections. This approach enhanced the authenticity of the detection system by strengthening the CNN model through PCA utilization [ 50 ]. The Kaggle dataset was employed to assess the accuracy of DR diagnosis, and the results showed promising performance compared to other chronic and semi-DR categorization methodologies. Automated detection of this disease can assist ophthalmologists in managing their workload by identifying patients who require essential ophthalmic care. Despite the potential of CNNs in DR diagnostic frameworks, they still pose challenges for clinical applications [ 40 ]. DL-based methods often use numerous image patches for image-level classification, embracing short-range dependencies and ensemble approaches such as SVM or majority voting. However, these models frequently overlook long-range connections.

3 Materials and Methods

The methodology employed in this study involves up-sampling and down-sampling techniques applied to the dataset, segmentation of the optic disc, preprocessing of images for transformation, and the use of a CNN-based model. The suggested model’s comprehensive design is depicted in Fig.  2 . To address data imbalance, the method initiates sampling and downscaling, effectively mitigating the issue in subsequent stages of the system. Next, optical disc segmentation is employed to prevent skewed findings, as the optical disc’s presence may lead to erroneous decisions due to its resemblance to lesions. Preprocessing follows, reducing irrelevant noise and highlighting pertinent information crucial for precise predictions [ 51 ]. This preprocessing procedure is straightforward and efficient, requiring minimal computing resources.

The figure illustrates the detection process of the suggested model for diabetic retinopathy

The provided figure Fig.  2 illustrates the detection process for diabetic retinopathy used in the study as its visual realization. It states the iterative stages in diabetic retinopathy detection including data balancing, image pre-processing and classification. The process has each operation illustrated which readers understand on the spot. The figure breaks down the detection process into separate steps, which give a structured representation of how the model works, thus enabling the understanding of the employed methodology.

The proposed methodology investigates an innovative technique for the identification and categorization of diabetic retinopathy, with a focus on assessing its degree of severity. Identifying DR illness can be challenging clinically, especially in mild and moderate stages. This approach aids clinical professionals in early disease prediction using fundus images, potentially improving patient satisfaction. The system acquires essential data, such as eye images, using a fundus camera integrated with the system. The acquired data undergoes preprocessing, including edge detection, resizing, and normalization. The CNN classifier receives a completely preprocessed picture and proceeds to assess the level of severity of diabetic retinopathy (DR).

The process flow diagram for the classification and image processing is shown in Fig.  3 , which includes several preprocessing steps, such as edge detection, scaling, interpolation, and normalization. The CNN algorithm is utilized for classifying the severity degree of DR. In the developed model, the detection, analysis, and treatment of diabetic retinopathy, along with the consultation process. The retinal image is acquired through a fundus camera and then undergoes preprocessing, including edge detection and resizing. The fully preprocessed image is further segmented to detect the different disease levels of diabetic retinopathy. If the patient is healthy, the diagnosis concludes; otherwise, if the image shows disease detection, a report is sent to connected doctors for telemedicine treatment and further care [ 52 ].

The proposed model for diabetic retinopathy uses data balancing, optical disc segmentation, and preprocessing to detect and classify the condition. This method improves patient care and diagnosis accuracy by overcoming noise and enhancing relevant information, thereby enhancing the overall effectiveness of the model

The detection process is summarized in Fig.  3 . It presents the main processing steps which include edge detection, scaling, interpolation, and normalization, followed by the classification using the CNN algorithm. Its simplicity notwithstanding the figure reflects the core of the detection process and thus allows readers to grasp the main features of the methodology easily.

3.1 Dataset

In this research, the IDRiD and MESSIDOR datasets from ADCIS were utilized. The MESSIDOR collection consists of 1200 raw retinal color fundus images, categorized into four classes (0–3) depending on the severity of Diabetic Retinopathy. Class 0 represents a normal eye image with no signs of disease, while Class 1, Class 2, and Class 3 depict varying levels of disease severity. The MESSIDOR dataset exhibits class imbalance, with most Class 0 images accounting for almost 50% and a limited number of Class 1 images, possibly contributing to model bias. The IDRiD dataset comprises 413 training retinal images and about 103 testing images. These images represent either the left or right eye, with no specific age group information provided [ 53 ]. Additionally, retina fundus images from the public dataset EyePACS, available on Kaggle.com, were used. The ophthalmologists assigned labels to these images and categorized them into five classes: normal, mild, moderate, severe, and proliferative diabetic retinopathy (DR). Detailed information about this dataset is presented in Table  1 .

Table  1 shows the Kaggle EyePACS dataset used in this study in detail. The dataset is classified into five classes per the severity of diabetic retinopathy which ranges from normal to proliferative diabetic retinopathy (PDR). The table shows the number of images for each class as well as its corresponding percentage on the total dataset. The table being presented gives a detailed proportion of the dataset composition and as such it presents valuable information with regards to the distribution of different severity levels which is very crucial for the training and evaluation of machine learning models. Scientists can use this information to evaluate the representativeness of the data set and the existing biases.

3.2 Data Upsampling and Downsampling

The research incorporates upsampling and downsampling techniques to achieve a balanced dataset and address data imbalance. Upsampling involves replicating photographs in regions where data is scarce, ensuring a balanced representation of classes and avoiding bias towards a specific category. On the other hand, downsampling involves removing samples from specific classes, preventing them from dominating the dataset and introducing bias. This approach guarantees that the classification algorithm is not adversely impacted by randomly reducing data from the dataset. In the study, the Kaggle dataset also exhibits imbalance, prompting the implementation of upsampling and downsampling methods on photos belonging to different series [ 26 ].

3.3 Segmentation of Optic Disc

In the fundus image of the eye retina, the presence of the optic disc poses a challenge in distinguishing bright lesions from dark lesions. The equivalent intensity values of the two lesions make it difficult for the neural network to differentiate between them reliably. As a result, the optic disc needs to be removed to minimize its detrimental impact on the predictions. However, this presents a unique set of problems as the optic disc exhibits comparable intensity changes as bright lesions [ 54 ]. The typical retinal fundus image is depicted in Fig.  4 a, and the retinal image after subdividing the optic disc in Fig.  4 b.

The figure highlights the success of a proposed system’s optic disc segmentation process to eliminate the optic disc’s impact on accurately identifying bright and dark lesions in the fundus image of the eye retina. The technique effectively finds the brightest point, isolating a single pixel as the brightest within the image. This segmentation process greatly improves the model’s ability to distinguish lesions, leading to better diagnosis and management of diabetic retinopathy disease without introducing any bias

This Fig.  4 b presents optic disc segmentation, an essential component of the diabetic retinopathy detection process. It visually depicts the segmentation method that removes the optic disc from retinal fundus images which helps in the accurate detection of lesions. The comparison of the before and after images highlights the effect of segmentation on improving the visibility of lesions, e.g., microaneurysms and hemorrhages. The figure, illustrating the outcomes of the segmentation method, serves as visual proof of the approach’s effectiveness, thus exposing its role in the precision improvement of diabetic retinopathy. A thorough examination of the dataset revealed that the brightest pixel inside the picture is usually located at the middle side of the optic disc. This is because of the imaging technology used to acquire retinal fundus images. To address this issue, a novel approach for optic disc segmentation was developed in the proposed system. Instead of relying on intensity values, the approach searches for the brightest point. Global maxima are employed to find the brightest spot in the picture, resulting in a single brightest pixel within the image. Once the brightest point is located, a circle is drawn around it, and the rest of the picture is removed. This approach effectively addresses the challenges of the optic disc segmentation process [ 54 ].

3.4 Image Preprocessing

Experiments revealed the optimal approach for enhancing effectiveness and utility in image preprocessing involves cropping, eliminating optic disc, resizing, interpolation, and normalizing [ 1 , 55 ]. Edges in a picture are described as abrupt changes/discontinuities that really can encode almost as much data as pixels. Smoothing the image to decrease noise and then identifying these abrupt changes in intensity are all part of the canny edge detection procedure. Reducing noise, discovering the intensity gradient, non-maximum suppressing, and hysteresis threshold are all steps in the process. Most accessible fundus photos are lossy compressed, causing minute structures like MAs to be distorted. That makes it challenging to isolate the most distinctive characteristics. Furthermore, due to the tiny length of MAs, it is critical to limit the noise impact. As a result, before moving on to the actual detection procedures, it required to examine a lot of image smoothing. An Image enhancement with something like a width of 7 and a variance of 1.0 is used in our implementation. If you need to reduce or increase the number of pixels in an image, resize it. When you enlarge an image, the pixel information changes. An image is expanded based on the estimate to get an image of a large size. The resolution of the photographs used is 2240 \(\times \) 1488 pixels. Computing such huge pictures is computationally unfeasible since more pixels means many heavy loads linked together with the image, which requires more processing power for modification. Interpolation is a technique for inserting additional data points into a set of existing discrete data points. It is an estimating function that can be utilized for locating lost data. The procedure used here is the method of non-adaptive interpolation, which is required when the image is scaled. Image interpolation operates in both planes. The suggested approach deemed the best for resizing employs inter-area- Area Interpolation. This approach uses the pixel area relationship to resample, resulting in noise-free output. The normalizing technique is used to alter the limits of pixel values. It may also be called contrast stretching as well as histogram stretching. It is done to reduce image noise and output the values back to the intensity. To do so, divide all the pixels by 255 to put them within the range of zero to one. Figure  5 a, b. show a normal fundus eye picture and a fundus image after preprocessing [ 56 ].

Retinal fundus image preprocessing

The image processing steps deployed on retinal fundus images before classification are depicted in Fig.  5 . It displays a normal eye fundus photograph and a preprocessed fundus image with cropping, resizing, and normalization adjustments made during the preprocessing. Visualizing the preprocessing pipeline with a figure, the latter allows the readers to understand how methods improve the quality and suitability of images for further analysis. It emphasizes the role of preprocessing in improving image quality and enabling precise diabetic retinopathy diagnosis.

3.5 Data Augmentation

The training dataset significantly impacts deep learning algorithms’ performance, making it crucial to have a larger dataset with intricate network architecture [ 57 ]. Medical imaging datasets are typically limited, so researchers use distinct data augmentation techniques, such as rotation, shifting, flipping, and cropping, to improve the training dataset. Cropping removes surplus areas, while rotation angles ranging from 0 to 180 degrees are applied to the patches. Shifting operations are also performed within a specified frame, and flipping techniques are employed to enhance the training dataset.

3.6 Feature Extraction Process

This process aims to find all micro-aneurysms (MAs) in the pre-processed picture and pick them. Micro-aneurysms show distinct patterns that are not related to the vessels. Micro-aneurysms can be classified depending on their form, size, and intensity level. Micro-aneurysms are microscopic red spots that range in size from 10 to 100 microns in diameter and are round. The prospective micro-aneurysms are differentiated by isolating them from the capillaries after pre-processing the picture. MAs and vessels are reddish and cannot be found on vessels. Blood vessels have an enormous surface area and are related components. Therefore, they may be distinguished from MA by their size. Experimentation is used to determine the threshold value. Objects with an area more significant than the threshold value are removed to eliminate blood vessels, as seen in Fig.  2 . Micro-aneurysms and noise may appear in the final image, which are disconnected vessels and other particles in the fundus picture. MAs have a diameter of 10–100 microns; therefore, they may be distinguished from noise by their area. Experimentation yielded two threshold values for removing noise objects with areas bigger and less than MAs. The generated picture contains objects with similar areas, some of which are micro-aneurysms [ 58 ].

3.6.1 Blood Vessels

The human eye’s digital fundus photography provides crisp pictures of the blood vessels in the retina. This approach provides a good view of the health of a DR patient. Figure  6 depicts an example of blood vessel identification using several forms of DR. The blood artery architecture was derived by applying diverse image processing methods to the green channel of the RGB fundus image [ 59 ]. Two-dimensional matching filters were used to locate blood vessels. The Gaussian distribution may be used as an approximation for the gray-level profile of a blood vessel cross-section. The method of vascular approximation was used, whereby the matched filter detection technique was applied to identify sections of blood vessels that demonstrate a piecewise linear pattern.

Results of blood vessel detection for normal and PDR

Figure  6 presents the outcomes of blood vessel detection in normal and PDR cases. It shows that the detection method is successful in identifying blood vessels on retinal images. Such identification is necessary for diabetic retinopathy diagnosis. The figure presents sample images with annotated blood vessels which serves as visual evidence of the accuracy of the detection process and highlights how much it matters to assess retinal health. The graph provides evidence of the capability of the method of detection to accurately detect blood vessels which, consequently influences the accuracy of the system.

3.6.2 Exudates

Exudates are buildups of lipids and proteins in retinal tissue, causing dazzling white or cream-colored lesions. These conditions indicate vascular permeability and potential retinal edema. Although not directly threatening to vision, they indicate fluid accumulation in the retina. Lesions near the macula center are considered sight-threatening. They are often encountered in conjunction with micro-aneurysms. As a result of these micro-aneurysms indicating increasing leakage, the traditional lesion is a circular ring of exudates having multiple micro-aneurysms at its core. Figure  7 is an example of exudate detection from several kinds of DR. The images depict black regions indicating the absence of exudates, while white regions indicate the presence of exudates. Eliminating prominent retinal components, such as the blood vascular network and the optic disc, is a pivotal step in the extraction procedure. After removing these structures, the exudates were discovered using image processing methods.

The findings about the identification of exudates in individuals with normal PDR

Figure  7 illustrates the procedure of identifying the exudates, supposed to be the lipid and protein aggregations in retinal tissue that imply the possibility of retinal edema. These lesions commonly seen near the macular center correspond to vascular permeability and fluid accumulation. In the Fig.  7 , exudates are marked as white or cream-colored lesions in contrast to a black background which signifies the presence or absence of the space occupying the lesion. The method is based on the segmentation of exudates in the image by eliminating other retinal components including blood vessels and the optic disc/ disk of the eye. This visualization will help us understand how exudates are recognized in retinal images which is fundamental for diagnosing diabetic retinopathy.

3.6.3 Detection of Microaneurysms

Detecting microaneurysms is crucial as they are the first recognized indicator of diabetic retinopathy. Studies have examined their incidence and regression in the early stages. Color fundus pictures and MA-tracker technology were used to calculate microaneurysm turnover. Results show stability over time, with only 29% of microaneurysms persisting in their initial site. Figure  8 illustrates the process of identifying microaneurysms in both normal individuals and patients with PDR. Microaneurysms were captured using the green channel of RGB fundus images, removing prominent elements like blood vessel trees and optic discs. Advanced image processing methodologies were employed to identify microaneurysm-containing regions within fundus images [ 60 ].

The findings of the identification of microaneurysms in individuals with normal retinal health and those diagnosed with PDR

Figure  8 shows the identification of microaneurysms, the initial signs of diabetic retinopathy. Microaneurysms appear as either circular or bead-like structures in the retinal vessels. the recognition step is composed of detecting them by the green channel of RGB fundus images and advanced image processing techniques for localization. The figure underlines microaneurysms in both normal and diabetic patients thus portraying their importance in disease diagnosis and treatment.

3.6.4 Hemorrhages

As the degree of DR intensifies, the presence of retinal hemorrhages becomes apparent. The authors propose augmenting ischemia, which is the inadequate oxygen supply, in the retinal tissue. As the quantity of retinal vessels rises, their susceptibility to injury and leakage escalates, leading to the exudation of fluid, lipids, and proteins. Figure  9 depicts the outcome of bleeding detection. The regions devoid of color in the image are indicative of hemorrhages. The detection of hemorrhages has two distinct components: (a) The identification of blood vessels, (b) the identification of blood vessels that exhibit hemorrhages. To get the image depicting hemorrhages, the image just displaying blood vessels was removed from the image, including both blood vessels and hemorrhages.

Results of hemorrhages detection for normal, PDR

Figure  9 displays the detection of retinal hemorrhages, which are visible in the disease progress. Hemorrhages are presented as colorless areas in the retinal tissue. Vessel segmentation is followed by defining regions inside those vessels having hemorrhages. This visualization aids in comprehending how hemorrhages are identified in retinal images, signifying the gravity of diabetic retinopathy and the necessity for prompt action.

