A Sparse Spatial Linear Regression Model for fMRI Data Analysis

In this study we present an advanced Bayesian framework for the analysis of functional Magnetic Resonance Imaging (fMRI) data that simultaneously employs both spatial and sparse properties. The basic building block of our method is the general linear model (GML) that constitute a well-known probabilistic approach for regression. By treating regression coefficients as random variables, we can apply an appropriate Gibbs distribution function in order to capture spatial constraints of fMRI time series. In the same time, sparse properties are also embedded through a RVM-based sparse prior over coefficients. The proposed scheme is described as a maximum a posteriori (MAP) approach, where the known Expectation Maximization (EM) algorithm is applied offering closed form update equations. We have demonstrated that our method produces improved performance and enhanced functional activation detection in both simulated data and real applications.