Published Paper Details:

Diabetes Mellitus is a metabolic disease chronic in nature involving millions across the globe and may lead to complications that are critical in nature like Diabetic Retinopathy (DR), an illness which causes permanent vision impairment. Prompt detection and precise DR classification are indispensable for effective therapy and prevention of blindness. Manual techniques employed till now for DR are tedious, with potential for variability, so motivating the implementation of computer-assisted techniques in medical imaging. This work provides an extensive review of the recent developments in the era of deep learning, including Convolutional Neural Networks (CNNs), for detecting and classifying DR from retinal fundus images. We review publicly accessible datasets and discuss a range of deep learning models applied in this area. The primary contributions of this study are: (1) providing state-of-the-art deep learning methods for DR diagnosis, (2) comparing their performance on standard datasets, and (3) outlining current shortcomings and possible directions for future work in automated retinal image analysis.