4 Classification Process

Classification seems to be a challenge in disease prediction to determine a specific class using sample points. Labels and targets are other terms for the classes. Classification is the process of estimating the mapping function from data input to outputs [ 61 ]. There are two classification problems in medical imaging, binary and multi-classification. Binary classification has two target classes: cancer classification benign vs. malignant. Over two classes are included in the multi-classification process. For example, consider diabetic retinopathy categorization (normal, mild, moderate, severe, and PDR). The classification model needs sufficient training samples to determine how and why the input function corresponds to a target class. Labeled data from all four categories of diabetic retinopathy will be utilized as training data in diabetic retinopathy classification. After training, the classifier can be used to categorize unlabeled data. Classification is a supervised learning technique in which the output is pre-defined concerning the input data. To achieve classification, the DL classifier is employed as discussed below.

Medical images are high-quality images that retain specific information. Deep learning models are used to train these images, but they often suffer from the loss of contextual information. Techniques like support vector machines (SVM) and majority voting have shown efficacy in this context, but they also result in the loss of spatial data. Researchers have proposed techniques like context-aware learning and patch probability fusion to preserve contextual information. Context-aware learning combines four feature vectors from patch-wise methods but faces challenges in capturing spatial data constrained within certain boundaries. Patch probability fusion uses a patch-wise network to extract spatial characteristics, followed by an image-wise network for classification. However, this method lacks the preservation of distant contextual information. To address these issues, a transfer learning-based technique for classification was implemented on a CNN-based model to preserve contextual information of features. The methodology extracts and classifies several characteristics based on the derived feature set.Three pre-trained networks, namely AlexNet [ 62 ], VGGNet [ 63 ], and ResNet [ 64 ], were used for feature fusion as shown in Fig.  10 .

Classification process model diagram

The efficiency of deep learning models for medical image identification is well acknowledged on a global scale. Training deep learning models requires a larger quantity of data, which is often inadequate for the dataset of medical pictures. Training deep learning models from scratch is a time-consuming process. To address these challenges, the researchers used transfer learning as a methodology. Transfer learning involves using a model trained on a particular task dataset to address comparable problems, with the ability to make modest adjustments to the hyperparameters.

The models often used for image classification have shown satisfactory accuracy in the ILSVRC (ImageNet Large Scale Visual Recognition Challenges) [ 65 ]. The study investigates VGGNet and ResNet models, focusing on shallower versions like VGGNet-16 and ResNet-18, to mitigate overfitting issues in limited training data. The global average pooling layer is replaced with the first fully connected layer, and a single fully connected layer is retained after the GAP layer. The pre-existing models, namely AlexNet, VGGNet-16, and ResNet-18, are only used to extract features. Subsequently, a non-linear support vector machine (SVM) is employed to categorize cataracts. VGGNet [ 63 ] is a visual geometry group at Oxford University research initiative supervised by Zisserman and Simonyan (VGG). It came in second place in the ILSVRC-2014 Challenging Challenge. In contrast to the AlexNet architecture, which incorporates larger kernel-size filters \(11\times 11\) and \(5\times 5\) , the VGG16 model consists of 16 convolutional layers and primarily employs 3 \(\times \) 3 kernel-size filters. The VGG model has a substantial parameter count of 138 million, which presents challenges in training the network due to the process’s associated difficulties and time-consuming nature. Combining the two methodologies yields imprecision among the top five rankings, amounting to 6.8%. The model has several convolutional layers preceding the fully connected layers. Every convolutional layer is annotated with its kernel size, number, and stride thus showing the direction of the information flow through the network. Max-pooling layers are added for subsampling feature maps and fully connected layers handle classification tasks. Acquiring this architecture is pivotal for getting how the model adheres to retinal pictures to categorize diabetic retinopathy.Though deep learning algorithms can solve various classification issues, the fundamental issue with medical image classification is the lack of labeled data. Transfer learning is extensively used to overcome the lack of labeled data by reusing previously trained deep CNN for a similar job. As a result, it may be used to reduce training overhead and train with a smaller dataset. There are concerns about using transfer learning to overcome the lack of labeled data in DL-based medical images. Categorization. This study used pre-trained VGGNet-16 to identify using fixed size \(3 \times 3\) kernel filters in this experiment. Table  2 illustrates the architecture of the suggested model. The input layer is \(512\times 512\) , and the architecture has 16 levels. For all convolutional layers, save the first layer, which has a stride of 2, a \(3\times 3\) kernel size filter, unified bias, and stride of 1 are used. Each max-pooling layer in the network employs a \(2\times 2\) kernel size filter with a stride of 2. The resulting features obtained from the pooling operation are then flattened before being sent to the subsequent levels in the network. To mitigate the issue of overfitting, all layers within the model use the rectified linear unit (ReLU) activation function. Additionally, a dropout rate of 0.5 is implemented before the first two fully connected layers. After the convolution layers, two interconnected layers are introduced, each consisting of 1024 neurons. Finally, for early-stage detection and classification of diabetic retinopathy, a softmax function is utilized as a single neuron output layer. The process of fine-tuning the hyperparameters is used in the training of the VGGNet, which is based on transfer learning. The learning rate was 0.0001, and the model underwent training using the Adam optimization algorithm. The network weights are randomly initialized using a batch size of 32, followed by training the network for 300 epochs. The momentum magnitude is defined as 0.9, and the categorical cross-entropy is chosen as the objective function. Data augmentation techniques are used separately for each grade to solve the issue of data imbalance.

Table  2 provides a detailed breakdown of the architecture employed for the suggested VGGNet model. It enumerates the layers by type, kernel size, number, stride and output dimensions, respectively. This concise summary, in turn, enables us to understand the architecture of the VGGNet model and see how the information is processed through its layers.

5 Result Comparison

The proposed study aims to develop advanced models for multi-stage cataract grading and provide an efficient and timely diagnosis. The experimental results of the proposed methods indicate that these models have a strong capability to predict and grade early-stage cataracts with higher prediction performance. The proposed models are evaluated on a private dataset using different statistical measures, including F-score, specificity, precision, AUC, sensitivity, and accuracy. Applied preprocessing techniques such as image resizing, contrast-limited adaptive histogram equalization, top-bottom hat transformation, non-local mean denoising, green channel extraction, and image filtering to diminish noise and expand the retinal image quality. Proposed novel data augmentation operations like Gaussian scale-space-based theory and other data augmentation techniques on each cataract grade to address dataset unbalancing and annotated training dataset insufficiency issues. Proposed a novel ensemble technique of transfer learning and machine learning models for diagnosing cataracts, and to achieve better diagnostic performance and reduce computational time.

5.1 Configurations

The testing was conducted utilizing a Quadro K620 GPU with 8 GB of RAM, running on the Ubuntu 16 operating system. The Keras framework ( http://keras.io/ ) was used for the experiments. The dataset, including retinal data, was divided into two subsets: a training set and a testing set. A random selection process is used to allocate 80% of the pictures for training purposes, while the remaining 20% are reserved for testing as shown in Table 3 .

5.2 The Performance Metrics

To assess the efficacy of the proposed approach, an analysis was conducted on the performance of recognizing abnormal human behavior. This analysis included examining a confusion matrix and calculating several performance metrics afterward. The concept of precision refers to the level of accuracy or exactness in measurements or calculations. The positive predictive value (PPV) represents academic discourse accuracy by calculating the sum of true positives (TP) and false positives (FP) in a component tag. Precision can be measured as in Eq.  1 .

The F1 score is computed as the harmonic mean of recall and accuracy as in Eq.  2 .

Accuracy evaluates system performance by calculating the ratio of true positives (TP) and negatives (TN) to the total number of components, including TP, TN, FP, and FN. The accuracy may be determined using the equation presented in Eq.  3 .

Specificity measures the negative rate relative to the total components in the negative class, including both TN and FP instances. The mathematical expression is shown in Eq.  4 .

The calculation of the area under the curve (AUC) involves the execution of a specific integral across the interval defined by the two given points. The measurement of this quantity may be obtained by the use of the mathematical equation presented in Eq.  5 .

Different transfer learning models were implemented in addition to the suggested VGGNet technique, as indicated in Table  4 . The performance metrics of various deep learning architectures for diabetic retinopathy classification are compared by the given table. The metrics are sensitivity, specificity, accuracy, precision, F1-score, and area under the curve. Each row is distinct in terms of architecture, and the values in each column are the performance of that architecture for diabetic retinopathy classification. Studying this table helps in the comprehension of the degree of correctness of different models in the precise detection of diabetic retinopathy.

5.3 Result Analysis

The first pairplot which is an analysis of the dataset distribution in multi dimensions, that is the comparison of ‘Feature 1’ against ‘Feature 2’ is a fundamental exploration of the dataset. This scatter plot depicts the joint distribution of the two features, capturing the way they interrelate at different severity levels of diabetic retinopathy. The pair plot colors the data points based on the level of severity, adding another layer of information to aid in the identification of possible clusters and patterns that are associated with the different severity categories as shown in Fig.  11

The first pair plot compares ’Feature 1’ and ’Feature 2’ to examine the dataset’s simple matters. It shows the joint distribution of the two features and their interdependence over different disease severity grades of diabetic retinopathy. The color-coding on the plot helps identify possible clusters and patterns with different severity degrees

Each dot of the pair plot stands for one data point, the position of which is determined by the values of ’Feature 1’ and ’Feature 2’ as shown in Fig.  11 . The color of the dots is dependent on their severity level allowing analysts to distinguish the combinations of features which occur more frequently in particular severity categories. For example, if a particular part of the plot is made up mostly of red dots (representing serious cases), the corresponding feature attribute values may be indicative of severe diabetic retinopathy. Instead, if another region is occupied by green dots (representing mild cases) it means that a different pattern is identified with less severity levels. The other pairplot which is compared against itself ’Feature 1’ gives a more detailed view of the distribution of this feature with each severity level shown in Fig.  12 . This plot exhibits how ’Feature 1’ varies with the reference to both axes considering the severity level as the categorical variable. Through the visualization of the point distribution as well as their coloring by severity, the analysts can tell whether there exists a general pattern or outlier in some of the severity categories.

The second pair plot, visualizing ’Feature 1’ versus itself, shows the distribution of this feature among different severity grades. This plot shows how Feature 1 changes along both axes while severity level is the categoric variable. By visualizing the point distribution and color-coding them by severity, analysts can identify general patterns or outliers within each severity category

For instance, if ’Feature 1’ has widely varied values among the severe cases but less variability among mild cases, it implies that the feature is of more importance in the prediction of severe diabetic retinopathy as shown in Fig.  13 . Concerning this, if the distribution of ’Feature 1’ is almost the same across severity levels, it suggests that this feature is not so tightly related to the condition severity.

The third pair plot charts the relationship between the ’Feature 2’ and itself and it is similar to the exploration of the distribution of this feature across all the severity levels. Using this graphic, analysts can see any distinct patterns or correlations between ’Feature 2’ and retinopathy severity

The Fig.  13 shows third pair plot focusing on Feature 2 versus itself gives a similar exploration of the distribution of this feature across different severity levels. This plot enables analysts to observe any unique patterns or linkage between ’Feature 2’ and diabetic retinopathy severity. By analyzing the differences in the distribution of ‘Feature 2’ within each category of severity, the analysts can acquire clues about ‘Feature 2”s possible predictive power.

For example, in the severe cases of diabetic retinopathy, some of the ranges among ’Feature 2’ values are predominant; meaning the ’Feature 2’ is a reliable indicator of the level of severity of the disease. As a whole, these pairplots for a complete visual analysis of the dataset allowing analysts to discover critical information about the complex relationship between features and the severity of diabetic retinopathy. Using these visualizations researchers can better understand the underlying patterns in the data, pinpoint possible risk factors or predictive markers, and thus, ultimately improve decision-making and patient management in diabetic retinopathy. To mitigate computational expenses, this research uses four widely used pre-trained models, AlexNet, GoogLeNet, ResNet, and VGGNet. These models have been recognized for their exceptional learning capabilities. We hypothesized that the classification approach would achieve optimum performance by integrating the feature set derived from each CNN model. Transfer learning included transmitting the acquired learning data to the composite extracted characteristics after training the suggested model. The proposed model showed he maximum training and validation accuracy as shown in the Fig.  14

( a ) Shows the training and the validation accuracy and ( b ) shows loss of the proposed fine-tuned VGGNet framework

The results of the experiments show that the suggested architecture is more accurate than existing techniques. The suggested model on a balanced supplemented dataset is more accurate in identifying and diagnosing diabetic retinopathy. The training plots and validation accuracy of the proposed VGGNET model exhibit a gradual rise over time, as seen in Fig.  14 a, b. The training plots and validation accuracy of the proposed model Alex-Net exhibit an upward trend with time, as seen in Fig.  15 a, b. VGGNet’s fine-tuned version performs admirably. With each epoch that passes, the training and validation accuracy rate improves. The loss curve shows that the training and validation losses decrease with each epoch.

( a ) Depicted the training and the validation accuracy and ( b ) cross-entropy loss of the proposed Alexnet algorithm

The model under consideration achieved a 97.61% accuracy rate, making it a reliable tool for diabetic retinopathy identification and classification. Data were divided into training and testing datasets, with 20% of available data used for evaluation and 80% allocated for training purposes. Both the original and supplemented datasets were used to evaluate the suggested model as shown in Fig.  16 .

The proposed system proved immensely helpful and efficient for the early detection of diabetic retinopathy and its stages in patients belonging to remote areas.

The evaluation of the performance of the proposed model using various parameters is depicted graphically

The observations’ accuracy using the methods in the Table  6 above ranges from 82.3 to 85.3 to 90.4, 78.7 to 95.68 to 95.03%. The accuracy achieved by this approach, which is substantially greater than baseline methods, was 96.6%.

The mean accuracy values for ResNet, GoogleLeNet, and AlexNet were 92.40%, 93.75%, and 94.62%, respectively.

Still, there are some difficulties that we faced in implementing this system. The overall implementation of devices was a significant task as fundus cameras and classification systems were complex to handle compared to other devices customarily used. An expert was needed to capture a good-quality eye image for handling and using the fundus camera. The installation process was also quite complex as it must be handled carefully. Meanwhile, collaboration with expert doctors was also a difficult task. Some limitations are associated with proposed models irrespective of their diagnostic performance and effectiveness. The dataset utilized in this study is acquired from the local hospital in China, as the retina size may vary from one region to another; thus, collecting a diverse dataset covering all regions is the key constraint of this study. The adopted dataset is minimal compared to the natural images dataset. The proposed study is only focused on cataract detection and grading. The other overlapping diseases may also be considered. The limitation of deep learning-based fundus image analysis is the inherited computational complexity and requires GPUs to train the network. Research is an evolving area, and modern available techniques can be improved further with technological advancement. Various retinal datasets (either public or private) can be combined and annotated by professional ophthalmologists to evaluate the effectiveness of proposed models on diverse datasets. The experimental results obtained in this study may also be improved by combining various high and low-level features. The localization of other handcrafted features, along with deep learning-based extracted features, can further help improve accuracy. Since the growths of vessels are so essential in cataracts, using a stochastic approach may help estimate vessel growth. This detail may be included and used by medical practitioners to monitor the disease’s progression.

5.4 Sensitivity Analysis Comparison for Various Proposed DCNN-Based Architectures

Sensitivity analysis is a statistical method used in cataract diagnosis to assess the impact of uncertainties in input image data on detection models. It helps researchers identify significant features that affect the model’s performance and optimize parameters and configuration. Statistical measures like MAE, MSE, RMSE, sensitivity, area under the curve (AUC), and R-squared \((R^2)\) can quantify the prediction error between predicted values and target classes. This helps assess the model’s accuracy and overall performance, improving the reliability and effectiveness of the detection and classification process.

A VGGNet classifier achieved an MAE of 0.018, MSE of 0.013, RMSE of 0.117, and an R-squared score of 0.046. It also had an AUC of 0.0.981 and a sensitivity of 0.983, indicating that the classifier correctly identified all positive samples, as plotted in Fig.  13 . The CNN classifier performed well, achieving high AUC values and sensitivity. Table  5 presents the results of the prediction error obtained by applying DL-based algorithms.

