Diabetic Retinopathy Detection Using Retinal Images

Abstract

Diabetic Retinopathy is an ocular condition that affects individuals with diabetes. Elevated blood sugar levels harm ocular blood vessels and can possibly result in blindness. Exudates, which are brilliant yellow lesions, and red patches called microaneurysms are indicators of diabetic retinopathy. This research suggests a straightforward and successful method for treating diabetic retinopathy. It has been noted that early diagnosis of exudates and microaneurysms may preserve the patient's vision. The empirical investigation has made use of real-time and publicly accessible datasets of colored fundus camera images. Exudates and microaneurysms in the fundus pictures have been used in the proposed study to grade the severity of diabetic retinopathy, i.e., whether it is mild, moderate, or severe. A potential automated method makes use of features extraction, image processing, and machine learning models to precisely forecast the existence of exudates and microaneurysms, which can be utilized for grading. There are two sections to the research: one focuses on exudates, and the other on microaneurysms. While microaneurysms are graded by counting them, exudates are graded according to how far out from the macula they exhibit. Support vector machines had the maximum accuracy of 92.1% when grading using exudates, and when grading using microaneurysmsIn terms of predicting the disease's severity levels, decision trees had the best accuracy—99.9%.