DIABETIC RETINOPATHY DIAGNOSIS USING RESNET

Abstract: Diabetic Retinopathy (DR) is a leading cause of preventable blindness globally, characterized by damage to the blood vessels in the retina. Early detection and timely intervention are critical to preserving vision; however, manual screening of retinal fundus images is time-consuming and prone to human error. In this paper, we propose a highly accurate automated diagnostic system utilizing a ResNet architecture for the efficient detection of Diabetic Retinopathy. Our approach processes and classifies retinal images, distinguishing between healthy eyes and those affected by DR. The proposed model was trained and rigorously evaluated, achieving a remarkable training accuracy of 97.3% and a testing accuracy of 94.6%. The results substantiate that our proposed deep learning framework not only offers superior diagnostic accuracy but also minimizes false negatives, making it a robust and scalable solution to assist ophthalmologists in clinical diagnostic workflows and early screening of Diabetic Retinopathy.

Keywords: Diabetic Retinopathy, Deep Learning, ResNet, Medical Image Classification, Retinal Fundus Images, Automated Diagnosis.