Level-set based Carotid artery segmentation for Stenosis grading

Abstract

A semi-automated method is presented for the determination of the degree of stenosis of the internal carotid artery (ICA) in 3D contrast-enhanced (CE) MR angiograms. Hereto, we determined the central vessel axis (CA), which subsequently is used as an initialization for a level-set based segmentation of the stenosed carotid artery. The degree of stenosis is determined by calculating the average diameters of cross-sectional planes along the CA. For twelve ICAs the degree of stenosis was determined and correlated with the scores of two experts (NASCET criterion). The Spearman’s correlation coefficient for the proposed method was 0.96 (p<0.001), versus 0.89 and 0.88 (p<0.001) for the manual scores, and a smaller bias and tighter confidence bounds for the automated method were found.