Automation of CT-Based Liver Segmentation Is Accurate and Saves Time

The semi-automated segmentation method is based on variational interpolation, and 3-D minimal path surface segmentation. Contrast-enhanced CT images, in venous phase, were used to segment total and sub-segmental liver volumes.

The retrospective study included 41 subjects all of whom underwent liver CT for preoperative planning, and presented with pathologies such as colorectal cancer metastases, hepatocellular carcinoma, and benign liver lesions. Two image analysts made independent semi-automated segmentations, and two additional analysts performed manual segmentations. The repeatability and agreement of both methods were evaluated using Intraclass Bland–Altman analysis, and Correlation Coefficients (ICC).

The results of the study showed that semi-automated segmentation method can shorten interaction time significantly compared to manual segmentation, while still providing the same high level of repeatability and agreement.

Related Links:
University of Montreal