CNN Based Diabetic Retinopathy Status Prediction Using Fundus Images

Abstract

One of the most diabetes complication is Diabetic Retinopathy (DR) that causes major loss of vision or blindness. In present day medical science, estimation of images has become key instrument for exact identification of disease. So we have designed a computational model for predicting Diabetic Retinopathy (DR) status which is based on retinal image and neural network. Our computational model has been consisting of a feature extraction phase and a classification phase. In feature extraction phase we have extracted the most appropriate features from digital fundus images by Blood Vessels and Micro aneurysms detection. For this research work we have used Diabetic Retinopathy dataset provided by Kaggle Community. Finally, we have used CNN to predict the Diabetic Retinopathy (DR). In our proposed methodology, we have achieved 95.41% accuracy.