Identification of diabetic retinopathy using convolutional neural network

Abstract

Due to high blood sugar levels, diabetic retinopathy (DR), a complication of diabetes, affects the retina in the back of the eye. It may cause blindness if undiagnosed and mistreated. The early detection and treatment of DR are made easier by retinal screening. This paper proposes using an image-based dataset to build different convolution neural network (CNN) models to detect DR in its early stages to ease the screening procedure. The accuracy achieved was 0.9615 using the VGG model and 0.9712 using the Inception-ResNet model. This study demonstrates the effectiveness of using deep learning techniques to aid in diagnosing and predicting diabetic retinopathy.