Interactive segmentation of medical images using belief propagation with level sets

In this paper, we propose an interactive segmentation method to apply user information during the segmentation of a specific anatomic structure. This method is formulated to use belief propagation to minimize a global cost function according to local level sets. The propagation starts with one user labeled point, and iteratively extends the user information from the labeled pixel to its neighborhood by calculating the beliefs of the pixels in the same level as the labeled pixel. Since the segmentation relies on both local user information and global image features, it is less interrupted by noise, and works well even the target is not obvious to its neighbor. The promising segmentation results also show that our method is robust to the objects with high shape variation and inhomogeneous intensity value appearance.