ARTIFICIAL INTELLIGENCE FOR DIABETIC RETINOPATHY
Editor:
Rachelle Srinivas, DO
Dr. Srinivas is an Ophthalmology resident at Henry Ford Health - Warren
Diabetic retinopathy is rapidly becoming a global crisis, with millions at risk of losing their vision. While early screening and treatment can prevent most cases of blindness, access remains severely limited in many parts of the world. But what if technology could change everything? Read more to find out how the latest technologies can help!
Prevalence
With the high prevalence of diabetes mellitus and diabetic retinopathy (DR) being one of the leading causes of blindness affecting adults, DR is a concerning issue worldwide. The burden of DR has been increasing, with a projected worldwide prevalence of approximately 160 million people affected by 2045 (1). Unfortunately, there is also a disproportionately high burden of disease affecting low- and middle-income countries (2, 3). Concurrently, many of the countries with the highest prevalence of vision-threatening DR have also been found to have the lowest number of ophthalmologists per 1000 affected patients (3).
Disparities in Care and the Role of Prevention and Screening
Primary and secondary prevention through screening and early treatment of DR is an effective method to combat the increasing incidence of DR (2). This approach can especially be effective in low—and middle-income countries, where tertiary prevention may not be as successful due to an overwhelming mismatch between trained providers and the need for treatment of DR (3).
With the recent rise in popularity of artificial intelligence (AI) with machine learning and deep learning and its incorporation in medicine, there have also been increased efforts focused on utilizing AI to improve access to care, and screening and prevention of disease (4). The incorporation of AI could help alleviate the disproportionate mismatch between disease prevalence and the limited number of ophthalmologists available to conduct timely examinations (3, 5).
AI has been proposed for screening purposes and to evaluate the progression of DR, and there are several AI models under evaluation for their ability and effectiveness as a screening tool for early identification and treatment referral of DR (5).
Incorporation of Prevention/Screening Methods
The Optomed Aurora, a portable fundus camera, has an in-built AI algorithm that can be used for DR screening (4). A recent study was conducted in a real-world clinical setting that included diabetic and non-diabetic patients. Images obtained with the camera were analyzed by a skilled operator and the AI program. Results were promising, with the AI algorithm successfully identifying DR in a majority of subjects and a sensitivity of 96.8% and specificity of 96.8% (4).
A recent study conducted by Abrahamoff et al. aimed to evaluate the performance of an AI system, IDx-DR, for use in primary care settings to detect more than mild DR. The study compared the detection of DR between the AI system and certified photographers (6). The AI system was found to have promising results, with a sensitivity of 87.2% and specificity of 90.7%. The IDx-DR has since been approved by the FDA for the detection of more than mild DR (6).
Another study in Zambia, a lower-middle income country, compared the detection of referable diabetic retinopathy, defined as moderate non-proliferative DR or worse, between an AI model and human graders. Similar to previous studies, they also found the AI system to have a sensitivity of 92.25% and specificity of 89.04% for detecting referable DR (2, 7).
In conclusion, the promising results from such studies make it likely that AI can soon be more widely implemented in low-resource areas to increase screening and detection of DR and subsequently lead to patients receiving timely follow-up and treatment.
References
1. Vision Loss Expert Group of the Global Burden of Disease Study; GBD 2019 Blindness and Vision Impairment Collaborators. Global estimates on the number of people blind or visually impaired by diabetic retinopathy: a meta-analysis from 2000 to 2020. Eye (Lond). 2024;38(11):2047-2057. doi:10.1038/s41433-024-03101-5
2. Wong TY, Sabanayagam C. Strategies to Tackle the Global Burden of Diabetic Retinopathy: From Epidemiology to Artificial Intelligence. Ophthalmologica. 2020;243(1):9-20. doi:10.1159/000502387
3. Teo ZL, Tham YC, Yu M, Cheng CY, Wong TY, Sabanayagam C. Do we have enough ophthalmologists to manage vision-threatening diabetic retinopathy? A global perspective. Eye (Lond). 2020;34(7):1255-1261. doi:10.1038/s41433-020-0776-5
4. Lupidi M, Danieli L, Fruttini D, et al. Artificial intelligence in diabetic retinopathy screening: clinical assessment using handheld fundus camera in a real-life setting. Acta Diabetol. 2023;60(8):1083-1088. doi:10.1007/s00592-023-02104-0
5. Li H, Jia W, Vujosevic S, et al. Current research and future strategies for the management of vision-threatening diabetic retinopathy. Asia Pac J Ophthalmol (Phila). 2024;13(5):100109. doi:10.1016/j.apjo.2024.100109
6. Abràmoff MD, Lavin PT, Birch M, Shah N, Folk JC. Pivotal trial of an autonomous AI-based diagnostic system for detection of diabetic retinopathy in primary care offices. NPJ Digit Med. 2018;1:39. Published 2018 Aug 28. doi:10.1038/s41746-018-0040-6
7. Bellemo V, Lim ZW, Lim G, et al. Artificial intelligence using deep learning to screen for referable and vision-threatening diabetic retinopathy in Africa: a clinical validation study. Lancet Digit Health. 2019;1(1):e35-e44. doi:10.1016/S2589-7500(19)30004-4
