Diagnosis of Diabetic Retinopathy Via Classification of Fundus Images Using Deep Learning Methods

In diabetics, a condition known as diabetic retinopathy (DR) can harm the retina is a tissue located at the rear of the eye that contains blood vessels., resulting in vision loss or possibly blindness. It often affects both eyes, is more common in people who have had diabetes mellitus for a longer period of time. Early detection and appropriate treatment can help to control diabetic retinopathy, but once it has progressed past a certain point, there is no cure and no way to stop it. Thorough dilated eye exam, a labour - intensive manual procedure requiring a qualified ophthalmologist and a significant investment of time and money, is used to diagnose dry eye disease. Consequently, we are able to automatically identify various stages of diabetic retinopathy using the Fundus image with the help of convolutional neural network. In order to categorise the severity stages of diabetic retinopathy, which range from 0 (No diabetic retinopathy) to 4 (Proliferative diabetic retinopathy), we present Deep Transfer Learning classification technique utilising convolutional neural network trained model. A user interface must be made first. The second stage is to use filtering and pre-processing procedures to get rid of the extraneous elements from the image. Applying an appropriate algorithm that can identify the image and determine the degree of diabetic retinopathy severity is the third stage. The fundus image's severity degree and a detailed explanation of the presence stage are shown.