Activation detection on FMRI time series using hidden Markov model

This paper introduces several unsupervised learning methods for analyzing functional magnetic resonance imaging (fMRI) data based on hidden Markov model (HMM). Unlike the conventional general linear model (GLM) method, which aims at modelling the blood oxygen level-depend (BOLD) response of a voxel as a function of time, HMM approach is focused on capturing the first order statistical evolution among the samples of a voxel time series. Therefore this approach can provide a complimentary perspective of the BOLD signals. For each voxel, a two-state HMM is created, and the model parameters are estimated from the voxel time series and the stimulus paradigm. No training data is needed. Two different methods are presented in this paper. One is based on the likelihood and likelihood ratio test, and the other is based on distance measures between the two state distributions. Experimental results are presented to validate the effectiveness of our approach