Classifier of BOLD signals from active and inactive brain states using FCM clustering and evolutionary algorithms

Abstract

A fuzzy classifier based on the fuzzy c-means (FCM) clustering has shown a decisive generalization ability in classification. This paper reports a result of test on a data set with high-dimensional feature values. For classifying the blood oxygen level dependent (BOLD) responses of the brain, a way of directly handling high-dimensional fMRI signals is applied. Our goal is to distinguish the BOLD responses to recalling tasks from those to resting (i.e., a binary classification problem). We use the signals from wide areas of the brain, which forms a set of high dimensional data vectors. The FCM classifier is compared with support vector machine (SVM). SVM is a high performance classifier and well suited for binary classification problems, since the size of the kernel matrix of SVM depends only on the number of instances. The error rate on the test set by the FCM classifier surpassed the SVM, though the SVM can easily handle sets of high dimensional feature vectors.