Investigation on retinal fundus images for early detection of diabetic retinopathy and classification using adaptive neuro fuzzy interference system

Abstract

Diabetic Retinopathy (DR) is a primary cause of sightlessness and there exist valuable behaviors that pull back the development of the disease given that it might be recognized in the initial stage. But DR is classically asymptomatic in its starting stage and so the diabetic patients cannot endure any eye test awaiting it is formerly too late for the finest possible treatment and harsh retinal damages have been formed. A proper method to enhance the image description for consequent segmentation and feature extraction is vital. The research is segmented into three phases: pre-processing phase for feature enhancement and segmentation; detection and classification phase that categorizes the images as healthy or abnormal images of Diabetic Retinopathy (DR); Classification of Non Proliferative Diabetic Retinopathy (NPDR)into mild, severe or moderate level; train the images using Artificial Neural Network (ANN) and Artificial Neuro Fuzzy Interference System(ANFIS) and results presented after analysis and comparison with various methods. newline