A hybrid deep learning model for detecting diabetic retinopathy

Diabetes affects large number of people all over the world and is a very common disease in India. People having diabetes are very likely to be affected by diabetic retinopathy which causes blindness. Diagnosis of this disease at an early stage can help in completely eliminating it and hence preserve the person's vision. In this paper, we propose a hybrid deep learning based approach for detection of diabetic retinopathy in fundus photographs. We use convolutional neural network with linear support vector machine to train the network on standard benchmark dataset EyePACS dataset. Experimental results show high sensitivity and specificity achieved in detecting diabetic retinopathy by our proposed model.