The Alex-net classifier demonstrated strong classification performance on presented datasets, with an AUC value of 0.9449 and a sensitivity score of 0.949. Although marginally improved, the classifier’s AUC and sensitivity scores were lower than those of the VGGNet classifier. This suggests that the former may not possess the same level of efficacy in accurately categorizing data. Overall, the Alexnet classifier’s high AUC and sensitivity scores make it a potentially valuable instrument for categorizing cataracts.

5.5 Performance Comparison with Other Cutting-Edge Research

The efficacy of the suggested approach was evaluated by comparing its findings with those obtained from four established methods, therefore assessing its relative robustness. The average accuracy of the baseline approaches discussed in Table 6 , with values of 91%, 88%, 86%, and 95%, respectively. On the other hand, the approach suggested has a 97% accuracy rate, surpassing that of conventional methods. Furthermore, the findings of this study provide evidence that our approach exhibits superior performance in terms of accuracy compared to other methods. Table  6 presents the findings of a comprehensive analysis conducted to evaluate the robustness of the suggested design using five contemporary approaches.

The proposed method outperforms four well-known methods in terms of accuracy, compared to baseline methods achieving an average accuracy of 91%, 88%, 86%, and 95%, respectively, can be observed from Table  6 . The proposed method achieves 97% accuracy, surpassing mainstream methods and outperforming others in terms of accuracy. Table  6 shows the results of a comparative examination of the proposed design using five current methodologies to determine its strength.

Transfer learning techniques are often used in machine learning and are specifically tailored to a particular issue. These methods may be repurposed for other related problems by adjusting the hyperparameters via a process known as fine-tuning. The primary network is designated for transfer learning, whereby the pre-existing weights of the network are adapted. The initial weights of the network undergo continuous modifications to extract characteristics relevant to the job at hand. Recent research has shown that transfer learning techniques, namely those involving fine-tuning, have proven effective in addressing a range of medical imaging objectives, such as detecting DR and categorizing skin cancer. The present research used the four prevailing CNN-based architectures, AlexNet, GoogLeNet, VGGNet, and ResNet, to classify fundus images. Data processing, often in areas such as healthcare where the details are all-important is an exercise requiring full investigation and visualization. The enclosed code develops pairplots using Python libraries such as Seaborn, Pandas, and Matplotlib to investigate the features‘s relationship with the severity of diabetic retinopathy.

6 Conclusion

The proposed research aims to create a tool based on computer-based diagnosis and analysis to identify and categorize diabetic retinopathy levels. After a thorough review of the available literature, it was discovered that there were few studies in which CNN was employed, and strong results were produced. Even among the subset of individuals who had favorable outcomes, the approach proved to be computationally demanding, hence requiring the use of advanced computing equipment. Consequently, an approach based on CNN addresses the limitations seen in prior models. To remove extraneous information, many pre-processing processes were conducted. Based on the disease level of diabetic retinopathy, color fundus pictures of eyes were divided into four categories: moderate DR, No DR, Mild DR, and Severe DR. The suggested approach proved successful in detecting and segmenting the disease. The reports generated by the automatic detection system are then sent to the doctors attached to the system for further analysis, medications, or surgeries. The collected images represent the limitations of the method, as the classification accuracy is significantly affected by the quality of the images. There is still room for growth in the given technique. Due to second-grade variability, some individuals may already have moderate diabetic retinopathy at baseline, particularly for subtle abnormalities like microaneurysms. Although users should be aware that a high projected risk could suggest already-existing diabetic retinopathy, the deep-learning system’s prediction that these individuals will acquire diabetic retinopathy is semantically accurate. Future research may examine possible remedies, such as the simultaneous application of a diabetic retinopathy grading methodology using this risk classification tool. It may also be investigated if the deep-learning system may assist in predicting individuals who already have diabetic retinopathy, regardless of how severe it is or how it is graded on a scale of 1–10. The study found that diabetic retinopathy grades were based on one- or three-field color fundus images, making it difficult to detect macular edema. The lack of optical coherence tomography made it difficult to assess the progression to vision-threatening diabetic retinopathy. Patients were often referred for ophthalmology follow-up when moderate or severe diabetic retinopathy was found. A patient-level analysis, ideally in prospective situations, could help assess clinical significance. This system may be efficient for detecting diabetic retinopathy early in areas lacking good health facilities, but adaptation may be problematic due to the need for trained staff and system maintenance.

Availability of data and materials

This study used retina fundus images from the public dataset EyePACS, and can be accessed on Kaggle.com

Atteia, G., El-kenawy, E.-S.M., Samee, N.A., Jamjoom, M.M., Ibrahim, A., Abdelhamid, A.A., Azar, A.T., Khodadadi, N., Ghanem, R.A., Shams, M.Y.: Adaptive dynamic dipper throated optimization for feature selection in medical data. Comput. Mater. Continua 75 (1), 1883–1900 (2023)

Article   Google Scholar  

Kashif, J., et al.: Transfer learning-based model for diabetic retinopathy diagnosis using retinal images. Brain Sci. 12 (535), 1–12 (2022)

Google Scholar  

Iqbal, S., Qureshi, A.N., Ullah, A., Li, J., Mahmood, T.: Improving the robustness and quality of biomedical cnn models through adaptive hyperparameter tuning. Appl. Sci. 12 (22), 11870 (2022)

Hartnett, M., et al.: Perceived barriers to diabetic eye care: qualitative study of patients and physicians. Arch. Ophthalmol. 123 (15), 387–391 (2005)

Azar, A.T., Khan, Z.I., Amin, S.U., Fouad, K.M.: Hybrid global optimization algorithm for feature selection. Comput. Mater. Continua 74 , 2021–2037 (2023)

Iqbal, S., Qureshi, A.N., Li, J., Mahmood, T.: On the analyses of medical images using traditional machine learning techniques and convolutional neural networks. Arch. Comput. Methods Eng. 30 (5), 3173–3233 (2023)

Michael Liew, M., Bunce, C.: A comparison of the causes of blindness certifications in England and wales in working age adults (16–64 years). BMJ Open 4 (2), 1–20 (2014)

IIqbal, S., Qureshi, A.N., Li, J., Choudhry, I.A., Mahmood, T.: Dynamic learning for imbalance data in learning chest x-ray and ct images. Heliyon (2023)

Ali, S., Li, J., Pei, Y., Khurram, R., Rehman, K.U., Mahmood, T.: A comprehensive survey on brain tumor diagnosis using deep learning and emerging hybrid techniques with multi-modal mr image. Arch. Comput. Methods Eng. 29 (7), 4871–4896 (2022)

Hassan, S.A., Akbar, S., Rehman, A., Saba, T., Kolivand, H., Bahaj, S.A.: Recent developments in detection of central serous retinopathy through imaging and artificial intelligence techniques-a review. IEEE Access 9 , 168731–168748 (2021)

Sidrah, T.J., Tariq, K.: Frequency of diabetic retinopathy and factors for suboptimal diabetic control in type 2 diabetic patients in a trust hospital of pakistan. Pakistan J. Ophthalmol. 37 (2), 183–187 (2021)

Ramadan, R.A., Khedr, A.Y., Yadav, K., Alreshidi, E.J., Sharif, M.H., Azar, A.T., Kamberaj, H.: Convolution neural network based automatic localization of landmarks on lateral x-ray images. Multimed. Tools Appl. 81 (26), 37403–37415 (2022)

Rehman, K.U., Li, J., Pei, Y., Yasin, A., Ali, S., Mahmood, T.: Computer vision-based microcalcification detection in digital mammograms using fully connected depthwise separable convolutional neural network. Sensors 21 (14), 4854 (2021)

Jamal, A., Alkawaz, M.H., Rehman, A., Saba, T.: Retinal imaging analysis based on vessel detection. Microsc. Res. Tech. 80 (7), 799–811 (2017)

Saba, T., Akbar, S., Kolivand, H., Bahaj, S.A.: Automatic detection of papilledema through fundus retinal images using deep learning. Microsc. Res. Tech. 84 (12), 3066–3077 (2021)

Krishna, N.V., et al.: The communal system for early detection microaneurysm and diabetic retinopathy grading through color fundus images. Int. J. Sci. Eng. Technol. 2 (4), 228–232 (2013)

Jabbar, A., Naseem, S., Mahmood, T., Saba, T., Alamri, F.S., Rehman, A.: Brain tumor detection and multi-grade segmentation through hybrid caps-vggnet model. IEEE Access 11 (1), 72518–72536 (2023)

Suma, K., Saravana Kumar, V.: A quantitative analysis of histogram equalization-based methods on fundus images for diabetic retinopathy detection. In: Computational Intelligence and Big Data Analytics, pp. 55–63. Springer (2019)

Kornblith, S, Shlens, J., Le, Q.V.: Do better imagenet models transfer better? In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2019)

Jadhav, A.S. et al.: Computer-aided diabetic retinopathy diagnostic model using optimal thresholding merged with neural network (2020)

Walter, T., et al.: A contribution of image processing to the diagnosis of diabetic retinopathy-detection of exudates in color fundus images of the human retina. IEEE Trans. Med. Imaging 21 (10), 1236–1243 (2002)

Mahmood, T., Li, J., Pei, Y., Akhtar, F., Imran, A., Rehman, K.U.: A brief survey on breast cancer diagnostic with deep learning schemes using multi-image modalities. IEEE Access 8 , 165779–165809 (2020)

Al-Badri, A.H., Ismail, N.A., Al-Dulaimi, K., Rehman, A., Abunadi, I., Bahaj, S.A.: Hybrid cnn model for classification of rumex obtusifolius in grassland. IEEE Access 10 , 90940–90957 (2022)

Kimar, D., Taylor, G., Wong, A.: Discoveryradiomics with clear-dr: interpretable computer aided diagnosis of diabetic retinopathy. IEEE Access 7 , 25891–25896 (2019)

Eltanboly, A., et al.: A computer-aided diagnostic system for detecting diabetic retinopathy in optical coherence tomography images. Med. Phys. 44 (3), 914–923 (2017)

Abunadi, I.: Deep and hybrid learning of mri diagnosis for early detection of the progression stages in Alzheimer’s disease. Connect. Sci. 34 (1), 2395–2430 (2022)

Ali, S., Shaukat, Z., Azeem, M., Sakhawat, Z., Mahmood, T., ur Rehman, K.: An efficient and improved scheme for handwritten digit recognition based on convolutional neural network. SN Appl. Sci. 1 , 1–9 (2019)

Skouta, A., Elmoufidi, A., Jai-Andaloussi, S., Ouchetto, O.: Deep learning for diabetic retinopathy assessments: a literature review. Multimed. Tools Appl. 82 (27), 41701–66 (2023)

Sabeena, A.S., Jeyakumar, M.K.: Gd-stfa: gradient descent sea turtle foraging algorithm enabled deep q network for diabetic retinopathy detection. Multimed. Tools Appl. 1–20 (2023)

Ramachandran, S.K., Manikandan, P.: An efficient alo-based ensemble classification algorithm for medical big data processing. Int. J. Med. Eng. Inform. 13 (1), 54–63 (2021)

Das, S., Saha, S.K.: Diabetic retinopathy detection and classification using cnn tuned by genetic algorithm. Multimed. Tools Appl. 81 (6), 8007–8020 (2022)

Khanna, M., Singh, L.K., Thawkar, S., Goyal, M.: Deep learning based computer-aided automatic prediction and grading system for diabetic retinopathy. Multimed. Tools Appl. 82 (25), 39255–302 (2023)

Rachapudi, V., Rao, K.S., Rao, T.S., Dileep, P., Deepika Roy, T.L.: Diabetic retinopathy detection by optimized deep learning model. Multimed. Tools Appl. 82 (18), 27949–71 (2023)

Minija, S.J., Rejula, M.A., Ross, B.S.: Automated detection of diabetic retinopathy using optimized convolutional neural network. Multimed. Tools Appl. 83 (7), 21065–80 (2023)

Saranya, P., Prabakaran, S.: Automatic detection of non-proliferative diabetic retinopathy in retinal fundus images using convolution neural network. J. Ambient Intell. Humaniz. Comput. 1–10 (2020)

Saranya, P., Umamaheswari, K.M.: Detection of exudates from retinal images for non-proliferative diabetic retinopathy detection using deep learning model. Multimed. Tools Appl. 1–21 (2023)

Erciyas, A., Barişçi, N.: A meta-analysis on diabetic retinopathy and deep learning applications. Multimed. Tools Appl. 1–20 (2023)

Al-Karawi, A., Avşar, E.: A deep learning framework with edge computing for severity level detection of diabetic retinopathy. Multimed. Tools Appl. 1–22 (2023)

Thomas, N.M., Jerome, S.A.: Diabetic retinopathy detection using eadbsc and improved dilated ensemble cnn-based classification. Multimed. Tools Appl. 1–23 (2023)

Kale, Y., Sharma, S.: Detection of five severity levels of diabetic retinopathy using ensemble deep learning model. Multimed. Tools Appl. 82 (12), 19005–19020 (2023)

Ohri, K., Kumar, M.: Domain and label efficient approach for diabetic retinopathy severity detection. Multimed. Tools Appl. 1–30 (2023)

Kwasigroch, A., Jarzembinski, B., Grochowski, M.: Deep cnn based decision support system for detection and assessing the stage of diabetic retinopathy. In: 2018 International Interdisciplinary PhD Workshop (IIPhDW), pp. 111–116 (2018)

Dehghani, A., Moin, M.-S., Saghafi, M.: Localization of the optic disc center in retinal images based on the harris corner detector. Biomed. Eng. Lett. 2 (3), 198–206 (2012)

Dubey, S., Dixit, M.: Recent developments on computer aided systems for diagnosis of diabetic retinopathy: a review. Multimed. Tools Appl. 82 (10), 14471–14525 (2023)

Parthiban, K., Kamarasan, M.: Diabetic retinopathy detection and grading of retinal fundus images using coyote optimization algorithm with deep learning. Multimed. Tools Appl. 82 (12), 18947–18966 (2023)

Pratt, H., et al.: Convolutional neural networks for diabetic retinopathy. Proc. Comput. Sci. 90 , 200–205 (2016)

Belderrar, A., Hazzab, A.: Real-time estimation of hospital discharge using fuzzy radial basis function network and electronic health record data. Int. J. Med. Eng. Inform. 13 (1), 75–83 (2021)

Mansour, R.: Deep learning based automatic computer-aided diagnosis system for diabetic retinopathy. Biomed. Eng. Lett. 8 (1), 41–57 (2018)

Hasan, M., Uguz, H.: A new early-stage diabetic retinopathy diagnosis model using deep convolutional neural networks and principal component analysis. Traitement du Signal 37 (5), 711–722 (2020)

Gupta, S., Thakur, S., Gupta, A.: Comparative study of different machine learning models for automatic diabetic retinopathy detection using fundus image. Multimed. Tools Appl. 1–32 (2023)

Belay, H., Ayalew, S., Hailu, D.: Hydrological performance evaluation of multiple satellite precipitation products in the upper blue nile basin. J. Hydrol. Reg. Stud. 27 , 1–34 (2020)

Oacgade, S., et al.: Retinal fundus multi-disease image dataset: a dataset for multi-disease detection research. Data 6 (14), 1–14 (2021)

Jinfeng, G., et al.: Ensemble framework of deep cnns for diabetic retinopathy detection. Comput. Intell. Neurosci. 1–10 , 2020 (2020)

Wang, Y., et al.: Automatic segmentation of optic disc and cup in retinal fundus images using improved two-layer level set method. Math. Probl. Eng. 1–15 , 2019 (2019)

MathSciNet   Google Scholar  

Mahmood, T., Li, J., Pei, Y., Akhtar, F., Jia, Y., Khand, Z.H.: Breast mass detection and classification using deep convolutional neural networks for radiologist diagnosis assistance. In: 2021 IEEE 45th Annual Computers, Software, and Applications Conference (COMPSAC), pp. 1918–1923. IEEE (2021)

Alyoubi, L., Maysoon, F., Wafaa, M.: Diabetic retinopathy fundus image classification and lesions localization system using deep learning. Sensors 21 (3704), 1–22 (2021)

Mahmood, T., Li, J., Pei, Y., Akhtar, F., Rehman, M.U., Wasti, S.H.: Breast lesions classifications of mammographic images using a deep convolutional neural network-based approach. PLoS One 17 (1), e0263126 (2022)

