Foremost section study and kernel support vector machine through brain images classifier

Automated and accurate classification of MR brain imagesis extremely important for medical analysis and interpretation. Overthe last decade numerous methods have already been proposed. In this paper, we presented a novel method to classify a givenMR brain image as normal or abnormal. The proposed method first employed wavelet transform to extract features from images, followed by applying principle component analysis (PCA) to reducethe dimensions of features. The reduced features were submittedto a kernel support vector machine (KSVM). The strategy of K-fold stratified cross validation was used to enhance generalization of KSVM. We chose seven common brain diseases (glioma, meningioma, Alzheimer's disease, Alzheimer's disease plus visual agnosia, Pick'sdisease, sarcoma, and Huntington's disease) as abnormal brains, andcollected 160 MR brain images (20 normal and 140 abnormal) fromHarvard Medical School website. We performed our proposed methodswith four different kernels, and found that the GRB kernel achievesthe highest classification accuracy as 99.38%. The LIN, HPOL, and IPOL kernel achieves 95%, 96.88%, and 98. 12%, respectively. We alsocompared our method to those from literatures in the last decade, and the results showed our DWT+PCA+KSVM with GRB kernelstill achieved the best accurate classification results. The averagedprocessing time for a 256£256 size image on a laptop of P4 IBM with3 GHz processor and 2GB RAM is 0.0448 s. From the experimentaldata, our method was effective and rapid. It could be applied to thefield of MR brain image classification and can assist the doctors todiagnose where a patient is normal or abnormal to certain degrees.