Gang Wrote:
> No, 3dMEMA does not allow for multiple inputs from
> each subject. So the results from 3dMVM are not
> powerful enough?

I just get too many false positive voxels with 3dMVM, the correct regions are detected but also a lot of other stuff, in particular at the border of the brain, and thresholding doesn't clean it up.
The reason is most likely that these voxels have very high beta values for certain parameters (large vessels = high amplitude etc) but they also have a very high variability and poor intra-individual t-values, which 3dMEMA neatly takes into account.

Maybe a good approach would be to assess the parameters with 3dMEMA first and only pass the "useful/reasonable" ones to the 3dMVM analysis.