Deep Learning Approach for Detection of Diabetic Retinopathy
Kaustubh Ratna, Akash Shedage, Raghav Agal, Bhushan Maheshwari, Amber Aggarwal, Somya Goyal
Abstract
Early Detection of Diabetic Retinopathy is paramount in prevention of vision loss. This study proposes a novel approach to the automatic detection of diabetic retinopathy (DR) using fundus images. This paper compares other techniques and does comparative study of Convolutional Neural Networks (CNNs) and the ResNet architecture to analyze fundus images and classify them based on the severity of DR. The study uses the APTOS 2019 Blindness Detection dataset, a publicly available dataset of over 5,000 high-resolution retinal images collected from patients with varying degrees of DR. The study concludes that the proposed approach provides a promising solution for the automatic detection and classification of DR.