Spatially adaptive mixture modeling for analysis of fMRI time series

Markov chain Monte Carlo methods are finally used to simulate from the joint posterior distribution and Posterior Mean estimates are computed to summarize this posterior distribution. Figure 2 shows normalised contrasts maps of auditive sentence versus auditive calculation, where modeling of spatial correlation seems to lead more sensitive results as activations in the parietal cortex are highlighted with SSMM and USMM whereas they are not with IMM. [...]these activations are coherent with the anatomy since they follow a sulcus.