Diagnosis of Glaucoma and Cataract Disease with Deep Learning Models

Although the eye is one of the most important organs of human beings, it can also be exposed to many diseases. Some of these diseases are frequently encountered in society. Two of these are very important eye diseases known as cataracts and glaucoma. Convolutional Neural Networks (CNN) can be used to aid in the early detection and expert diagnosis of these diseases. In this study, cataract, glaucoma and normal eye fundus images have been classified using Convolutional Neural Networks, which is one of the deep learning methods. The performances of Googlenet, Densenet-201, Xception and Inception-V3 networks have been compared using Transfer Learning. Adam, rmsprop and sgdm optimization methods have been applied for each network. This study has been performed using 262 Cataracts, 200 Glaucoma and 2816 normal fundus image data in the Dataset. Images have preprocessed with resizing, background removal, random rotation and resizing. As a result of the simulations made in the Matlab environment, the best results have obtained with the Xception network architecture with rmsprop optimizer compared to other network ones.