A Multi-Dimension Method for Brain Tissue Segmentation

The paper presents a novel method of multi-dimension brain tissue segmentation. Firstly, the ventricle extraction and strip-skull are adopted by level set method and regions merging with tags. Secondly, the noise of the spinal fluid is eliminated through improved level set method. Finally, the gray matter and white matter are extracted through covered background method and the brain tissue is segmented. The method is generally applies to image 2D and 3D segmentation, The accuracy, commonality and practicality are proved by the results in the experiments.