Dear Elisabeth,
Concerning the 2nd aspect, there's indeed a difference beween regressors, the scaling options, and the modulation. Whether to go with this or that approach is likely dependent on your particular hypotheses, so there's no simple answer. Recently I replied to a similar question on GIFT mailing list in the context of structural ICA, pointing to the most relevant papers and links, thus please see
http://sourceforge.net/p/icatb/mailman/message/33813970/
http://sourceforge.net/p/icatb/mailman/message/33816379/
Another relevant paper would be http://www.ncbi.nlm.nih.gov/pubmed/21684724 , some recent papers http://www.ncbi.nlm.nih.gov/pubmed/25339897 and http://www.ncbi.nlm.nih.gov/pubmed/25414635 . There are also several papers which focus on how to measure ICV / evaluation of different algorithms (e.g. based on Freesurfer vs. SPM).
Best,
Helmut
|