Diabetic Retinopathy Classification using Deep Learning Techniques

Abstract

Diabetic Retinopathy is a disease that damages the eyes and is caused by a consequence of diabetes. If blood sugar levels aren't controlled for an extended period of time, the disease can develop. It is mainly caused due to the damage of blood vessels in the retina. Retinopathy is the main cause of blindness in the world. Doctors can diagnose blindness before it occurs using Artificial Intelligence and Deep Learning. Medical imaging plays a very crucial role in a variety of medical issues and at all major levels of health issues. Medical imaging can be used to identify a variety of common eye illnesses. However, for a variety of reasons, including uneven lighting, picture blurring, and low contrast and brightness, poor-quality retinal images are ineffective for further diagnosis, particularly in automated analyzing systems. Ophthalmologists' manual Diabetic Retinopathy diagnostic procedure is time-consuming, requires more work, costly, and might result in misdiagnosis. Basing on the vision like having trouble in reading distant objects or seeing distant objects, blindness or any other changes may happen in eye retina that affects diabetes. Diabetic retinopathy is one of the most frequent eye illnesses, affecting mostly diabetics. This model can assist the opthmologists for clinical diagnosis and detect and classify the diabetic retinopathy. There are three phases in this diabetic retinopathy detection and classification technique (i) enhancement (ii) Feature Extraction and (iii) Retinopathy Detection and Classification. Feature extraction involves blood vessels extraction and exudates extraction. First two phases assist the opthmologists by providing clear images of the retina and blood vessels and exudates extracted images. In this work, from the presented retinal fundus pictures, the Res-Block model is used to classify and diagnose diabetic retinopathy.