Research Of MR Brain Image Segmentation Based On Support Vector Machine

Medical image segmentation is one of the hotspots in the research of medical image. Although current segmentation methods have achieved a certain effect, the previous segmentation techniques are mostly based on the traditional methods, and they are based on the asymptotic theory when the number of the samples tends to infinity. Least Squares Support Vector Machine(LS-SVM) theory has been developed rapidly, which has been applied in image segmentation and obtained satisfactory results.LS-SVM is based on the theory of VC dimension of statistical learning theory and the principle of structural risk minimization. According to the limited information of the samples, LS-SVM finds the best compromise between model complexity and learning ability to obtain the best generalization ability. In this article, LS-SVM is applied in the image segmentation of MR brain images.MR brain images are from McGill University Brain Web with simulation of different noise levels. Experiments show that the application of LS-SVM in MR brain image segmentation can get good result.During the process of feature extraction, we take the average of the two kinds of regional pixels as Eigen value. Firstly, the gray-scale features are calculated while texture features are calculated in 5×5 circular area and 7×7 expanding. Secondly, this method takes the average of the gray-scale and texture features as support for eigenvector. Lastly, classification model is generated by using these eigenvector. We apply this classification model to segment the MR image.Experiments of different numbers of training samples validated that the method based on LS-SVM is better while training samples number is small. Traditional statistical learning methods find the optimal value when the number of samples tends to infinity, while the LS-SVM obtain the optimal solution under the limited information of the samples, so their generalization ability is better than the traditional learning methods. Result of performance comparing between LS-SVM and other segmentation methods shows the advantage of LS-SVM based image segmentation method.