The segmentation of liver and vessels in CT images using 3D hierarchical seeded region growing
Zhanpeng Huang, Faling Yi, Shizhong Jiang, Zhao Jie, Bao Su-su · 2011
Seeded region growing (SRG) is becoming a popular method because of its ability to involve high-level knowledge of anatomical structures in seed selection processes. As medical images are mostly fuzzy, defining the homogeneity criterion depending on the image properties is a challenging task. We developed a novel 3D hierarchical SRG algorithm which learns its homogeneity criterion automatically. In our approach, several seed points are selected firstly. Then the CT images are divided into sub-blocks with a defined size and the homogeneity criterion is estimated automatically through investigation of the statistical characteristics in the local regions of the seed points. In order to utilize the texture of the liver image and reduce computation cost, the hierarchical sub-blocks merging techniques are used during the region growing procedure. Experiments results show the proposed method can efficiently segment the live region and vessels from serial abdominal CT images with little user interaction.