Microaneurysms (MAs) detection using mathematical morphology

Diabetic retinopathy (DR) is one of the abnormalities on the retina. The diabetic retinopathy is a disease caused by diabetes and is considered as the major cause of blindness in working age population. The aim of this paper is to analyse the mathematical morphology namely extended minima function to detect the MAs. In this paper, the proposed methodology consists of five stages: pre-processing, detection of blood vessels, elimination of blood vessels, MAs detection and evaluation. The images were obtained from two database which are e-ophtha and ImageRet that provides the ground truth collected from several experts and a strict evaluation protocol. A performance analysis was performed based on sensitivity, specificity and accuracy of detection in retina image. The proposed method showed that the candidate of MAs has been successfully detected with average sensitivity obtained was 99.98 meanwhile 9.00 for specificity and accuracy 99.88