This is not the correct procedure. Try looking at the multi-modal integration tutorial on our web site. Basically you should not register the difference image. Instead, register the middle time point of the raw data of the fMRI to the anatomical. Motion correct the time series to the middle time point. When you analyze the data in whatever way you are using, analyze the motion corrected data. You can then use the registration computed from the raw data for anything derived from the motion corrected data. doug

francesca strappini wrote:

Yes, It's a whole brain acquisition and the functional and the anatomical data come from the same subject. I try to explain all the story. I acquired 12 runs of polarangle per subject in two different days. I have 3 conditions and 4 runs for every condition. In each day the subjects did all the conditions (6 runs per day). I calculated a sort of subtraction between these conditions, unfortunately it's not a simple subtraction and for this reason I didn't use fscalc. What I tried to do is to register this subtraction to the surface but the process failed probably because the result of the subtraction is as very small activation only in the occipital cortex (in fact the conditions are slightly different). Do you think that is there something else that I can try?

Thanks!

Francesca