Automatic graph-based method for classification of retinal vascular bifurcations and crossovers

Abstract

Implementing an automatic algorithm for classification of retinal vessel landmarks as bifurcation and crossovers will help the experts to analyze retinal images and detect the abnormalities of vascular topology in less time. It also can be used as the initial step of an automatic vessel classification system which is worthwhile in automatic screening programs. In this paper, we proposed a graph based method for automatic classification of vessel landmarks which consist of three steps: generating vasculature graph from centerline image, modifying the extracted graph to reduce the errors and finally classifying vessel landmarks as bifurcations and crossovers. We evaluated the proposed method by comparing the results with manually labeled images from DRIVE dataset. The average accuracy for detection of bifurcations and crossovers are 86.5% and 58.7% respectively.