Sujith, S., Vipula, S.: Automatic detection of diabetic retinopathy in nondilated rgb retinal fundus images. Int. J. Comput. Appl. 47 (19), 1–7 (2012)

Imran, Q., Jjun, M., Qaisar, A.: Recent development on detection methods for the diagnosis of diabetic retinopathy. Symmetry 11 (749), 1–34 (2019)

Shengchun, L., et al.: Microaneurysms detection in color fundus images using machine learning based on directional local contrast. BioMed. Eng. Online 9 (21), 1–25 (2020)

Alfian, G., et al.: Deep neural network for predicting diabetic retinopathy from risk factors. Mathematics 8 (1620), 1–19 (2020)

Krizhevsky, A., Sutskever, I., Hinton, G.E.: Imagenet classification with deep convolutional neural networks. Adv. Neural. Inf. Process. Syst. 25 (1), 1097–1105 (2012)

Simonyan, K., Zisserman, A.: Very deep convolutional networks for large-scale image recognition (2014). arXiv:1409.1556

He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learning for image recognition. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 770–778 (2016)

Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., Fei-Fei, L.: Imagenet: a large scale hierarchical image database. In: 2009 IEEE Conference on Computer Vision and Pattern Recognition, pp. 248–255 (2009)

Nanni, L., Ghidoni, S., Brahnam, S.: Handcrafted vs non-handcrafted features for computer vision classification. Pattern Recogn. 71 , 158–172 (2017)

Gargeya, R., Leng, T.: Automated identification of diabetic retinopathy using deep learning. Ophthalmology 124 , 962–969 (2017)

Zeng, X., Hao Chen, Y., Luo, W.Y.: Automated diabetic retinopathy detection based on binocular siamese-like convolutional neural network. IEEE Access 7 , 30744–30753 (2019)

Li, X., Ding, Q., Sun, J.-Q.: Remaining useful life estimation in prognostics using deep convolution neural networks. Reliab. Eng. Syst. Saf. 172 , 1–11 (2018)

Zhang, W., Zhong, J., Yang, S., Gao, Z., Jing, H., Chen, Yu., Yi, Z.: Automated identification and grading system of diabetic retinopathy using deep neural networks. Knowl.-Based Syst. 175 , 12–25 (2019)

Lin, G.M., Chen, M.J., Yeh, C.H., Lin, Y.Y., Kuo, H.Y., Lin, M.H., Chen, M.C., Lin, S.D., Gao, Y., Ran, A., Cheung, C.Y.: Transforming retinal photographs to entropy images in deep learning to improve automated detection for diabetic retinopathy. J. Ophthalmol. (2018)

Download references

Acknowledgements

The authors would like to thank China’s National Key R &D Program for providing the experimental facilities to perform these experiments. The author would also like to thank the Artificial Intelligence and Data Analytics (AIDA) Lab, CCIS, Prince Sultan University Riyadh, Saudi Arabia, for support. The authors are thankful for the support.

This study is supported by the National Key R &D Program of China with project no. 2020YFB2104402.

Author information

Authors and affiliations.

Faculty of Information Technology, Beijing University of Technology, 100124, Beijing, China

Ayesha Jabbar & Jianqiang Li

College of Electronics and Information Engineering, Shenzhen University, Shenzhen, China

Ayesha Jabbar & Kashif Jabbar

Faculty of Information Sciences, University of Education, Lahore, 54000, Pakistan

Shahid Naseem

Beijing Engineering Research Center for IoT Software and Systems, Beijing, 100124, China

Jianqiang Li

Artificial Intelligence and Data Analytics (AIDA) Lab, CCIS Prince Sultan University, Riyadh, 11586, Saudi Arabia

Tariq Mahmood, Amjad Rehman & Tanzila Saba

Faculty of Information Sciences, University of Education, Vehari Campus, Vehari, 61161, Pakistan

Tariq Mahmood

You can also search for this author in PubMed   Google Scholar

Contributions

Conceptualization, Ayesha, Shahid, Tariq and Jianqiang; data curation, Ayesha, Tanzila, and Amjad; formal analysis, Shahid, Tariq, Amjad, Kashif and Tanzila; Funding acquisition, Jianqiang.; methodology, Ayesha, Shahid, Tariq, Kashif, Jianqiang and Amjad; software, Tariq, Amjad, and Tanzila; Supervision, Shahid and Tariq.

Corresponding author

Correspondence to Tariq Mahmood .

Ethics declarations

Conflict of interest.

The authors declare that they have no conflict of interest.

Use of AI tools declaration

The authors declare they have not used artificial intelligence (AI) tools in the creation of this article.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Jabbar, A., Naseem, S., Li, J. et al. Deep Transfer Learning-Based Automated Diabetic Retinopathy Detection Using Retinal Fundus Images in Remote Areas. Int J Comput Intell Syst 17 , 135 (2024). https://doi.org/10.1007/s44196-024-00520-w

Download citation

Received : 13 November 2023

Accepted : 07 May 2024

Published : 29 May 2024

DOI : https://doi.org/10.1007/s44196-024-00520-w

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Retinopathy
• Micro-aneurysms
• Retinal fundus
• Health issues
• Find a journal
• Publish with us
• Track your research

• Open access
• Published: 03 March 2023

The experience of diabetic retinopathy patients during hospital-to-home full-cycle care: A qualitative study

• Mengyue Zhang 1 ,
• ChunHua Zhang 1 ,
• Chen Chen 1 ,
• Linjie Liu 2 ,
• Youping Liang 2 ,
• YiRong Hong 2 ,
• Yanyan Chen 2 &
• Yinghui Shi 2  

BMC Nursing volume  22 , Article number:  58 ( 2023 ) Cite this article

2343 Accesses

1 Citations

1 Altmetric

Metrics details

Diabetic retinopathy (DR) is one of the major blinding eye diseases worldwide. Psychological, emotional and social problems of DR patients are prominent. The aim of this study is to explore the experiences of patients with different phases of DR from hospital to home based on the “Timing It Right” framework, and to provide a reference for formulating corresponding intervention strategies.

The phenomenological method and semi-structured interviews were used in this study. A total of 40 patients with DR in different phases were recruited from a tertiary eye hospital between April and August 2022. Colaizzi’s analysis method was used to analyse the interview data.

Based on the “Timing It Right” framework, different experiences in five phases of DR before and after Pars Plana Vitrectomy (PPV) were extracted. The patients experienced complicated emotional reactions and inadequate coping skills during the pre-surgery phase, increased uncertainty during the post-surgery phase, insufficient confidence and the decision to change during the discharge preparation phase, eagerness for professional support and moving forward in exploration during the discharge adjustment phase, and courageous acceptance and positive integration during the discharge adaptation phase.

The experiences of DR patients with vitrectomy in different phases of disease are ever-changing, and medical staff should provide personalized support and guidance to help DR patients get through the hard times smoothly and enhance the quality of hospital-family holistic care.

Peer Review reports

Introduction

Diabetic retinopathy (DR) is one of the most common microvascular complications among diabetics and the main leading cause of blindness in adults [ 1 ]. Proliferative diabetic retinopathy (PDR) is one of the most visually impairing complications of DR [ 2 ]. According to the latest data, 463 million adults currently have diabetes worldwide, with China ranking first [ 3 ]. Approximately 60% of diabetics will develop PDR [ 4 ]. The global prevalence of DR is 34.6% among diabetes patients and more than 50% of visual impairment or blindness cases caused by DR come from the Asia-Pacific region [ 5 , 6 ]. These sobering statistics indicate that DM has emerged as a major global health problem and it is urgent to step up efforts to combat it.

Regular screening and intensive glucose management can effectively delay the progression of DR [ 5 , 7 ]. However, diabetics in China often have poor awareness of ophthalmic screening and poor self-management ability. Only 39.7% of diabetics have good glycaemic control and meet the recommended target levels [ 8 ], and up to 67% of diabetics have PDR at the time of their first ophthalmology visit [ 9 ]. The present study deeply explored the feeling of illness and experience of patients with DR using a qualitative research methodology; thus, providing a basis for intervention studies.

Vitrectomy is one of the mainstays of DR treatment. However, patients are only able to recover or retain partial useful vision after surgery, and most patients continue to have psychological, emotional, and social problems [ 2 , 10 ]. Meanwhile, visual impairment due to DR has a huge impact on quality of life and can cause concern and emotional distress by limiting mobility, activity, and socialization [ 11 , 12 ]. According to statistics, almost 25% and 13.5% of patients with DR show symptoms of depression and anxiety, respectively, in China [ 13 ]. The psychological, emotional, and social problems are more prominent in patients with PDR [ 14 ], which may reduce compliance, reduce the level of blood glucose management, and accelerate disease progression in patients with DR [ 15 ]. Therefore, the psychosocial situation of DR patients requires attention to meet their relevant needs.

Until now, many studies have focused on interventions for patients with DR and their effectiveness [ 16 , 17 ]. Some studies explored the disease-related experiences of DR patients [ 18 , 19 ]. However, little is known about the full cycle of experiences of DR patients treated by vitrectomy during the hospital-to-home in China. Cameron and Gignac (2008) proposed the theory of “Timing It Right” (TIR) [ 20 ], which divided the disease process into five stages: (1) event/diagnosis, (2) stabilization, (3) preparation, (4) implementation and (5) adaptation. The first two phases occur during acute care, the third occurs during acute care and/or inpatient rehabilitation, and the final two phases occur in the hospital-family community. Each stage focuses on information, emotion, tools, and assessment needs, and further emphasizes that patients’ care needs vary with time. Using the TIR theoretical framework and a qualitative approach allow for a better understanding of the experiences and feelings of DR patients from their perspective during disease different stages, thus enabling the needs of DR patients to be more fully responded to.

Therefore, this study aims to explore the experiences and feelings of DR patients during different disease treatment stages through in-depth interviews based on the theory of “Timing It Right”. This is helpful to provide a reference for formulating effective and sustainable hospital-family holistic care intervention programs.

This study was part of a larger research project to develop a hospital-family holistic caring intervention program for patients with DR and discuss the effect of its application on DR patients through a quasi-experimental study.

Primary objective

The primary objective of this qualitative study was to explore the experiences and feelings of DR patients at different phases and understand their relevant needs.

Secondary objective

To provide a reference for formulating the intervention program that meets the needs of these patients.

This qualitative design was adopted using the Descriptive Phenomenology approach. Descriptive phenomenology emphasizes a process of “returning to the thing itself” and gives more attention to the life experience of patients, which helps to explore the experiences and feelings of patients[ 21 ]. Since the study focused on exploring the experiences and feelings of patients with DR in different phases, the descriptive phenomenology research methodology was appropriate.

Participants

Patients with DR were voluntarily recruited from the Fundus Surgical Department, Eye Hospital of Wenzhou Medical University, Zhejiang, China. Purposeful sampling combined with a maximum variation sampling approach was utilized. Participants with different sociodemographic characteristics were chosen when possible.

Recruitment of participants continued until data saturation was reached [ 22 ], which occurred with the 7th, 12th, 12th, 11th and 10th participants for each phase respectively; two researchers agreed no new themes were identified from the interview data. The inclusion criteria were (1) met the relevant diagnostic criteria for DR in the Clinical Diagnosis and Treatment Guidelines for DR in China [ 23 ], (2) underwent elective vitrectomy, (3) age ≥ 18 years, (4) had normal verbal communication skills and (5) voluntary participation in this study. Exclusion criteria were (1) other eye diseases or a history of ocular trauma, (2) severe cardiovascular disease or serious organic disease, (3) cognitive impairment, mental disorder, or not fully capable of acting and (4) being receiving or having received care interventions for other chronic diseases.

Based on the TIR framework, combined with the characteristics of DR and vitrectomy, and consulting with specialists in the Fundus Surgical Department, Endocrinology Department and other related fields, five phases were designated: pre-surgery, post-surgery, discharge preparation, discharge adjustment, and discharge adaptation. The pre-surgical period was from the time the DR patient decided to undergo vitrectomy to the time of surgery and this period usually lasted 3–7 d. The post-surgery phase was the period between the patient’s surgery and post-surgery stabilization. This period usually lasted 1–2 d. The discharge preparation phase was from the time the patient achieved medical criteria for hospital discharge to the time of discharge, and usually lasted 1 d. The discharge adjustment phase was the period from the patient’s discharge to home until 3 months after the vitrectomy. The discharge adaptation phase was the period 3 months to 6 months after the vitrectomy. According to the short hospitalization period and quick turnaround of DR patients undergoing vitrectomy, the post-surgery and discharge preparation phases were the same patient. There was no duplication in the other phases of participants. Each phase was denoted as A, B, C, and D. The staging criterion was based on the most diseased eye.

Data collection

A semi-structured interview guide (Table  1 ) was used to collect information through face-to-face interviews. This interview guide was developed by the physicians, nurses, and patients and was based on a literature review and project team discussions. The final interview guide was revised after consultation with qualitative nursing experts and fundus surgical specialists, and pre-interviews with three patients. All the interviews were conducted by a postgraduate nursing student who was trained in qualitative research. A research assistant played an auxiliary role which included recording the interviews.

Patients who were interested in participating in the study and met the inclusion criteria were informed of the purpose and significance of the study and signed an informed consent form. The interview consisted of five phases. The first interview phase occurred during the pre-hospitalization period and was conducted in the outpatient waiting room. The second and third phases were conducted during the inpatient period, the interview location was a quiet place such as a ward or duty room, and the interview time was during non-treatment and patient rest periods. Patients had returned to the community during the latter two phases and when convenient the latter two interviews were conducted in the outpatient waiting room at the time of patient review or were recorded by video. Each person was interviewed one time, and each interview lasted approximately 20–40 min. The entire interview was recorded, and the researcher listened carefully and recorded the interviewees’ expressions, movements, and emotional reactions.

Ethical consideration

Ethical approval for this study was granted by the ethics committee of the hospital (approval number: 2022-045-K-30-01). Informed consent was provided and obtained from all participants before the study commenced. To protect the privacy of interviewees, the interviews were presented anonymously, and names were replaced with letters.

Data analysis

The recordings were transcribed word by word within 24 h after the interview. Colaizzi’s seven-step method was used for data analysis [ 24 ]: (1) reading all interview materials carefully, (2) extracting and labelling meaningful statements, (3) coding meaningful statements preliminarily, (4) classifying the codes into themes and subthemes, (5) merging the formed themes with the research content and describing them in detail, (6) discussing the structural framework of the experience and the feelings of the patients with DR at different phases and (7) returning the theme to the interviewee for confirmation. Two female researchers independently analysed and coded the original data. In case of disagreements, the group discussed and reached a consensus. The data were analysed using NVivo 12.0.

The trustworthiness of this qualitative study was ensured by maintaining the credibility, dependability, confirmability, and transferability of the data [ 25 , 26 ]. The study interviewer was a master’s degree student in nursing. The interviewer received systematic qualitative training to master qualitative research methods, was experienced in eye hospital practices, and established a good relationship with the patients before the interviews commenced. This facilitated the acquisition of real information. The researcher maintained a neutral attitude during the interview, did not lead or hint, did not interrupt the interviewee at will, and only asked timely follow-up questions, rhetorical questions, and clarifications until no new information emerged. Therefore, credibility was ensured. The collection, analysis, and interpretation of data were continually reviewed and detailed to ensure its dependability. The data extracted from the survey results were described in detail to achieve confirmability. Regarding transferability, this study described in detail the inclusion criteria, exclusion criteria, and demographic characteristics involved. Simultaneously, the Consolidated Criteria for Reporting Qualitative Research (COREQ) checklist was used to report the findings [ 27 ] (See Appendix I for details).

A total of 40 respondents were enrolled, and the number of respondents at each phase was 7, 12, 11 and 10, including 25 males and 15 females. The average duration of diabetes was 13 years. The general characteristics of the study respondents are presented in Table  2 . The themes in the interviews are presented in Table  3 .

The pre-surgery phase: complicated emotional reactions and inadequate coping skills

Fear and worry.

In DR fundus haemorrhage, the patient’s sudden blurring of vision and the floating of black shadows in front of the eyes cause the patient to experience fear, anxiety, and other negative emotions.

