Diabetic Retinopathy Identification

In recent years, there is an increasing enthusiasm in the healthcare research community for the automated diagnosis of various health related diseases. One of the major reasons behind this is that the medical assistance and infrastructure available at hand is unable to match the exponential rate at which the diseases are proliferating. This is where technologies like Deep Learning come into play. Medical Science has various branches, one such being the field of Ophthalmology. It is a branch of medicine that primarily deals with the diagnosis and treatment of eye disorders, which if not treated appropriately and in a timely fashion may lead to loss of vision. One such leading cause of blindness is Diabetic retinopathy, also known as diabetic eye disease. It is a medical condition in which long-term poor control of diabetes mellitus causes damage to the retina. The traditional approach of diagnosis requires the doctors to analyse the colour fundus images of retina in order to determine the level of severity of disease. However, this approach is extremely timeconsuming and requires highly-experienced eye specialists for accurate analysis. But the required resources like infrastructure and skilled professionals is not only insufficient but also expensive in developing countries like India. Due to these factors, automated diagnosis is the need of the hour. Although there are existing models based on deep learning; we aim to develop it using Transfer Learning. Thus, our model intends to reduce the diagnosis time, facilitate deployable options and achieve higher accuracy.