Diagnosis of Diabetic Retinopathy from Retinal Images

Abstract

Diabetic retinopathy is the condition where the retina is damaged due to leaking fluid from the blood vessels into the retina which causes blindness. Fundus Camera is used to acquire the retinal images based on monocular indirect ophthalmoscopy. The main aim of the project is to detect the abnormalities like microaneurysm and exudates from retinal fundus images and to classify the severity based on the features extraction. The diagnosis of diabetic retinopathy is achieved by preprocessing the input image and by the segmentation of microaneurysm and exudates. The detection of microaneurysm is done by edge detection and exudates using thresholding technique. The optic disc is eliminated from the segmented image by the creation of binary mask. The features like correlation, energy, entropy, sum average, cluster shade, maximum probability, Inverse Difference Normalized (IDN), Inverse Difference Moment Normalized (IDMN), homogeneity, contrast are extracted from the gray scale image using Gray Level Co-occurrence Matrix (GLCM). The classification of severity is accomplished by Support Vector Machine (SVM) algorithm.