Liver Extraction based on blood vessel using multislice CT datasets

摘要

At liver transplantation planning stage, liver volumetry extraction would be helpful for surgeons. The main problem at liver extraction is wide variability of anatomic structures of liver in shape and size. to overcome that problem, we propose a method based on hepatic blood vessel as it defers from a person to another as well. in this method we used CT datasets at portal venous phase to extract portal and hepatic veins. then we construct polyhedron structure of liver using endpoints of extracted vessels. then, we extended that region based on liver map and volume of extracted region during dilation. to stop dilation, we draw tree structure of increased area. when it changes direction from increasing to decreasing, that time we stop dilation to get final liver shape. results of that method agree with manual labeling done by an expert and reviewed by a medical doctor at rate 97.04%. this method seems promising to extract various liver shapes and sizes.