A2: ‘One day I woke up and my eyes suddenly went blind. I didn’t realize the haemorrhage was so severe.’

A5: ‘When I got up to wash my face, I found that the cobwebs inside my eyes had fallen out. I was scared to death.’

Vision loss caused many inconveniences in the working lives of patients with DR, such as difficulties driving, walking, using WeChat, injecting insulin, etc. This intensified patient anxiety.

A3: ‘Last time I almost hit the old lady, so I am afraid to walk alone now.’

A6: ‘I can’t see with my eyes. I can’t do anything.’

Regret and self-blame

In the early stages of diabetes, many patients do not know or believe that diabetes can cause diabetic retinopathy, diabetic foot and other related complications, and do not realize the importance of diabetes glucose management. When vision is impaired, patients with DR begin to regret their previous behaviour.

A3: ‘I didn’t know regret until something went wrong with my eyes. I regret that I didn’t look at my eyes earlier.’

A6: ‘It may not be so serious if it is controlled at the beginning.’

A7: ‘Why didn’t I control it before? Why was my mouth so greedy? Why was my self-control so poor?’

Shadow and daunt

Some patients with DR have undergone multiple panretinal photocoagulation (PRP) and anti-vascular endothelial growth factor (anti-VEGF) treatments before being treated with vitrectomy.

A1: ‘The two eyes have already been injected several times, two or three times, and the PRP treatment has been done several times.’

The repeated PRP treatments and the pain associated with the PRP usually cast a psychological shadow on DR patients, making them fearful of subsequent treatments.

A2: ‘When I had PRP treatment before, I felt a little scared. The cloth came up in layers and just showed you an eye.’

A5: ‘I’m definitely not doing PRP treatment. The eye is like being gouged out. It’s more painful than giving birth. I give up on the left eye, I’m still left with that one after this one goes blind.’

The post-surgery phase: the increased uncertainty

Physical discomfort after surgery.

The treatment of choice for DR is usually vitrectomy followed by insertion of either a gas or silicone oil tamponade. The procedure utilises local anaesthesia. After the procedure, the anaesthetic effect wears off and patients complain of eye pain.

B5: ‘I want to lower my head a bit, my eyes are not very comfortable, my eyes are swollen and painful. The whole procedure is also very painful. I feel like my eye is bursting open.’

B8: ‘The whole eye is going to explode, now the pressure of the eye is so heavy, as if a mountain is pressed, the eye cannot open.’

Surgical trauma and elevated intraocular pressure (IOP) are the main factors causing postoperative pain. Additionally, if gas or silicone oil is filled in the eye, the patient usually needs to maintain the head-down and side-lying positions alternately after surgery. A prolonged prone position will cause pain in the patient’s head, chest, abdomen and extremities, compress the eye orbit, affect blood circulation and aggravate eye swelling [ 28 ].

B10: ‘I never thought I would be like this after the surgery. I couldn’t eat, I want to vomit, and my blood pressure is still high. So, I am in a bad mood. I am told to sit during the day and lie on my side at night, I am so tired. My eyes are swollen and cannot be opened.’

Uncertainty about disease prognosis

Due to the filling of gas and silicone oil, DR patients will not have a significant change in vision immediately after surgery compared to pre-surgery. Meanwhile, those who have had silicone oil injected need a second surgery to remove the silicone oil. The uncertainty of the time of the second surgery and the uncertainty of the recovery of vision will increase the patient’s uncertainty about their disease prognosis.

B1: ‘I don’t open my eye. I worry whether it can open. I always expect my eyes to recover better… I’m afraid that eye still cannot see after the surgery.’

B5: ‘I don’t know what the condition of my retina is, do you have the report card from the surgery? What is the condition of my eyes now? Are 800 points of laser considered too much? I am worried that I will see less after the surgery.’

B11: ‘Do I get the PRP treatment next or the anti-VEGF treatment? How can I best maintain my vision?’

Heavy family burden

The course of the disease with DR is long and the condition often recurs. Patients mostly need a combination of PRP, anti-VEGF and vitrectomy treatments. Treatment requires tremendous energy and financial resources, which puts heavy care and financial burdens on patients and their families.

B3: ‘Last year the doctor said it would take 20,000. I only had more than 10,000 in my pocket, so I walked away. This year the eye is really no good, my brothers and sisters lent a little to me to do surgery.’

B12: ‘The doctor told me to do the other eye as well. I’m not blind. I’m not doing it for now. This surgery costs more than 20,000 yuan, and my salary is only 1000 yuan a month.’

In addition, the social roles of DR patients at this age are more complex, as they are both sons and daughters and parents. As sons and daughters, they need to support the elderly; as parents, they must raise their children. The multiple roles make their caregiving burden heavy and many neglect themselves.

B3: ‘My two couples earn 100,000 yuan in school, the children’s school needs 50,000 yuan, the family also needs to spend money, not much money left in a year.’

B9: ‘Our generation is in the situation of “the elderly above and children below”, we are very tired, so we do not have time to pay attention to our own health. It is too late once we get sick.’

The discharge preparation phase: the insufficiency of confidence and the decision to change

Lack of self-care confidence.

Most patients with DR have a short hospitalization period. They lack sufficient knowledge about eye care, handling postoperative complications and so on. As the time of discharge approaches, patients become increasingly worried that they will not be able to take good care of themselves and then develop a sense of helplessness.

B7: ‘I still don’t know how to protect my eyes, for example, I want to eat melon seeds and peas, but I don’t know if I can eat them. Are these peas bad for the eyes? I’m afraid that if I don’t take good care of my eyes, they will bleed again.’

B10: ‘Today the doctor said that the surgery was successful, and that post-operative infection should be prevented. But my body is too sick, my body is not immune, my immunity is not as strong as others, and I am afraid of infection when I go home.’

B12: ‘The hospital in our town does not provide ophthalmology services, what should we do if we have problems after discharge? Can we add a WeChat? We can consult when we encounter problems.’

Deciding to make a change for the eyes

After the patients experienced the inconvenience of blurred vision in their work lives and also had a more comprehensive understanding of the cause and treatment of DR, more than half (10/12) of the interviewees said they realized the importance of blood glucose management and decided to make changes.

B6: ‘Before I thought that as long as the hospital was there, I would not be afraid, and I would be able to come to the hospital for treatment. After this surgery, I realize that my previous knowledge of diabetes was very inadequate, and I will put it in my mind in the future.’

B9: ‘I will definitely manage my blood glucose in the future. It comes out of my eyes and directly affects my work, and I’ve got it in mind.’

B11: ‘We must keep our blood glucose under control. If we don’t control it, we may have big problems with our eyes again later. Whether it is high blood glucose or high blood pressure, it may be harmful to the eyes.’

The discharge adjustment phase: eager for professional support and moving forward in exploration

Eager for professional support.

DR patients treated with vitrectomy have a long recovery after surgery, while most patients have a short hospitalization period. An inflammatory response such as conjunctival hyperaemia and corneal oedema still occurs within a short period after discharge. Combined with inadequate self-care ability among DR patients, they urgently want to obtain home support services.

C4: ‘I want to listen to online lectures because sometimes there is something uncomfortable in my eyes and I can learn what causes it. …The consultation channels I also need, otherwise, I am panicking when I have a problem. I want someone who knows more about it to help me.’

C11: ‘I felt fine for the first two days after surgery, then I didn’t know what caused my eyes to get red and swollen, and I really wanted to ask the doctor, but I didn’t know who to ask, and I didn’t have the hospital’s phone number, so I stopped the medication.’

C9: ‘I often ask my doctor in charge. We have added WeChat. I ask him any question and he answers me.’

Trouble with low vision and impaired mental health

DR patients whose eyes are filled with silicone oil or gas after surgery still have poor vision. It can make the patient puzzled.

C2: ‘After the surgery on this eye, I still couldn’t see clearly, so I wondered why it was the same before and after the surgery.’

C7: ‘My son and my husband say that I can’t see because of the silicone oil, but I don’t know if that’s the reason.’

At the same time, patients who had high expectations of the surgery before the operation and whose vision recovery was not satisfactory after the surgery will feel a greater sense of psychological disparity. They gradually lose their confidence, and their mental health is impaired.

C6: ‘What is the meaning of my life when I can’t see with my eyes.’

C7: ‘I’m so annoyed that the treatment cost so much money and I can’t see so well. I’m really annoyed.’

C8: ‘Even if the silicone oil is removed, my eyes may still be blind…the vision recovery is so different from what the doctor said before the surgery. I can’t accept it at all.’

Patients did not know how to relieve the pain of impaired vision. However, they did not want to impose it on their family either. Some patients balanced their inner helplessness by complaining about their fate.

C6: ‘I don’t know what to do. I don’t know where the eye problem came from. Why God treated me this way.’

Making changes and moving forward in exploration

Most of the respondents mentioned the importance of blood glucose management, and the vision change made them more alert. They started to explore the experience of eye protection and blood glucose management for themselves.

C5: ‘I also bought a glucometer and an automated sphygmomanometer, and now I’m eating more regularly, and my eating habits have slowly adjusted.’

C8: ‘Since the surgery, I’ve been drinking less, and I’ve taken care of myself. I usually sleep at 11 pm, and I don’t stay up until 2 or 3 am.’

C10: ‘Do not smoke, do not touch the kitchen fumes. Cannot eat spicy and stimulating food, we must protect the eyes after surgery, do not let the sweat flow into the eyes in summer.’

The discharge adaptation phase: courageous acceptance and the positive integration

Compromise acceptance and positive transformation.

The effect of postoperative vision recovery in patients with DR is different. After the second surgery, some DR patients continue to have low vision, and the distress of low vision makes them reach their lowest point emotionally, showing a loss of self-worth and self-denial.

D1: ‘I can’t read, I can’t write, and I can’t enjoy the scenery, so I have no interest in life.’

Patients mostly emerge from their negative emotions over time. They actively adjust their mindset, seek knowledge about ophthalmology and have regular reviews. They are hopeful for the future.

D1: ‘I’m lucky to be alive now.’

D3: ‘You are sick, no one else is to blame. If you can be cured, it’s good, if you can’t be cured, you learn to accept it.’

D4: ‘If you really can’t keep your eyes, you have to face it. I go to the endocrinology every month to draw blood and prescribe medication now. I am relieved when the doctor said it is okay.’

Gratitude and active integration

During this period, DR patients gradually change from their previous fear and worry to positive confrontation and gratitude. Most DR patients are trying to reintegrate into their current lives and work.

D5: ‘I was really lucky to meet Director Wu. We are very happy that the surgery was done well. Now I just live my life as usual, watching TV is basically no problem, and I can cook and eat by myself.’

D6: ‘I was a little cranky before, but after the silicone oil was taken, my eyes can see clearly, so I am very happy now. I’m still young. After my eyes rest for some time, I’m going to catch up on my work.’

Pay attention to patients’ adverse emotions and provide timely information and emotional support

The results of the study show that when DR patients have fundus haemorrhage, patients will experience sudden blurred vision, dark shadows and other subjective feelings, which often make patients feel fear and anxiety. This is consistent with the findings of Shi et al. [ 18 ]. At the same time, DR is characterized by complex treatment, long disease course, and unsatisfactory efficacy. Multiple repeated treatments leave a psychological shadow on the patients, which makes them fearful of the next surgical treatment. The emotional response of DR patients not only affects their treatment effect but also influences disease prognosis and accelerates disease progression [ 29 ]. Therefore, healthcare professionals should comfort and care for patients, guide them to vent their negative emotions and relieve their fear and anxiety. Additionally, they should communicate with patients promptly, inform them of the causes, risk factors, treatment methods and relevant prognostic information, reduce their fear of follow-up treatment and improve their coping ability.

Additionally, DR patients become concerned about the surgical effect, disease prognosis and outcome in the post-surgery phase, with an increased sense of uncertainty. During the discharge adjustment period, the persistent low vision distress and the discomfort of special body positions make DR patients feel negative emotions such as anxiety and depression. Their mental health is impaired. Some DR patients in the adaptation phase after secondary surgery still have poor postoperative outcomes. They have their hopes dashed, show a loss of confidence in continuing treatment and a gradual loss of interest in life, and reach their lowest point emotionally. However, the psychology of DR patients is closely related to their self-care ability, blood glucose management and disease prognosis [ 29 , 30 ]. Therefore, we must address the psychology of DR patients, focusing on DR patients with persistent low vision distress. Various methods should be used to alleviate patients’ negative emotions, such as cognitive behavioural therapy to assess and identify patients’ negative thoughts and help them to effectively overcome their negative thoughts [ 31 ]; as well as the emotional freedom technique to encourage DR patients to tap on acupoints to quickly release their negative emotions [ 32 ]. Simultaneously, it is also possible to hold patient meetings, to use the power of peer support to promote mutual communication with each other, to carry out emotional catharsis, and to relieve their negative emotions [ 33 ].

To stimulate self-efficacy and positive psychological adjustment to improve quality of life

The results of this study showed that patients mostly expressed the importance of their eyes, the inconvenience caused by their limited vision, and their determination to make changes for their eyes in the discharge preparation phase. They planned to regularly review and manage their blood glucose. This high level of self-confidence in DR patients who decide to make changes to maintain their vision is called self-efficacy. Self-efficacy has a beneficial effect on both glycaemic control and quality of life in DR patients [ 34 ]. The study also showed that DR patients had a high level of self-efficacy during the discharge preparation phase and were determined to change. They actively explored self-care modalities that were appropriate for them during the adjustment period. However, the long disease course of DM, the severe complications and the complexity of self-management behaviours make patients with DM prone to diminished self-efficacy [ 35 ]. Therefore, during the post-discharge, adjustment and adaptation period, healthcare professionals should conduct activities such as continuous empowerment and supportive education for DR patients from the perspective of self-efficacy [ 36 , 37 ]. The aim is to involve patients in healthcare decision-making, help them to establish correct perceptions, make self-decisions, conduct self-manage and improve their quality of life.

Provide DR patients with knowledge and skills guidance to enhance their self-care ability through diversified forms of education

This study found that from the discharge preparation period, DR patients questioned their self-care ability and showed a strong desire for DR knowledge and skills guidance. During the adjustment period, some DR patients expressed the futility of seeking help and demonstrated self-care helplessness at home. They hope to receive continuous and professional care guidance through diversified forms of guidance such as WeChat, telephone, and online lectures. Studies have confirmed that DR patients are prone to complications such as bleeding, infection and increased IOP in the early postoperative period. Moreover, improper postoperative care also increases the risk of recurrent retinal detachment, late recurrent vitreous haemorrhage and related secondary surgery [ 38 , 39 ]. Therefore, attention should be paid to improving the self-care ability of patients. Healthcare professionals should develop appropriate disease knowledge instruction plans for DR patients at different stages of the disease and provide DR knowledge and skills instruction through multiple channels such as oral, written, video, and audio. In the discharge preparation phase, patients should be evaluated using a self-care ability scale and personalized guidance provided based on assessment results to help them make the transition from hospital to home care [ 40 ]. Diversified information exchange is carried out after discharge, and multimedia mobile platforms are fully utilized, such as WeChat public platform and WeChat group to push relevant knowledge and to provide online question-and-answer services. At the same time, patients are regularly evaluated and instructed on their self-care ability by combining outpatient diabetes specialists, case management, home visits and remote follow-ups.

Building a multidisciplinary platform to achieve high-quality continuity of care

This study found that DR patients face many difficulties in-home care, such as basic diabetic care, eye care, and mental adjustments which are aggravated by persistent impaired vision. They are very eager to receive professional help in many ways. The care of patients with DR involves multiple disciplines such as ophthalmology, endocrinology, and nutrition. A previous study implemented multidisciplinary teamwork continuity of care for patients with DR, and it improved their blood glucose levels and quality of life [ 41 ]. Therefore, nursing staff should actively play a leading role in building a multidisciplinary team, providing continuity of care services, and discussing and developing vision maintenance, blood glucose management and follow-up treatment plans with ophthalmologists, endocrinologists, optometrists, dieticians and psychological counsellors. At the same time, the active role of specialist nurses is critical. Monthly telephone follow-ups and home visits are conducted by diabetic and ophthalmic specialist nurses to truly develop and implement continuity of care plans. Additionally, multidisciplinary outpatient clinics should be actively created, the construction of the hospital-community-family care model strengthened, and a comprehensive social support system to provide quality and efficient continuity of care services for DR patients established.

