Chinese Journal of Computers   Full Text
Title	A Variational Level Set Method of Multiphase Segmentation for 3D Images
Authors	.PAN Zhen-Kuan1) LI Hua1) WEI Wei-Bo1) GUO Zhen-Bo1) ZHANG Chun-Fen2)
Address	1)(College of Information Engineering, Qingdao University, Qingdao, Shandong 266071) 2)(School of Computer Science and Telecommunications Engineering, Jiangsu University, Zhenjiang, Jiangsu 212013)
Year	2009
Issue	No.12(2464—2474)
Abstract &   Background	Abstract The variational level set method is a new modeling method for image segmentation which has been extended to multiphase segmentation of images successfully. But the generic expressions of region/phase partitioning, image models in every region, energy functional, and their extensions to applications in segmentation of high dimensional images are current research focuses in image processing based on variational method, level set method and methods using PDEs(Partial Differential Equations). The purpose of this paper is to develop a novel variational level set method for 3D image multiphase segmentation. For this goal, a generic characteristic function for region partitioning based on n-1 level set functions for n regions and their Heaviside functions is designed, and a generic energy functional includes region-based model, edge-based model and constraint term enforcing the level set functions as signed distance functions are suggested also. The corresponding PDEs of evolution of level set functions are discretized using finite difference method and semi-implicit iterations and used to multiphase segmentation for 3D images finally. Keywords 3D reconstruction; multiphase segmentation of images; variational method; level set method; finite difference method Background The problem of multiphase image segmentation and 3D reconstruction of multiple objects in image processing and computer vision is very difficult which has been received considerable attention recent years. The 3D reconstruction of multiple organs based on 3D CT images is a part of the project of “Computational Models for Surgical Simulation” which is supported by the Program for New Century Excellent Talents in Universities of China. The purpose of this project is to develop a comprehensive simulation environment for surgical training, which include 3D reconstruction of organs, deformation computing due to interactive forces, collision detection, feedback force computation, visualization and so on. This paper focuses on developing a generic model for 3D reconstruction based on 3D images using variational level set method and a generic characteristic function formulation with strong region competition is proposed. The project has been studied 3 years in the research group, and over 20 papers on these topics have been published.