Diabetic Retinopathy Classification using Deep Learning Technique

Diabetic retinopathy is condition of the eye which leads to damage of the eye's blood vessels. if not identified early enough, DR can result in impaired vision or even blindness. It is common eye condition that affects diabetics. The severity of DR is automatically determined in this paper. Non-proliferative diabetics retinopathy is the starting stage of diabetic retinopathy, is characterized by red lesion, cotton-wool patches, and exudates. There are five stages of diabetic retinopathy: mild, moderate, severe, proliferative, and no disease. However, when the illness progresses to the level of proliferative diabetic retinopathy. The growth of scar tissue and new blood vessel bleeding begins, and total vision loss is unavoidable. In our method using different pre-processing flow and using convolution neural network of LeNet Architecture with ADAM optimizer gives the better results for classifying the DR types.