Tracking of Cerebral Arterial Segment by Particle Filtering in CT Angiography

In this paper a method to extract a segment of cerebral arteries from CT angiography (CTA) is proposed. Since the CTA shows both bone and vessels, the examination of vessels is a challenging task. It is also very difficult to separate out an arterial segment not corrupted by vein contamination. Since a vessel segment is considered as an ellipse travelling in three dimensional (3-D) space, its extraction is taken as tracking of the ellipse in spatial sequence and is solved by the application of a particle filter. The proposed tracking method is equipped with adaptive properties to both bone contact and vein contamination, leading to less user intervention. The experimental results on actual clinical data showed its improved robustness compared to the conventional vessel tracking methods and confirmed its feasibility.