Space-time statistical model for functional MRI image sequences

Abstract

Changes in cerebral blood oxygenation and flow during activation of human brain can be measured using functional magnetic resonance imaging (fMRI) data acquired during periodic sensory stimulation. Ideally, spatial and temporal correlations in the acquired data should all be taken into account to derive statistical parametric maps (SPM) and to identify significant changes in fMRI signal. This paper proposes a multivariate statistical model for brain activation detection accounting for both the spatial and temporal correlations. This model considers a spacetime variant error and a spatial Markov random field process is used to yield an unbiased estimate of the SPM. As the number of pixels is large enough, the asymptotic theory is used to derive a threshold allowing the identification of activated areas in the SPM. The method is illustrated on sensorimotor experiments performed on normal subjects using 1.5T gradient-echo MRI.