Automatic Detection of Coronary Vessels Using Mutli-scale Texture Dictionaries

Abstract

In this paper we present a new automatic method for coronary artery vessel detection. We employ a texture modelling approach based on image textons as texture features, in the context of a classification experiment, where we attempt to discriminate between vessel and non-vessel like shapes in X-ray angiogram images. Experiments were conducted on a real patient database. The results show that the proposed model can perform well and distinguish vessel areas from others in an efficient manner, and outperforms other existing methods.