Intra-hepatic vessel segmentation and classification in multi-phase CT using optimized graph cuts

Abstract

The segmentation and classification of the major intra-hepatic blood vessels are critical for the robust identification of the segmental anatomy of the liver. We propose a novel 4D graph-based method to segment and label the hepatic and portal veins. The algorithm uses multi-phase CT images to model the differential enhancement of the liver structures and Hessian-based vesselness likelihood to avoid the common pitfalls of graph cuts-based intra-hepatic vessel segmentation. A hybrid classification step identifies the right, middle and left hepatic, and portal veins. We tested the method on CT data from nine patients and comparatively found that the new vesselness and enhancement graph costs are effective in reducing the effects of heterogeneous noise and vessel fragmentation.