Automated Detection of Diabetic Retinopathy using Deep Learning in Retinal Fundus Images: Analysis

Abstract

Diabetes frequently results in diabetic retinopathy (DR), which develops retinal lesions that impair vision. It can cause blindness if not caught in time. Treatments simply maintain vision because DR is an irreversible process. The risk of blindness can be considerably decreased with early DR detection and treat- ment. The traditional diagnosis of DR retinal fundus pictures by an ophthalmologist is time-consuming, labor-intensive, expensive, and prone to error in comparison to computer-assisted diagnostic techniques Recent advances in deep learning have propelled them to the top of the list of the most widely used methods. Deep learning is particularly effective at classifying and analysing medical images.Convolutional neural networks, a more common and effective deep learning technique, handle medical images very effectively. The proposed model makes use of Inception V3which is Convolutional Neural Network that provides accuracy of 93% which is the highest accuracy when compared to AlexNet, DenseNet121, RestNet50 and EfficientnetBO in detecting diabetic retinopathy.