Automatic Detection of White and Red Lesions for Diabetic Retinopathy Screening

Abstract

Exudates are the earliest and one of the most prevalent symptoms of Diabetic Retinopathy (DR), which is a serious impediment of diabetes mellitus and a main cause of blindness worldwide. Certain areas of the retina with such symptoms are to be photocoagulated by laser to stop the disease growth and prevent blindness. Outlining these areas is reliant on outlining the exudates, the blood vessels, the optic disc and the macula and the region between them. The earlier the detection of exudates in fundus images, the stronger the preserved sight level. So, early detection of exudates in fundus images is of great importance for early diagnosis and appropriate treatment. In this paper, a robust and computationally competent approach for the localization of the different features in a fundus retinal image is presented. Since many features have the common intensity properties, geometric features and correlations are used to differentiate them. First, the blood vessels are removed based on Mathematical Morphology and the exudates are segmented by using column wise neighborhood filter. We have proposed a new constraint for optic disk detection by using circular fitting method which is based on brightest point. After filtering the optic disc, the borders are detached in order to detect the exudates. We furthermore show that many of the features such as the blood vessels, exudates and microaneurysms can be detected quite accurately using this method. The performance of the proposed system is carried out on the DRIVE Database.