Publication

Abstract : Diabetic Retinopathy is a diabetes disablement that affects eyes. According to the WHO information, by the year 2020/21, more than one billion people will experience visual impairment or blindness, and nearly one billion people with diabetes. Computer-aided diagnosis (CAD) tools have the potential support for the ophthalmologist to diagnose sight-threatening diseases such as cataract, glaucoma, diabetic retinopathy etc. The purpose of the proposed work is to detect diabetic retinopathy, where we aimed to diagnose using clinical imaging that incorporate the use of deep learning in classifying full-scale Diabetic Retinopathy in retinal fundus images of patients with diabetes. A comparative analysis is done with various deep learning models like CNN, MobileNetv2, ResNet50, Inceptionv2, VGG-16, VGG-19, and Dense Net and the best model is proposed which is used to make predictions and attain accuracy using lesser number of images. Automatic detection with more accuracy will make screening for retinal diseases as cost effective and efficient and can prevent eye disorders in the earlier stage.