Limitations

This study had several limitations. First, the interviewees were limited to one tertiary teaching hospital for collection and the results may not be generalizable. Further surveys could be selected for sampling in non-tertiary hospitals. Second, this study only interviewed DR patients, and the generalization of the results may not be comprehensive. Follow-up studies are needed to further understand the relevant experiences of healthcare professionals and caregivers of DR patients, to gain a deeper understanding of the needs of DR patients, and to provide a higher quality and more comprehensive basis for the development of relevant nursing intervention programs. Third, considering the short hospitalization period of DR patients in this hospital, the same patients were interviewed in the post-surgery phase and discharge preparation phase, which may cause some recall bias. The results can be supplemented by future studies in wards that do not use the day surgery model.

Based on the “Timing It Right” framework, this study conducted in-depth interviews with 40 DR patients at different phases. We found that the feelings and experiences of DR patients were dynamic, ranging from complicated emotional reactions and inadequate coping skills during the pre-surgery period, to increased uncertainty in the post-surgery period, reflecting a strong need for emotional and informational support. From a lack of confidence in self-care during the discharge preparation period to a desire for professional support in the discharge adjustment period, reflecting a need for support for continuity of care; during the adaptation period, most patients compromised, accepted, positively transformed, and actively integrated into their current life trajectory. Therefore, nursing staff should provide appropriate psychological support and professional guidance in a phased and planned manner, especially after the patients are discharged from the hospital, to ensure the continuity of care.

Data Availability

The datasets used and analysed during the current study are available from the corresponding authors on reasonable request.

Abbreviations

• Diabetic retinopathy

Pars Plana Vitrectomy

Panretinal photocoagulation

Anti-vascular endothelial growth factor

Park JY, Hwang JH, Kang MJ, Sim HE, Kim JS, Ko KS, EFFECTS OF GLYCEMIC VARIABILITY ON THE PROGRESSION OF DIABETIC RETINOPATHY AMONG PATIENTS WITH TYPE 2 DIABETES. Retina. 2021;41(7):1487–95.

Article   CAS   PubMed   Google Scholar  

Berrocal MH, Acaba-Berrocal L. Early pars plana vitrectomy for proliferative diabetic retinopathy: update and review of current literature. Curr Opin Ophthalmol. 2021;32(3):203–8.

Article   PubMed   Google Scholar  

Saeedi P, Petersohn I, Salpea P, Malanda B, Karuranga S, Unwin N, et al. Global and regional diabetes prevalence estimates for 2019 and projections for 2030 and 2045: results from the International Diabetes Federation Diabetes Atlas, 9 edition. Diabetes Res Clin Pract. 2019;157:107843.

Shani M, Eviatar T, Komaneshter D, Vinker S. Diabetic Retinopathy -Incidence and Risk factors in A Community Setting- A longitudinal study. Scand J Prim Health Care. 2018;36(3):237–41.

Article   PubMed   PubMed Central   Google Scholar  

Yau JWY, Rogers SL, Kawasaki R, Lamoureux EL, Kowalski JW, Bek T, et al. Global prevalence and major risk factors of diabetic retinopathy. Diabetes Care. 2012;35(3):556–64.

Chua J, Lim CXY, Wong TY, Sabanayagam C. Diabetic Retinopathy in the Asia-Pacific. Asia Pac J Ophthalmol (Phila). 2018;7(1).

Vujosevic S, Aldington SJ, Silva P, Hernández C, Scanlon P, Peto T, et al. Screening for diabetic retinopathy: new perspectives and challenges. Lancet Diabetes Endocrinol. 2020;8(4):337–47.

Yang S, Hsue C, Lou Q. Does patient empowerment predict self-care behavior and glycosylated hemoglobin in chinese patients with type 2 diabetes? Diabetes Technol Ther. 2015;17(5):343–8.

Sapkota R, Chen Z, Zheng D, Pardhan S. The profile of sight-threatening diabetic retinopathy in patients attending a specialist eye clinic in Hangzhou, China. BMJ Open Ophthalmol. 2019;4(1):e000236.

Fenwick EK, Pesudovs K, Rees G, Dirani M, Kawasaki R, Wong TY, et al. The impact of diabetic retinopathy: understanding the patient’s perspective. Br J Ophthalmol. 2011;95(6):774–82.

Assi L, Chamseddine F, Ibrahim P, Sabbagh H, Rosman L, Congdon N, et al. A Global Assessment of Eye Health and Quality of Life: a systematic review of systematic reviews. JAMA Ophthalmol. 2021;139(5):526–41.

Fenwick EK, Cheng GHL, Man REK, Khadka J, Rees G, Wong TY, et al. Inter-relationship between visual symptoms, activity limitation and psychological functioning in patients with diabetic retinopathy. Br J Ophthalmol. 2018;102(7):948–53.

Xu L, Chen S, Xu K, Wang Y, Zhang H, Wang L, et al. Prevalence and associated factors of depression and anxiety among chinese diabetic retinopathy patients: a cross-sectional study. PLoS ONE. 2022;17(4):e0267848.

Article   CAS   PubMed   PubMed Central   Google Scholar  

Rees G, O’Hare F, Saeed M, Sudholz B, Sturrock BA, Xie J, et al. Problem-solving therapy for adults with diabetic retinopathy and diabetes-specific distress: a pilot randomized controlled trial. BMJ Open Diabetes Res Care. 2017;5(1):e000307.

Fenwick E, Rees G, Pesudovs K, Dirani M, Kawasaki R, Wong TY, et al. Social and emotional impact of diabetic retinopathy: a review. Clin Exp Ophthalmol. 2012;40(1):27–38.

Atkinson-Briggs S, Jenkins A, Ryan C, Brazionis L. Mixed diabetic retinopathy screening coverage results in indigenous australian primary care settings: a nurse-led model of integrated diabetes care. J Adv Nurs. 2022;78(10):3187–96.

Zhe C. Application of intensive nursing intervention in patients with diabetic retinopathy. Chin Nurs Res. 2022;36(01):175–7.

Google Scholar  

Shi J, Zhang C, Zhao Q, Zhang X, Guo L, Jia T. Experience of patients with diabetic retinopathy: A qualitative study. Journal of Advanced Nursing. 2022.

Liu M, Wan J, Bian W, Su J. Qualitative study on adaptive challenges in patients with diabetic retinopathy. Chin Nurs Res. 2019;33(16):2783–7.

Cameron JI, Gignac MAM. “Timing it Right”: a conceptual framework for addressing the support needs of family caregivers to stroke survivors from the hospital to the home. Patient Educ Couns. 2008;70(3):305–14.

Lopez KA, Willis DG. Descriptive versus interpretive phenomenology: their contributions to nursing knowledge. Qual Health Res. 2004;14(5):726–35.

Yu X, Huang Y, Liu Y. Nurses’ perceptions of continuing professional development: a qualitative study. BMC Nurs. 2022;21(1):162.

Ophthalmology Society of Chinese Medical Association. Clinical guidelines for diagnosis and treatment of diabetic retinopathy in China. Chin J Ophthalmol. 2014;2014(11):851–65.

Ming L. Using an example to illustrate Colaizzi’S phenomenological data analysis method. J Nurs Sci. 2019;34(11):90–2.

Doyle L, McCabe C, Keogh B, Brady A, McCann M. An overview of the qualitative descriptive design within nursing research. J Res Nurs. 2020;25(5):443–55.

Lincoln YS, Guba EG. Naturalistic inquiry: sage; 1985.

Tong A, Sainsbury P, Craig J. Consolidated criteria for reporting qualitative research (COREQ): a 32-item checklist for interviews and focus groups. Int J Qual Health Care. 2007;19(6):349–57.

Cai H, Hu B, Lu Q, Sun Z, Bi H. Analysis of ache reasons of old patients after retinal reattachment surgery and nurslng care measures. Int Eye Sci. 2010;10(08):1612–4.

Li S, Liu H, Zhu X. The effect of psychotherapy on anxiety, depression, and quality of life in patients with diabetic retinopathy: a protocol for systematic review and network meta-analysis. Med (Baltim). 2021;100(51):e28386.

Article   Google Scholar  

Zhang B, Wang Q, Zhang X, Jiang L, Li L, Liu B. Association between self-care agency and depression and anxiety in patients with diabetic retinopathy. BMC Ophthalmol. 2021;21(1):123.

Safren SA, Gonzalez JS, Wexler DJ, Psaros C, Delahanty LM, Blashill AJ, et al. A randomized controlled trial of cognitive behavioral therapy for adherence and depression (CBT-AD) in patients with uncontrolled type 2 diabetes. Diabetes Care. 2014;37(3):625–33.

Zhang Y, Zhao Y, Xu L, Hu S, Li W. Impact of Emotional Freedom techniques on self-efficacy of elderly patients with type 2 diabetes melfitus. Chin J Nurs. 2017;52(10):1207–10.

Zhao Z, Xie W, Yang Q. Effectiveness of peer support for improving self-management in patients with type 2 diabetes: a Meta-analysis of randomized controlled trial. Chin J Nurs. 2016;51(03):356–64.

Lyu Q-Y, Huang J-W, Li Y-X, Chen Q-L, Yu X-X, Wang J-l, et al. Effects of a nurse led web-based transitional care program on the glycemic control and quality of life post hospital discharge in patients with type 2 diabetes: a randomized controlled trial. Int J Nurs Stud. 2021;119:103929.

Wang D, Wang C, Wang Z. Advances in research on the impact of self-efficacy on self-management behaviours of people with type 2 diabetes. Chin J Prev Control Chronic Dis. 2018;26(07):546–9.

Lambrinou E, Hansen TB, Beulens JW. Lifestyle factors, self-management and patient empowerment in diabetes care. Eur J Prev Cardiol. 2019;26(2suppl):55–63.

Cheng L, Sit JWH, Choi K-C, Chair S-Y, Li X, Wu Y, et al. The effects of an empowerment-based self-management intervention on empowerment level, psychological distress, and quality of life in patients with poorly controlled type 2 diabetes: a randomized controlled trial. Int J Nurs Stud. 2021;116:103407.

Zhao XXS, Chen Y. Antivascular endothelial growth factor agents pretreatment before vitrectomy for complicated proliferative diabetic retinopathy: a meta-analysis of randomised controlled trials. Br J Ophthalmol. 2018;102(8):1077–85.

Sivaprasad S, Prevost AT, Vasconcelos JC, Riddell A, Murphy C, Kelly J, et al. Clinical efficacy of intravitreal aflibercept versus panretinal photocoagulation for best corrected visual acuity in patients with proliferative diabetic retinopathy at 52 weeks (CLARITY): a multicentre, single-blinded, randomised, controlled, phase 2b, non-inferiority trial. Lancet. 2017;389(10085):2193–203.

Kearney BY, Fleischer BJ. Development of an instrument to measure exercise of self-care agency. Res Nurs Health. 1979;2(1):25–34.

Du R, Lu S, Pan H. Practice of multi-disciplinary collaborative transitional care in patients with diabetic retinopathy. J Nurs Sci. 2020;35(13):86–9.

CAS   Google Scholar  

Download references

Acknowledgements

We would like to thank all the patients who participated in this study and the experts who helped us in the design of the interview outline.

This study was supported by General Research Project of Zhejiang Provincial Education Department(Y202250278) and Wenzhou Basic Scientific Research Project (Y20210496).

Author information

Authors and affiliations.

School of Ophthalmology and Optometry, Wenzhou Medical University, 270 West Xueyuan Road, 325027, Wenzhou, Zhejiang, China

Mengyue Zhang, ChunHua Zhang & Chen Chen

The Eye Hospital of Wenzhou Medical University, 270 West Xueyuan Road, 325027, Wenzhou, Zhejiang, China

Linjie Liu, Youping Liang, YiRong Hong, Yanyan Chen & Yinghui Shi

You can also search for this author in PubMed   Google Scholar

Contributions

All authors have made substantial contributions to the following: Yinghui Shi designed the study. Mengyue Zhang, Linjie Liu performed the research and acquired the data. Mengyue Zhang, Chunhua Zhang, Chen Chen analysed and interpreted the data. Mengyue Zhang wrote the paper. Youping Liang and YiRong Hong revised the paper, Yanyan Chen supervised the study. All authors read and approved the final manuscript.

Corresponding authors

Correspondence to Yanyan Chen or Yinghui Shi .

Ethics declarations

Ethics approval and consent to participate.

The study clung to the tenets of the Declaration of Helsinki. This study was granted by the Ethic Committee of Eye Hospital of Wenzhou Medical University (approval number: 2022-045-K-30-01). Informed consent was provided and obtained from all participants before the study commenced.

Consent for publication

Not Applicable.

Competing Interest

No conflict of interest has been declared by authors.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary Material 1

Rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Zhang, M., Zhang, C., Chen, C. et al. The experience of diabetic retinopathy patients during hospital-to-home full-cycle care: A qualitative study. BMC Nurs 22 , 58 (2023). https://doi.org/10.1186/s12912-023-01206-y

Download citation

Received : 13 December 2022

Accepted : 13 February 2023

Published : 03 March 2023

DOI : https://doi.org/10.1186/s12912-023-01206-y

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Patient experience
• Qualitative research
• Timing it right

BMC Nursing

ISSN: 1472-6955

• General enquiries: [email protected]

Type 2 Diabetes

• Heart Disease
• Digestive Health
• Multiple Sclerosis
• Diet & Nutrition
• Supplements
• Health Insurance
• Public Health
• Patient Rights
• Caregivers & Loved Ones
• End of Life Concerns
• Health News
• Thyroid Test Analyzer
• Doctor Discussion Guides
• Hemoglobin A1c Test Analyzer
• Lipid Test Analyzer
• Complete Blood Count (CBC) Analyzer
• What to Buy
• Editorial Process
• Meet Our Medical Expert Board

Type 2 diabetes is a chronic condition affecting the body's ability to process sugar (glucose) for energy , leading to dangerously high levels of blood glucose ( hyperglycemia ). It's the most common form of diabetes. Symptoms of type 2 diabetes may include excessive thirst, frequent urination, and extreme fatigue. The condition can be diagnosed with a simple blood test. In many cases, type 2 diabetes can be managed through lifestyle modifications like diet and exercise, though medication may also be necessary. 

Understanding Type 2 Diabetes

Frequently asked questions.

Type 2 diabetes is caused by insulin resistance (when cells become less sensitive to insulin), or when the pancreas produces less insulin than necessary for proper glucose balance.   While family history and genetics play a role in the development of type 2 diabetes, lifestyle factors such as consuming a diet rich in processed foods, low physical activity, and obesity can contribute, too.

If you're able to closely follow a comprehensive treatment plan (typically including both medication and lifestyle changes), type 2 diabetes can be reversible.   Reversible isn't the same as curable, but it does mean a reduced risk of future complications. It may be possible for some people to wean off medication and manage their diabetes solely through diet and exercise.

Type 1 diabetes is an autoimmune condition in which the body's immune cells attack the pancreas, stopping insulin production.   It usually develops during childhood, but may occur after age 30, too. Type 2 diabetes is a chronic disease in which the body's cells become desensitized to insulin. It typically sets in during adulthood, but may affect children, too.

It seems that if you have a family member who has been diagnosed with type 2 diabetes, you're more likely to develop the condition yourself, suggesting a genetic component.   However, having a genetic disposition is not a guarantee you'll be diagnosed—your lifestyle plays an important part, too. What you eat, how much you exercise, your weight and age all determine which genes can be turned on or off—a concept called epigenetics.

Causes and Risk Factors

The hemoglobin A1C test is a blood test used to give a glimpse of your average blood sugar levels over the past three months. It's often used in conjunction with at-home glucose monitoring to provide a picture of your overall blood sugar management. It can also be helpful in diagnosing prediabetes or diabetes.

Glucose is broken down (metabolized) from carbohydrates in the diet and serves as the body's primary source of energy. When it enters the bloodstream, it's called blood glucose, or blood sugar. In people with diabetes, the body has difficulty regulating the balance of glucose in the blood versus the cells.

