Detection of Diabetic Retinopathy with Integration of Deep Learning

Diabetic Retinopathy (DR) is a disease of retina. If not diagnosed at an early stage, it causes blindness. To detect this disease some manual classification methods are used. However, these methods are highly time consuming. Also inter observer variability is another problem to deal with. Hence, computer automated diagnostic systems are preferred for retina images classification. In this study, it is aimed to classify DR existence and its level using transfer learning for Convolutional Neural Network (CNN) using the data of Kaggle and Messidor-2 datasets which contain five-classes and four-classes, respectively. The classification accuracy was obtained as 74.72% in Kaggle dataset for normal, mild, moderate, severe and end-stage classes.