Paper Title
A Hybrid Method Based On Svm And Morphological Segmentation For Classifying Mri Brain Image

Recently, the global population ageing problem is becoming more and more important, and brain degeneration disease is the major disease pattern for ageing people. For detecting brain disease, magnetic resonance (MR) imaging is one of most advanced medical imaging methods. Further, automatic and semi-automatic brain MR images segmentation has been commonly used in medical image processing and classification. There are two major drawbacks in segmentation methods: (1) automatic segmentation is limited to the treatment of some certain types of images, (2) single segmentation method has its disadvantage and restriction on segmentation. In order to overcome the drawbacks, this study proposed a hybrid semi-automatic segmentation method which includes three stages: (1) the regions of interest of brain were segmented out by segmentation algorithm, (2) deconstruct and calculate the feature value by using wavelet transform, and (3) use classification algorithm to classify images into normal and abnormal brain images. Finally, the experimental results show that the proposed method is superior to the listing methods. Keywords- Magnetic resonance imaging, Wavelet packet transform, Support vector machine;