Hard Exudates Detection for Diabetic Retinopathy Early Diagnosis Using Deep Learning

Diabetic retinopathy is the complication of the eye caused due to diabetes mellitus. It is caused due to high glucose level in blood which damages the blood vessels at the back of the eye namely retina. The abnormal blood vessels swell and leak into retina. The microstructures such as microaneurysm, exudates (hard/soft) will occupy the retina area which leads to vision threatening. Nowadays, in medical field, computer-aided systems are performing a promising work in terms of object recognizing, localization, classification, segmentation and analysis of images with the help of deep neural networks. Since diabetic retinopathy is irreversible, detection of early stages of diabetic retinopathy is needed. If it left without proper diagnosis and treatment, itwill lead to vision loss. Detection of hard exudates indicates the presence of diabetic retinopathy. The early sign of diabetic retinopathy is the formation of hard exudates in retina. In this method, the main aim is to detect the presence of hard exudates in the retinal image using deep convolution neural network.