Automatic Detection of Dark and Bright lesions using Segmented retinal Images (P-0588) (MFN 4786)

Abstract

The main purpose of our system is to help detect blindness in Diabetic patients. Diabetes is one of most commonly occurring disease and Diabetic retinopathy is one of the leading causes of blindness in diabetic patients. In Diabetic retinopathy the blood vessels present in the eye are affected as a result of chronic diabetes. In early stages fragile blood vessels in retina are further weakened. This results in the blood vessels leaking various fluids, lipids, proteins and blood on the retina. With these spots on the retina the vision becomes blurred with dark spots in between. If the disease is left untreated at this stage it progresses further and enters later stage. In this stage new fragile vessels are formed to replace older ones however they can also rupture leading to complete blindness. Using Image Processing Techniques our System identifies the various characteristics of these lesions and then using various machine learning algorithms it classifies them. Our System will assist a trained ophthalmologist in the identification of the disease.