Automatic segmentation for medical image with the optimized tree structured part model

Organ disease, such as liver and spleen, is the common disease with high morbidity worldwide, and the operative therapy is one of the major method for the organ disease therapy. The computer assisted surgery before the operation has the instructive effect on the clinical therapy, disease diagnosis, and surgical planning. This paper presents the optimized tree structured part model for automatic organ segmentation. The Optimized Tree Structured Part model (OTSPM) contains two parts. The first part uses the structure to discriminatively capture the topological shape variation. The other part is used to get the local part feature. For liver segmentation, the paper propose a convex concave point (CCP) method to automatically choose the most salient point to represent the local part feature, which explicitly describes the partial structure. Compared with the traditional shape model method, this improved method can get better organ segmentation effect. The model can be effectively applied to organ segmentation and it also can get high accuracy than traditional model.