Semi-automatic Liver Segmentation From Computed Tomography (CT) Scans based on Deformable Surfaces

This paper presents a semi-automatic segmentation method to extract the liver volume from CT scans, based on discrete deformable surfaces with both a priori shape and intensity information. A shape model of the liver, represented as a triangular mesh, is built and is manually located in the region of interest, to be iteratively deformed in function of the image intensity and edges. In the deformation process, the distances between the vertices of the mesh are evaluated to maintain a homogeneous distribution of the surface points, while a low-pass filter is applied to assure surface smoothness. Some experimental results of the proposed segmentation method are presented to illustrate its performance.