300. RudzkiM

The paper presents an automatic method for liver vasculature segmentation in volumetric images obtained during Computed Tomography (CT) patient examination. The method, apart from the CT image, also requires an additional image containing binary liver mask. A modified multiscale vesselness filter (VF) performs vasculature detection. VF response is used to determine the starting voxels for the segmentation and to calculate parameters characterizing detected blood vessels. The segmentation is done using region growing algorithm with dynamic thresholds calculated for each voxel according to the VF response. Testing was done on a set of 40 CT liver studies in portal phase. The images were of various voxel size and liver-vessels contrast. Obtained results are following: in one case the vessel tree could not be detected, in 4 cases the segmentation was of poor quality. In the remaining 35 cases the vessel tree was segmented above: 1st bifurcation (2 cases), 2nd bifurcation (11 cases) and 3rd bifurcation (22 cases).