Automatic Diagnosis of Non-proliferative Diabetic Retinopathy Using Artificial Neural Network

Abstract

Diabetic retinopathy is an eye disease caused by diabetes mellitus which affects the retina. It leads the retina blood vessels to swell, these damages the retina and may lead to blindness if it does not early and accurately detect. The most effective treatment is early detection through regular screenings, but fundus images are poor quality to be diagnostic. In this research, we present an automatic diagnosis system to classify in which stage the non-proliferative diabetic retinopathy using artificial neural network. The grading of the severity level of DR is based on detecting and analyzing the early clinical signs associated with the disease, such as microaneurysms’ proposed method consists of five stages: pre-processing, segmentation of optic disk, detection of MAs, feature extraction and classification. Mathematical morphology operation is used for pre-processing. K-mean technique used for segmentation of MAs, Artificial neural network is used for classification of the disease stage. A database of 261 color images are used in order to evaluate the performance of the developed system. The system achieves 99.0% of sensitivity ,95.833% of specificity and 98.18% of accuracy.