A Novel Approach for the Screening and the Classification of Macular Ischemia Caused by Diabetic Retinopathy Disease Using Retinal Image Datasets

Abstract

Diabetic Retinopathy (DR) is the retinal disease which is caused due to change in the micro retinal blood vascular structure. It may cause change in the diameter or the blockage of the blood vascular structure which leads to the occurrence of serious macular ischemia disease. The early detection of the blockage area in the vascular system is necessary otherwise it may result in loss of vision. The ophthalmologist screening of the retinal images differ from person to person and hence there requires the need of the automated system program for the analysis and classification of the fundus images. Here an automated system is proposed for the detection and the classification of the Macular Ischemia based on a graph trace method for automatic A/V (Arteries and Veins) classification by extracting the graph from the segmented vascular structure. The A/V classification is done by measuring the intensity features of the segmented blood vessels. KNN classifier is then used to extract the thickness feature of the vascular structure to allocate the blockage of the veins in the retinal blood vessels. For Grading of the disease severity, SVM Algorithm is use. In this methodology, the concept of evaluating two database at the same time is used in order to reduce the computational time i.e., two patients fundus images can be evaluated at the same time. The performance of the proposed system is evaluated or calculated by testing it on Drive database using parameter such as sensitivity, specificity and accuracy.