Vasculature segmentation of CT liver images using graph cuts and graph-based analysis

Extracting hepatic vasculature from 3D imagery is important for diagnosis of liver disease and planning liver surgery. In this paper we propose to segment vessels from liver CT images using a 3D graph-cuts based method that utilizes probabilistic intensity information and surface smoothness as constraints. A semi-automatic graph-based technique is then employed to efficiently separate the hepatic vessel systems. The complete vascular analysis method has been evaluated on 6 liver CT datasets using manual segmentation as the reference and showing that the method is reasonable robust to parameter choice and gives results of similar accuracy to previous methods in a time-efficient manner.