An automatic approach to segment retinal blood vessels and its separation into arteries/veins

Abstract

The retinal fundus image consists of blood vessels which are further classified as arteries and veins. The measurement of retinal microvasculature changes by classifying arteries and veins using image processing opens window to find biomarkers and gives signs related to diabetic retinopathy, hypertensive retinopathy, hyperglycemia and blood pressure, etc. The purpose of this paper is to find major vessels in retinal image and automatically distinguish them into artery and vein. This paper gives an automated approach for artery-vein classification by analyzing graphical vasculature tree extracted from retinal image. Here, the proposed method distinguish the graphical retinal network by classifying each graphical node as end point, intersection point, and separate point node furthermore labeling each graphical links as artery or vein. Finally, artery-vein classification is performed on the basis of structural as well as intensity-based features. We have tested results of this method on publically available DRIVE database.