ARTIFICIAL INTELLIGENCE–ENHANCED RETINAL IMAGING IN DIABETIC RETINOPATHY: OPPORTUNITIES AND LIMITATIONS

Background: Diabetic retinopathy (DR) remains a leading cause of vision loss worldwide, making early detection and effective screening essential for preventing irreversible complications. Traditional imaging methods, such as fundus photography and optical coherence tomography (OCT), identify established retinal lesions but often fail to capture subtle early microvascular changes. Objective: This review examines current retinal imaging techniques and evaluates the integration of artificial intelligence (AI) to enhance the detection of DR. Methods: The analysis focuses on studies investigating AI applications with fundus photography, OCT, and optical coherence tomography angiography (OCTA), highlighting their diagnostic performance and potential to improve screening programs. Results: OCTA enables high-resolution, non-invasive visualization of retinal and choroidal vasculature, allowing early detection of microaneurysms, capillary non-perfusion, and other biomarkers. AI integration improves diagnostic accuracy, sensitivity, and specificity, reducing the burden on clinicians. However, clinical adoption is limited by small, homogeneous datasets, lack of standardized imaging protocols, and limited explainability of AI algorithms. Conclusion: AI-enhanced retinal imaging shows significant promise for early detection and improved management of DR. Future efforts should focus on multicenter validation, data standardization, and development of explainable AI models to enable safe, effective, and equitable implementation in routine clinical practice.