Prediction of Diabetic Retinopathy Using Deep Learning

Diabetic patients who have Diabetic Retinopathy (DR), a retinal condition, are at a higher risk of going blind. The hazards can be decreased by early discovery and appropriate treatment. With image segmentation, feature extraction, and binary classification, an autonomous diabetic retinopathy detector has been suggested. Deep learning techniques like CNN and ResNet is used to implement all functions relevant to the automatic detection of diabetic retinopathy. This paper suggests a technique which is intended for drawing out of Blood Vessels from the Medical Image of Human Eye-Retinal Fundus which finds its application in Ophthalmology in Detecting DR. The outcome is that DR has been predicted in the affected fundus image and the DR is not predicted in the healthy fundus image with 93% of accurateness.