Segmentation of Carotid Arteries in CTA Images

Abstract

Stenos is of the internal carotid artery (ICA) is implicated in approximately one quarter of stroke cases. The degree of stenos is is currently used to decide whether to undertake a surgical procedure to reduce the risk of further stroke. However it is known that the degree of stenos is is not a good predictor of stroke risk. It is hoped that prediction might be improved by incorporation of other geometric factors. This paper describes a data driven approach using classical methods from the field of mathematical morphology to automatically segment the carotid artery tree in computed tomography angiography (CTA) images following user initialization. The resulting segmentation may be used to characterize the the arterial geometery in a variety of more complex ways than is possible using manual approaches.