Smart Diagnostic Analysis of Retinal Diseases

Abstract

Diabetic Retinopathy (DR) is most normal retinal disease. Diabetic Retinopathy is an eye newlinedisease caused by the microvascular complication of diabetes and it is one of the main sources newlineof vision impairment. The automatic detection and diagnosis of Diabetic Retinopathy (DR) is newlinevision and to help the ophthalmologists in mass screening of newlinediabetes sufferers. Diabetic retinopathy is a progressive eye disease and should be detected as newlineearly as possible. We proposed smart diagnostic analysis system for detection and classification newlineof various diabetic retinopathy lesions i.e. Microaneurysms and Haemorrhage (MAs and H) newlineappears as red or dark dots, while yellow or bright spots for Hard Exudates and Cotton Wool newlineSpots (HE and CWS). Several image processing techniques have used for separate finding newlinediabetic retinopathy lesions but the method can be used for smart screening of grading of newlinediabetic retinopathy with additive features on basis of abnormalities. In this thesis, we newlinesuggested another smart method in which every possible lesion present in a retinal fundus newlineimage detected by Gabor filter bank. Then feature sets are computed for each candidate lesion newlineusing different properties and features followed by classification of lesions. There are three newlinephase of the proposed system i.e. extracts all possible candidate lesions present in a fundus newlineimage using filter bank then feature sets are computed for each candidate lesion using different newlineproperties and features followed by classification of lesions. The evaluation of proposed newlinemethod is performed using retinal image standard and genuine databases with the help of newlinedifferent performance parameter and the results show the validity of proposed system. newline