STATISTICAL-METHODS OF ESTIMATION AND INFERENCE FOR FUNCTIONAL MR IMAGE-ANALYSIS

Two questions arising in the analysis of functional magnetic resonance imaging (fMRl) data acquired during periodic sensory stimulation are: i) how to measure the experimentally determined effect in fMRI time s eries; and ii) how to decide whether an apparent effect is significant , Our approach is first to fit a time series regression model, includi ng sine and cosine terms at the [fundamental) frequency of experimenta l stimulation, by pseudogeneralized least squares (PGLS) at each pixel of an image. Sinusoidal modeling takes account of locally variable he modynamic delay and dispersion, and PGLS fitting corrects for residual or endogenous autocorrelation in fMRI time series, to yield best unbi ased estimates of the amplitudes of the sine and cosine terms at funda mental frequency; from these parameters the authors derive estimates o f experimentally determined power and its standard error. Randomizatio n testing is then used to create inferential brain activation maps (BA Ms) of pixels significantly activated by the experimental stimulus. Th e methods are illustrated by application to data acquired from normal human subjects during periodic visual and auditory stimulation.