Recognition of Diabetic Retinopathy Levels Using Machine Learning

Diabetes is one of the major causes of Diabetic retinopathy (DR), it causes damage to eye retina and may lead to blindness. According to the International Diabetics Federation around 77 million cases of diabetes are registered in India in 2020. It is estimated that around 8.5% population over 18 years and around 45% population over 60 years is being affected by diabetes in India. The percentage of people being affected by the diabetes is increasing exponentially and the consequences hamper the human life. Therefore awareness and early detection are the key factors in the prevention of diabetic retinopathy. In this paper, we have presented a survey on the latest work carried by the various researchers in the domain along with their contributions. In this paper we have highlighted various research datasets available to work in this field. In this paper we present a model to predict the various class labels of the diabetic retinopathy. In the proposed paper we use data augmentation methods along with histogram equalization. Randomized Hough Transform is used to find the edges. Deep CNN model is used to classify the images. Accuracy of the classifier is around 95% with sensitivity around 90%.