Carbohydrates are one of the three macronutrients (along with protein and fat) that the body uses as fuel. Carbs are found in fruit, grains, starchy vegetables, beans, legumes, dairy, and sweets. Carbs impact blood sugar more than other foods, which means people with diabetes may benefit from following a reduced-carb diet.

A group of conditions marked by an inability of insulin to control glucose. There are several types of diabetes mellitus, including type 1 diabetes, type 2 diabetes, gestational diabetes , latent autoimmune diabetes in adults (LADA), and monogenic diabetes.

Neuropathy is nerve damage that results from consistently elevated levels of blood glucose, which may show up as numbness or tingling in the feet and hands, muscle weakness, or difficulty walking.

The fasting blood glucose test measures the amount of sugar (glucose) in your blood. It's a simple blood test that requires you to fast for at least 8 hours (usually overnight) before getting your blood drawn. The test can be used to screen for and diagnose prediabetes and diabetes.

Hyperglycemia, or high blood sugar, is a condition that occurs when there is too much glucose in the bloodstream. Symptoms may include excessive thirst, increased urination, extreme fatigue, nausea, increased hunger, and blurry vision. Hyperglycemia may set in when insulin function is impaired due to either type 1 or type 2 diabetes, but it also may result from very high levels of stress.  

Hypoglycemia, or low blood sugar, is a condition that occurs when there isn't enough glucose in the bloodstream. Symptoms may include dizziness, shakiness, rapid heartbeat, sweating, and headache. Hypoglycemia may be due to diabetes, or it could result from other causes, like delaying or skipping meals, vigorous exercise, certain medications, drinking too much alcohol.

A hormone produced by the pancreas that's responsible for shuttling glucose (blood sugar) into the cell for storage and energy). Insulin acts as a "key" that binds to insulin receptors on the cell surface, "unlocking" the cell membrane so glucose can enter. When insulin production is lower than normal, or if the cells become less sensitive to insulin, over time, this can result in type 2 diabetes.  

Prediabetes is a condition described as impaired glucose tolerance, and is a precursor to diabetes. It essentially means your body is having some difficulty balancing your blood glucose level. Prediabetes may not cause symptoms, but if left unchecked, it may progress to diabetes.   It can be diagnosed through a fasting blood glucose test and/or an A1C test.

More In Diabetes

National Institute of Diabetes and Digestive and Kidney Diseases. What is diabetes . December 2016.

National Institute of Diabetes and Digestive and Kidney Diseases. Symptoms & causes of diabetes . December 2016.

Prasad RB, Groop L. Genetics of type 2 diabetes-pitfalls and possibilities . Genes (Basel). 2015;6(1):87-123. doi: 10.3390/genes6010087

Goyal N, Kaur R, Sud A, Ghorpade N, Gupta M. Non Diabetic and Stress Induced Hyperglycemia [SIH] in Orthopaedic Practice What do we know so Far? . J Clin Diagn Res. 2014;8(10):LH01-3. doi:10.7860/JCDR/2014/10027.5022

U.S. Centers for Disease Control and Prevention. Insulin resistance and diabetes . Updated 2019.

National Institute of Diabetes and Digestive and Kidney Diseases. Insulin resistance and prediabetes . December 2016.

American Diabetes Association. Peripheral neuropathy .

The National Institute of Diabetes and Digestive and Kidney Diseases. Low Blood Glucose (Hypoglycemia) .

EDITORIAL article

Editorial: advances in the research of diabetic retinopathy.

• 1 Western University, London, ON, Canada
• 2 University of Campania Luigi Vanvitelli, Caserta, Italy
• 3 Strategic Center for Diabetes Research, College of Medicine, King Saud University, Riyadh, Saudi Arabia

Advances in the research of diabetic retinopathy

Diabetes remains a planetary crisis with its prevalence estimated to increase by nearly 50% in the next 25 years ( 1 , 2 ) One of the main challenges for the diabetic patients is the development of chronic complications, leading to end organ damage. Chronic diabetic complications are a major cause of mortality and morbidity for the people living with diabetes. Retinal damage in diabetics, also known as diabetic retinopathy (DR) is a leading cause of blindness in working-aged adults ( 3 , 4 ). Although DR begins with asymptomatic hyperglycemic damage to the retinal microvasculature, in particularly endothelial cells, it eventually causes a symphony of abnormalities at various levels, creating cellular dysfunction and damage, ultimately leading to functional and structural changes in the retina that result in vision impairment and blindness ( 5 – 7 ). Current treatment approaches serve as band-aid solutions that address the root of the problem in a very limited perspective ( 8 , 9 ). To develop a solid preventive and therapeutic approach for DR, a better understanding of this disease is essential.

This Research Topic presents a large number of articles including original research as well as review to improve our understanding of DR. The specific topics represent various levels of complexities. The publications address issues ranging from the molecular level to cellular level to animal levels and finally to DR patients. The articles investigate and discuss specific pathogenetic mechanisms, effects of current treatment modalities and treatment outcomes.

It was also to be noted that this collection also identifies potential upcoming treatment modalities and diagnostic approach using various RNA molecules as well as application of artificial intelligence and machine learning for DR diagnosis and assessment of prognosis.

Among the review topics Zhang et al. compared effectiveness of panretinal photocoagulation alone along with in combination of anti VEGF treatment. Guo et al. discussed uric acid abnormalities, an understudied area in DR. Similarly, Zheng et al. reviewed another relatively unappreciated topic, i.e, relationship of sleep quality with risk of developing DR. Furthermore, Zhang et al. reviewed the effects of calcium dobesylate treatments in patients with non-proliferative DR.

As we are entering in an area of RNA based therapy and diagnosis, two of the presented articles provided valuable insight in this area. Niu et al. reviewed exosomes/microRNAs in the treatment of DR and Kowluru discussed Long Noncoding RNAs and Mitochondrial Homeostasis in the Development of DR. Practical applications of these reviews these reviews were also supported by original research articles. In one such article Biswas et al. described the use of a serum lncRNAs panel to diagnose DR. In the other article Yang et al. characterised small RNAs and microRNAs in the vitreous Humor of proliferative DR. It was also revealed that plasma metabolomic profiling ( Sun et al. ) growth differentiation factor 15 levels ( Niu et al. ) may also be effective in the assessment of DR.

Several researchers presented original research describing roles of various additional molecules in DR, including secreted cystine rich acidic protein ( Luo et al. ), homocysteine ( Luo et al. ) and retinal inflammation and macrophage infiltration ( Meng et al. ).

The articles presented in this Research Topic further showed the impact of glycemic control on photoreceptor Layers and RPE in type 2 Diabetes ( Ishibashi et al. ). Furthermore, Xie et al. showed assessment of fundus structure using OCT may act as a predictive model for treatment effect for diabetic macular edema. Best corrected visual acuity in macular edema may also be predicted by OCT ( Li et al. ). Deng et al. proposed the use of hand-held ERG device for DR Screening. In addition, it was shown that artificial intelligence and machine learning may also predict DR in type 2 diabetic patients ( Zhao et al. ). It was also important to note the data of Kailuan eye study ( Yongpeng et al ), showing that DR is an independent risk factor for dry AMD.

In conclusion, this Research Topic brings insights and a wealth of knowledge regarding DR, involving its pathogenesis, diagnostic modalities, clinical presentation and treatment. We sincerely feel that these set of articles have set the stage for new knowledge creation and their clinical application in this area in future.

Author contributions

SC drafted the manuscript. ML and KS reviewed, provided input and approved the content.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

1. Saeedi P, Petersohn I, Salpea P, Malanda B, Karuranga S, Unwin N, et al. Global and regional diabetes prevalence estimates for 2019 and projections for 2030 and 2045: Results from the international diabetes federation diabetes atlas, 9 th edition. Diabetes Res Clin Pract (2019) 157:107843. doi: 10.1016/j.diabres.2019.107843

PubMed Abstract | CrossRef Full Text | Google Scholar

2. Cheung N, Mitchell P, Wong TY. Diabetic retinopathy. Lancet (Lond Engl) (2010) 376(9735):124–36. doi: 10.1016/S0140-6736(09)62124-3

CrossRef Full Text | Google Scholar

3. Leasher JL, Bourne RRA, Flaxman SR. Global estimates on the number of people blind or visually impaired by diabetic retinopathy: A meta-analysis from 1990 to 2010. Dia Care (2016) 39:1643–9. doi: 10.2337/dc15-2171

4. Teo ZL, Tham YC, Yu M, Chee ML, Rim TH, Cheung N, et al. Global prevalence of diabetic retinopathy and projection of burden through 2045. Ophthalmology (2021) 128(11):1580–91. doi: 10.1016/j.ophtha.2021.04.027

5. Khan ZA, Chakrabarti S. Chronic diabetic complications: Endothelial cells at the frontline. chapter 9. In: Rahman AU, Choudhary I, editors. Front Cardiovasc Drug Discov . Sharjah, U.A.E: Bentham Science publishers (2010). p. pp121–137.

Google Scholar

6. Gui F, You Z, Fu S, Wu H, Zhang Y. Endothelial dysfunction in diabetic retinopathy. Front Endocrinol (Lausanne) (2020) 11:591. doi: 10.3389/fendo.2020.00591

7. Brownlee M. The pathobiology of diabetic complications: A unifying mechanism. Diabetes (2005) 54:1615–25. doi: 10.2337/diabetes.54.6.1615

8. Duh EJ, Sun JK, Stitt AW. Diabetic retinopathy: current understanding, mechanisms, and treatment strategies. JCI Insight (2017) 2(14):e93751. doi: 10.1172/jci.insight.93751

9. Tomita Y, Lee D, Tsubota K, Negishi K, Kurihara T. Updates on the current treatments for diabetic retinopathy and possibility of future oral therapy. J Clin Med (2021) 10:4666. doi: 10.3390/jcm10204666

Keywords: diabetic retinopathy, pathogenesis, mechanisms, diagnosis, treatment

Citation: Chakrabarti S, Lanza M and Siddiqui K (2022) Editorial: Advances in the research of diabetic retinopathy. Front. Endocrinol. 13:1038056. doi: 10.3389/fendo.2022.1038056

Received: 06 September 2022; Accepted: 12 October 2022; Published: 25 October 2022.

Reviewed by:

Copyright © 2022 Chakrabarti, Lanza and Siddiqui. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: Subrata Chakrabarti, [email protected]

Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.

• Skip to main content
• Skip to FDA Search
• Skip to in this section menu
• Skip to footer links

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you're on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

U.S. Food and Drug Administration

•   Search
•   Menu
• Tobacco Products
• Public Health Education

Health Effects of Tobacco Use

How smoking can contribute to vision loss and blindness.

On this page:

How Can Smoking Affect My Eyesight and Vision?

Can smoking lead to vision loss and blindness, what are symptoms of eye diseases related to smoking , can smoking cause cataracts, can smoking cause age-related macular degeneration (amd), can smoking cause graves’ ophthalmopathy or thyroid eye disease, can smoking cause the onset or progression of diabetic retinopathy, can smoking cause glaucoma, how can quitting smoking protect my eyes.

When you smoke cigarettes, you can damage important parts of your eyes necessary for maintaining clear eyesight and vision . This damage can make your vision cloudy, reduce your field of vision, or even cause you to lose your eyesight completely. 

Smoking cigarettes can affect your:

• Retina : The delicate, light-sensitive tissue that lines the inside of the eye.
• Lens : The clear part of the eye that allows light to pass to the retina and allows the eye to focus on objects at varying distances. 
• Macula : The most sensitive part of the retina and is the part of the eye that supplies sharp vision. 1

Order Free Print Version | Download Infographic

Yes, smoking cigarettes can cause eye diseases that can lead to vision loss and blindness . Smoking cigarettes can put you at greater risk of developing two serious eye diseases:

• Cataracts : Clouding of the eye’s normally clear lens, causing loss of vision.
• Age-related macular degeneration (AMD) : Gradual destruction of the eye’s macula, which can lead to loss of vision in the center of the eye.

People who smoke cigarettes are two to three times more likely to develop cataracts 2 and up to four times more likely to develop AMD than people who don’t smoke. 3

Smoking cigarettes can cause two serious eye diseases: cataracts and age-related macular degeneration (AMD). Symptoms of these eye diseases related to smoking include :

Cataracts :

• Cloudy or blurry vision
• Colors that seem faded
• Sensitivity to light
• Trouble seeing at night
• Double vision
• Loss of the central vision you need to see details straight ahead
• Blurry or wavy areas in your central vision
• Having a hard time recognizing faces
• Needing more light to read or do other tasks in front of you

You should be aware that in the early stages of both eye diseases you may experience no symptoms at all. It’s important to get eye exams on a regular basis to increase the chances of catching and treating these conditions as early as possible. 

Yes, smoking cigarettes can cause cataracts. People who smoke cigarettes are two to three times more likely to develop cataracts than people who don’t smoke . 2  

A cataract is the clouding of the eye’s lens , which is normally clear. Cataracts cause loss of vision because the clouding prevents light from passing through the lens to the retina. 

Most cataracts develop slowly over the course of years and the risk for developing cataracts goes up as you get older. Smoking further increases your chance of developing cataracts.

You may not notice symptoms of cataracts at first. But as cataracts worsen, they can affect your ability to see clearly. 

Yes, smoking cigarettes can cause AMD . 1 People who smoke cigarettes are up to four times more likely to develop AMD than people who don’t smoke. 3

The retina is the delicate, light-sensitive tissue that lines the inside of the eye. AMD is an eye disease that affects your macula, the most sensitive part of the retina and the part of the eye that supplies sharp vision.

When you develop AMD, it gradually destroys the macula and can ultimately lead to loss of vision in the center of the eye. This can make it difficult to see or do things in front of you, such as recognize faces, read, drive, or perform chores around the house. 

AMD can progress quickly or slowly, depending on the person and their risk. Your risk for developing AMD goes up as you get older and is higher if you smoke cigarettes. 

Since AMD is a common condition, especially in older adults, and you may not notice symptoms for a long time, it is important to have regular eye exams.

More research is needed to find out whether smoking cigarettes can cause ophthalmopathy associated with Graves’ disease . Graves’ ophthalmopathy, also known as Graves’ eye disease or thyroid eye disease , can cause irritation, sensitivity, double vision, and other symptoms in the eyes as a result of an overactive thyroid caused by Graves’ disease.

The data from research conducted on the relationship between smoking and Graves’ ophthalmopathy is not strong enough to conclude that smoking is a cause of thyroid eye disease. However, some evidence suggests that smoking cigarettes can put you at an increased risk for Graves’ eye disease if you have Graves’ disease. 1,2 Since the role of smoking in Graves’ ophthalmopathy is unclear, more research is needed. 

More research is needed to find out whether smoking cigarettes can cause the onset or progression of retinopathy in people with diabetes . 1,2 Diabetic retinopathy affects the blood vessels in your retina, the delicate, light-sensitive tissue that lines the inside of the eye, and can cause vision loss and blindness in people who have diabetes. 

More research is needed to find out whether smoking cigarettes can cause glaucoma . Glaucoma is a group of eye diseases that can cause vision loss and blindness by damaging a nerve in the back of your eye called the optic nerve. 

The data from research conducted on the relationship between smoking and glaucoma does not tell us whether smoking is a cause of glaucoma. 1,2  

You can lower your risk of developing cataracts and age-related macular degeneration by quitting smoking cigarettes . 

It can be overwhelming to experience changes in your vision. If you smoke cigarettes and are concerned about your eyesight, speak with your health care provider and schedule an eye exam.

Get Help Quitting

• Patient Care & Health Information
• Diseases & Conditions
• Type 1 diabetes

What is type 1 diabetes? A Mayo Clinic expert explains

Learn more about type 1 diabetes from endocrinologist Yogish Kudva, M.B.B.S.

