Variational Level Set Method Of Liver CT Image Segmentation And Reconstruction

Multiphase segmentation of medical images and three-dimensional reconstruction are important in computer-aided diagnosis and virtual surgery. However, traditional technologies of image segmentation can not satisfy the actual demands because of the complexity and variety of medical image. The variational level set model for multiphase image segmentation on the plane is investigated. And in view of the less efficiency in calculation of the level set method, three rapid split projection methods are proposed in this paper. First of all, in the treatment of total variation, through the split method of introducing auxiliary variables, the computational complexity of nonlinear operators in the original energy functional is transformed into linear operator simple calculation. Then translate the complex constraint equation into a simple algebraic constraint equation. This avoids the complexity penalty calculation problems in the energy functional, reduces the computational complexity, and the level set properties can keep the signed distance function using the projection method in the process of evolution. And, in this paper, using the concept of the signed distance function and level set, avoids the influence of reinitialization contour line on segmentation results. By using the nonlinear anisotropic diffusion filter, treatment of liver image with complex structure, maintains the image edge features in smoothing noise at the same time. Moreover, by comparing with the traditional method, the experiments show that our algorithms presented in this paper have higher computational efficiency and accuracy, and better the extraction of liver contour. Finally, the general model of multiphase image segmentation based on the variational level set method is presented by studying the design schemes of characteristic function, formulas and calculation steps of the rapid split dual projection method, rapid split Bregman projection method and rapid split argumented Lagrangian projection method are given for multiphase image segmentation, and verified by CT images of liver tumor and three-dimensional reconstruction.