Discriminate Brain States from fMRI Images Using Fuzzy Support Vector Machines

It is useful to know the sequence of hidden brain states that subjects pass through when performing some complicated tasks. In this paper, we focus on classifying n-class brain states from fMRI data using fuzzy SVMs. First, each fMRI image is processed and transformed to normalized coordinates. Then, the features are extracted, based on the activities of voxel and the index of Brodmann's areas which are used as input vectors to train the classifiers of fuzzy SVMs. The results of the study on Chinese character vs. English character, which contain six types of brain states, indicate it is feasible for either single subject brain classification or multiple human subjects’.