A Quantitative Evaluation Function for Liver Vessel Segmentation on

A Quantitative Evaluation Function for Liver Vessel
Segmentation on Contrast Enhanced CT Images
Li Cheng (A0091485X)
The segmentation and analysis of liver vessel in volumetric datasets is of vital interest for
many medical applications. Unfortunately, this analysis requires a specialist to identify
specific features which is not always possible. Automation of this process will allow the
analysis to be performed in regions where specialists are non-existent and also large scale
analysis. Some algorithms have been designed to extract the liver vessel features from CT
images. However, to date, these algorithms have been evaluated using generic image
similarity measures without take into account the important properties of liver vessels.
Motivated by this, we propose a quality evaluation function, COSB, which is based on
connectivity, overlap volume rate, skeleton length coincidence and branches diameter error
to deal with this evaluation task. The performances of different measures are tested on ten
real CT scan datasets. The results show that COSB provides the best correspondence with
human perception when compared to the other remaining tested evaluation Functions. Thus,
It is suitable evaluation function for liver vessel segmentation evaluation task.