A geodesic selection based variational model for 3D liver segmentation

The segmentation of liver, especially the abnormal liver with tumors, is a challenging task. Due to the presence of tumor and intensity inhomogeneity in CT scans, the liver consists of multiple sub-regions with distinct appearances. Moreover, the liver and the tumor share the similar intensity and ambiguous boundaries with neighboring structures/organs, which make the separating surfaces ambiguous. Under the framework of active contour, we construct a joint model of delineating the whole liver with a single level set representation. The final result and computational efficiency is improved compared to segmenting the parts independently. The level set evolution on the multiple regions is guided to the desired boundary by a novel geodesic selection scheme. As a result, in each sub-region the most relevant appearance and boundary knowledge are automatically used. Besides, a weighted histogram for the local appearance description is introduced for the precise boundary detection. A quantitative and comparative performance assessment is carried out over a publicly available benchmark with competitive results.