Liver Vessel Tree Generation Based on Skeletonization and Graph Representation

Liver vessel tree generation is of importance for liver disease diagnosis and surgery plan. In this paper, we propose a method for generation of the liver vessel tree based on skeletonization and graph representation. First of all, basic methodology in the proposed method is introduced. Secondly, the 3D-thinning skeletonization algorithm together with the graph-based technique are employed on the liver vasculature to form an undirected cyclic graph. Thirdly, a new flow conservation based approach is applied to efficiently prune the graph into a directed acyclic tree. Experiments on a variety of liver CT datasets show the effectiveness and efficiency of the proposed method.