I'm Dr. Yogish C. Kudva an endocrinologist at Mayo Clinic. In this video, we'll cover the basics of type 1 diabetes. What is it? Who gets it? The symptoms, diagnosis, and treatment. Whether you're looking for answers for yourself or someone you love. We are here to give you the best information available. Type 1 diabetes is a chronic condition that affects the insulin making cells of the pancreas. It's estimated that about 1.25 million Americans live with it. People with type 1 diabetes don't make enough insulin. An important hormone produced by the pancreas. Insulin allows your cells to store sugar or glucose and fat and produce energy. Unfortunately, there is no known cure. But treatment can prevent complications and also improve everyday life for patients with type 1 diabetes. Lots of people with type 1 diabetes live a full life. And the more we learn and develop treatment for the disorder, the better the outcome.

We don't know what exactly causes type 1 diabetes. We believe that it is an auto-immune disorder where the body mistakenly destroys insulin producing cells in the pancreas. Typically, the pancreas secretes insulin into the bloodstream. The insulin circulates, letting sugar enter your cells. This sugar or glucose, is the main source of energy for cells in the brain, muscle cells, and other tissues. However, once most insulin producing cells are destroyed, the pancreas can't produce enough insulin, meaning the glucose can't enter the cells, resulting in an excess of blood sugar floating in the bloodstream. This can cause life-threatening complications. And this condition is called diabetic ketoacidosis. Although we don't know what causes it, we do know certain factors can contribute to the onset of type 1 diabetes. Family history. Anyone with a parent or sibling with type 1 diabetes has a slightly increased risk of developing it. Genetics. The presence of certain genes can also indicate an increased risk. Geography. Type 1 diabetes becomes more common as you travel away from the equator. Age, although it can occur at any age there are two noticeable peaks. The first occurs in children between four and seven years of age and the second is between 10 and 14 years old.

Signs and symptoms of type 1 diabetes can appear rather suddenly, especially in children. They may include increased thirst, frequent urination, bed wetting in children who previously didn't wet the bed. Extreme hunger, unintended weight loss, fatigue and weakness, blurred vision, irritability, and other mood changes. If you or your child are experiencing any of these symptoms, you should talk to your doctor.

The best way to determine if you have type 1 diabetes is a blood test. There are different methods such as an A1C test, a random blood sugar test, or a fasting blood sugar test. They are all effective and your doctor can help determine what's appropriate for you. If you are diagnosed with diabetes, your doctor may order additional tests to check for antibodies that are common in type 1 diabetes in the test called C-peptide, which measures the amount of insulin produced when checked simultaneously with a fasting glucose. These tests can help distinguish between type 1 and type 2 diabetes when a diagnosis is uncertain.

If you have been diagnosed with type 1 diabetes, you may be wondering what treatment looks like. It could mean taking insulin, counting carbohydrates, fat protein, and monitoring your glucose frequently, eating healthy foods, and exercising regularly to maintain a healthy weight. Generally, those with type 1 diabetes will need lifelong insulin therapy. There are many different types of insulin and more are being developed that are more efficient. And what you may take may change. Again, your doctor will help you navigate what's right for you. A significant advance in treatment from the last several years has been the development and availability of continuous glucose monitoring and insulin pumps that automatically adjust insulin working with the continuous glucose monitor. This type of treatment is the best treatment at this time for type 1 diabetes. This is an exciting time for patients and for physicians that are keen to develop, prescribe such therapies. Surgery is another option. A successful pancreas transplant can erase the need for additional insulin. However, transplants aren't always available, not successful and the procedure can pose serious risks. Sometimes it may outweigh the dangers of diabetes itself. So transplants are often reserved for those with very difficult to manage conditions. A successful transplant can bring life transforming results. However, surgery is always a serious endeavor and requires ample research and concentration from you, your family, and your medical team.

The fact that we don't know what causes type 1 diabetes can be alarming. The fact that we don't have a cure for it even more so. But with the right doctor, medical team and treatment, type 1 diabetes can be managed. So those who live with it can get on living. If you would like to learn even more about type 1 diabetes, watch our other related videos or visit mayoclinic.org. We wish you well.

Type 1 diabetes, once known as juvenile diabetes or insulin-dependent diabetes, is a chronic condition. In this condition, the pancreas makes little or no insulin. Insulin is a hormone the body uses to allow sugar (glucose) to enter cells to produce energy.

Different factors, such as genetics and some viruses, may cause type 1 diabetes. Although type 1 diabetes usually appears during childhood or adolescence, it can develop in adults.

Even after a lot of research, type 1 diabetes has no cure. Treatment is directed toward managing the amount of sugar in the blood using insulin, diet and lifestyle to prevent complications.

Products & Services

• A Book: The Essential Diabetes Book

Type 1 diabetes symptoms can appear suddenly and may include:

• Feeling more thirsty than usual
• Urinating a lot
• Bed-wetting in children who have never wet the bed during the night
• Feeling very hungry
• Losing weight without trying
• Feeling irritable or having other mood changes
• Feeling tired and weak
• Having blurry vision

When to see a doctor

Talk to your health care provider if you notice any of the above symptoms in you or your child.

The exact cause of type 1 diabetes is unknown. Usually, the body's own immune system — which normally fights harmful bacteria and viruses — destroys the insulin-producing (islet) cells in the pancreas. Other possible causes include:

• Exposure to viruses and other environmental factors

The role of insulin

Once a large number of islet cells are destroyed, the body will produce little or no insulin. Insulin is a hormone that comes from a gland behind and below the stomach (pancreas).

• The pancreas puts insulin into the bloodstream.
• Insulin travels through the body, allowing sugar to enter the cells.
• Insulin lowers the amount of sugar in the bloodstream.
• As the blood sugar level drops, the pancreas puts less insulin into the bloodstream.

The role of glucose

Glucose — a sugar — is a main source of energy for the cells that make up muscles and other tissues.

• Glucose comes from two major sources: food and the liver.
• Sugar is absorbed into the bloodstream, where it enters cells with the help of insulin.
• The liver stores glucose in the form of glycogen.
• When glucose levels are low, such as when you haven't eaten in a while, the liver breaks down the stored glycogen into glucose. This keeps glucose levels within a typical range.

In type 1 diabetes, there's no insulin to let glucose into the cells. Because of this, sugar builds up in the bloodstream. This can cause life-threatening complications.

Risk factors

Some factors that can raise your risk for type 1 diabetes include:

• Family history. Anyone with a parent or sibling with type 1 diabetes has a slightly higher risk of developing the condition.
• Genetics. Having certain genes increases the risk of developing type 1 diabetes.
• Geography. The number of people who have type 1 diabetes tends to be higher as you travel away from the equator.
• Age. Type 1 diabetes can appear at any age, but it appears at two noticeable peaks. The first peak occurs in children between 4 and 7 years old. The second is in children between 10 and 14 years old.

Complications

Over time, type 1 diabetes complications can affect major organs in the body. These organs include the heart, blood vessels, nerves, eyes and kidneys. Having a normal blood sugar level can lower the risk of many complications.

Diabetes complications can lead to disabilities or even threaten your life.

• Heart and blood vessel disease. Diabetes increases the risk of some problems with the heart and blood vessels. These include coronary artery disease with chest pain (angina), heart attack, stroke, narrowing of the arteries (atherosclerosis) and high blood pressure.

Nerve damage (neuropathy). Too much sugar in the blood can injure the walls of the tiny blood vessels (capillaries) that feed the nerves. This is especially true in the legs. This can cause tingling, numbness, burning or pain. This usually begins at the tips of the toes or fingers and spreads upward. Poorly controlled blood sugar could cause you to lose all sense of feeling in the affected limbs over time.

Damage to the nerves that affect the digestive system can cause problems with nausea, vomiting, diarrhea or constipation. For men, erectile dysfunction may be an issue.

• Kidney damage (nephropathy). The kidneys have millions of tiny blood vessels that keep waste from entering the blood. Diabetes can damage this system. Severe damage can lead to kidney failure or end-stage kidney disease that can't be reversed. End-stage kidney disease needs to be treated with mechanical filtering of the kidneys (dialysis) or a kidney transplant.
• Eye damage. Diabetes can damage the blood vessels in the retina (part of the eye that senses light) (diabetic retinopathy). This could cause blindness. Diabetes also increases the risk of other serious vision conditions, such as cataracts and glaucoma.
• Foot damage. Nerve damage in the feet or poor blood flow to the feet increases the risk of some foot complications. Left untreated, cuts and blisters can become serious infections. These infections may need to be treated with toe, foot or leg removal (amputation).
• Skin and mouth conditions. Diabetes may leave you more prone to infections of the skin and mouth. These include bacterial and fungal infections. Gum disease and dry mouth also are more likely.
• Pregnancy complications. High blood sugar levels can be dangerous for both the parent and the baby. The risk of miscarriage, stillbirth and birth defects increases when diabetes isn't well-controlled. For the parent, diabetes increases the risk of diabetic ketoacidosis, diabetic eye problems (retinopathy), pregnancy-induced high blood pressure and preeclampsia.

There's no known way to prevent type 1 diabetes. But researchers are working on preventing the disease or further damage of the islet cells in people who are newly diagnosed.

Ask your provider if you might be eligible for one of these clinical trials. It is important to carefully weigh the risks and benefits of any treatment available in a trial.

• Summary of revisions: Standards of medical care in diabetes — 2022. Diabetes Care. 2022; doi:10.2337/dc22-Srev.
• Papadakis MA, et al., eds. Diabetes mellitus. In: Current Medical Diagnosis & Treatment 2022. 61st ed. McGraw Hill; 2022. https://accessmedicine.mhmedical.com. Accessed May 4, 2022.
• What is diabetes? National Institute of Diabetes and Digestive and Kidney Diseases. https://www.niddk.nih.gov/health-information/diabetes/overview/what-is-diabetes. Accessed May 4, 2022.
• Levitsky LL, et al. Epidemiology, presentation, and diagnosis of type 1 diabetes mellitus in children and adolescents. https://www.uptodate.com/contents/search. Accessed May 4, 2022.
• Diabetes mellitus (DM). Merck Manual Professional Version. https://www.merckmanuals.com/professional/endocrine-and-metabolic-disorders/diabetes-mellitus-and-disorders-of-carbohydrate-metabolism/diabetes-mellitus-dm. Accessed May 4, 2022.
• AskMayoExpert. Type 1 diabetes mellitus. Mayo Clinic; 2021.
• Robertson RP. Pancreas and islet transplantation in diabetes mellitus. https://www.uptodate.com/contents/search. Accessed May 4, 2022.
• Levitsky LL, et al. Management of type 1 diabetes mellitus in children during illness, procedures, school, or travel. https://www.uptodate.com/contents/search. Accessed May 4, 2022.
• Hyperglycemia (high blood glucose). American Diabetes Association. https://www.diabetes.org/healthy-living/medication-treatments/blood-glucose-testing-and-control/hyperglycemia. Accessed May 4, 2022.
• Diabetes and DKA (ketoacidosis). American Diabetes Association. https://www.diabetes.org/diabetes/dka-ketoacidosis-ketones. Accessed May 4, 2022.
• Insulin resistance & prediabetes. National Institute of Diabetes and Digestive and Kidney Diseases. https://www.niddk.nih.gov/health-information/diabetes/overview/what-is-diabetes/prediabetes-insulin-resistance. Accessed May 4, 2022.
• Blood sugar and insulin at work. American Diabetes Association. https://www.diabetes.org/tools-support/diabetes-prevention/high-blood-sugar. Accessed May 4, 2022.
• Inzucchi SE, et al. Glycemic control and vascular complications in type 1 diabetes. https://www.uptodate.com/contents/search. Accessed May 4, 2022.
• Diabetes and oral health. American Diabetes Association. https://www.diabetes.org/diabetes/keeping-your-mouth-healthy. Accessed May 4, 2022.
• Drug treatment of diabetes mellitus. Merck Manual Professional Version. https://www.merckmanuals.com/professional/endocrine-and-metabolic-disorders/diabetes-mellitus-and-disorders-of-carbohydrate-metabolism/drug-treatment-of-diabetes-mellitus. Accessed May 4, 2022.
• Weinstock DK, et al. Management of blood glucose in adults with type 1 diabetes mellitus. https://www.uptodate.com/contents/search. Accessed May 7, 2022.
• FDA proves first automated insulin delivery device for type 1 diabetes. U.S. Food and Drug Administration. https://www.fda.gov/news-events/press-announcements/fda-approves-first-automated-insulin-delivery-device-type-1-diabetes. Accessed May 4, 2022.
• Boughton CK, et al. Advances in artificial pancreas systems. Science Translational Medicine. 2019; doi:10.1126/scitranslmed.aaw4949.
• Hypoglycemia (low blood sugar). American Diabetes Association. https://www.diabetes.org/healthy-living/medication-treatments/blood-glucose-testing-and-control/hypoglycemia. Accessed May 4, 2022.
• Diabetes in the workplace and the ADA. U.S. Equal Opportunity Employment Commission. https://www.eeoc.gov/laws/guidance/diabetes-workplace-and-ada. Accessed May 4, 2022.
• Cardiovascular disease and risk management: Standards of medical care in diabetes — 2022. Diabetes Care. 2022; doi:10.2337/dc22-S010.
• Diabetes technology. Standards of Medical Care in Diabetes — 2022. 2022; doi:10.2337/dc22-S007.
• FDA authorizes a second artificial pancreas system. JDRF. https://www.jdrf.org/blog/2019/12/13/jdrf-reports-fda-authorizes-second-artificial-pancreas-system/. Accessed May 4, 2022.
• Classification and diagnosis of diabetes: Standards of medical care in diabetes — 2022. Diabetes Care. 2022; doi:10.2337/dc22-S002.
• Retinopathy, neuropathy, and foot care: Standards of medical care in diabetes — 2022. Diabetes Care. 2022; doi:10.2337/dc22-S012.
• Glycemic targets: Standards of medical care in diabetes — 2022. Diabetes Care. 2022; doi:10.2337/dc22-S012.
• Pharmacologic approaches to glycemic treatment: Standards of medical care in diabetes — 2022. Diabetes Care. 2022; doi:10.2337/dc22-S009.
• Facilitating behavior change and well-being to improve health outcomes: Standards of medical care in diabetes — 2022. Diabetes Care. 2022; doi:10.2337/dc22-S005.
• Centers for Disease Control and Prevention. Use of hepatitis B vaccination for adults with diabetes mellitus: Recommendations of the Advisory Committee on Immunization Practices (ACIP). Morbidity and Mortality Weekly Report. 2011;60:1709.
• Management of diabetes in pregnancy: Standards of medical care in diabetes — 2022. Diabetes Care. 2022; doi:10.2337/dc22-S015.
• Older adults: Standards of medical care in diabetes — 2022. Diabetes Care. 2022; doi:10.2337/dc22-S013.
• FDA approves first-of-its-kind automated insulin delivery and monitoring system for use in young pediatric patients. U.S. Food and Drug Administration. https://www.fda.gov/news-events/press-announcements/fda-approves-first-its-kind-automated-insulin-delivery-and-monitoring-system-use-young-pediatric#:~:text=Today, the U.S. Food and,by individuals aged 2 to. Accessed May 8, 2022.
• What you need to know: Getting a COVID-19 vaccine. American Diabetes Association. https://www.diabetes.org/coronavirus-covid-19/vaccination-guide. Accessed June 1, 2022.

News from Mayo Clinic

• Driven by family, fueled by hope: Mayo Clinic researcher fights against Type 1 diabetes Aug. 06, 2023, 11:00 a.m. CDT
• Symptoms & causes
• Diagnosis & treatment
• Doctors & departments

Mayo Clinic does not endorse companies or products. Advertising revenue supports our not-for-profit mission.

• Opportunities

Mayo Clinic Press

Check out these best-sellers and special offers on books and newsletters from Mayo Clinic Press .

• Mayo Clinic on Incontinence - Mayo Clinic Press Mayo Clinic on Incontinence
• The Essential Diabetes Book - Mayo Clinic Press The Essential Diabetes Book
• Mayo Clinic on Hearing and Balance - Mayo Clinic Press Mayo Clinic on Hearing and Balance
• FREE Mayo Clinic Diet Assessment - Mayo Clinic Press FREE Mayo Clinic Diet Assessment
• Mayo Clinic Health Letter - FREE book - Mayo Clinic Press Mayo Clinic Health Letter - FREE book

We’re transforming healthcare

Make a gift now and help create new and better solutions for more than 1.3 million patients who turn to Mayo Clinic each year.