Bayesian spatiotemporal inference in functional magnetic resonance imaging

Mapping of the human brain by means of functional magnetic resonance imagin g (fMRI) is an emerging held in cognitive and clinical neuroscience. Curren t techniques to detect activated areas of the brain mostly proceed in two s teps. First, conventional methods of correlation. regression, and time seri es analysis are used to assess activation by a separate, pixelwise comparis on of the fMRI signal time courses to the reference function of a presented stimulus. Spatial aspects caused by correlations between neighboring pixel s are considered in a separate second step, if at all. The aim of this arti cle is to present hierarchical Bayesian approaches that allow one to simult aneously incorporate temporal and spatial dependencies between pixels direc tly in the model formulation. For reasons of computational feasibility, mod els have to be comparatively parsimonious, without oversimplifying. We intr oduce parametric and semiparametric spatial and spatiotemporal models that proved appropriate and illustrate their performance applied to visual fMRI data.