In MR image, the boundary of each encephalic tissue is very complicated and irregular. As a new kind of machine learning, support vector machine (SVM) based on Statistical Learning Theory has high generalization ability, especially for dataset with small number of samples in high dimensional space. But it is not very suitable for classification if the sample number is unbalance. In this paper, a weighted support vector machine (W-SVM) method was proposed for MR image segmentation. The irregular boundary of each encephalic tissue is extracted successfully. The experimental result shows that the segmentation performance based on W-SVM is effective and surpasses the segmentation performance based on SVM only.
展开▼
