Extraction of Liver Vascular Tree from Contrast Enhanced Computed Tomography Acquisitions

Extraction of liver vascular tree of the donor liver from abdominal computer tomography angiography (CTA) images is an important step that needs to be performed before surgical operation for living donated liver transplantations. However, due to the low contrast between vessels and other tissues of liver caused by various reasons, and varying vascular tree structures for each patient makes this process highly challenging. In this paper, the process of creating ground truth data for 20 CTA datasets and encountered challenges during this process are described. Moreover some of the well-established vascular tree segmentation methods and their performances are analyzed in terms of quantitative metrics such as Dice and qualitative analysis performed by 3D analysis of a radiologist.