Effective detection of retinal exudates in fundus images

Abstract

Diabetic-related eye diseases are the most common cause of blindness in the world. Early detection through regular screenings is the most effective treatment for these eye diseases. To improve the efficiency of such screenings, it is very important that effectively finding the presence of abnormalities in the retinal images captured during the screenings. In this paper, it is focused on automatically detecting one of the abnormal signs: the presence of exudates/lesions in the retinal images. A novel approach that combines median filtering and dynamic clustering analysis is proposed. Experimental results indicate that the new algorithm is easier, faster and more effective for lesion detection from retinal images of various qualities. ©2009 IEEE.