ASSESSMENT OF DIABETIC MACULAR EDEMA ON OCT BASED CLASSIFICATION IN TERTIARY EYE CARE HOSPITAL

Abstract

Diabetic macular edema (DME) is the leading cause of vision-loss among diabetics, and its early detection and precise classification are crucial for effective treatment. Optical Coherence Tomography (OCT) has emerged as an essential tool in the diagnosis and monitoring of DME, as it offers noninvasive but high-resolution images of retinal structures. The purpose of this study was to assess conventional and artificial intelligence (AI)-based OCT grading for the diagnosis of DME by considering their diagnostic performance, correlation with visual acuity (VA), and their ability to predict outcomes after anti-VEGF treatment. A cohort of 150 DME diabetic patients was imaged using OCT, and the data were processed with traditional (B-scan) and artificial intelligence (CNN, U-Net) approaches. These AI models showed better results in accuracy, sensitivity, specificity, and F1-score than the conventional methods. A correlation between SD-OCT and visual acuity was also observed: patients with more severe forms of DME (CME and SRD) were observed to have lower visual acuity. Further, AI-based models did yield some encouraging results regarding prognosis and the response to treatment, especially to anti-VEGF therapy. These results indicate that AI incorporated with OCT may be able to improve DME diagnosis, treatment and personalized healthcare.