Dear Miao,
With the default settings some parts of the cerebellum will end up in the analysis indeed, but there's no real need to mask them, we just accept it's there even if the main focus is on cerebrum.
In case you really want to exclude the cerebellum (e.g. to ensure that differences in cerebellum don't affect occipito-temporal regions) then you would have to mask the cerebellum well before on the individual unsmoothed w(m)c1 files. This would be time-consuming if you do so manually. But you could try to isolate cerebellum and brainstem with SUIT http://www.icn.ucl.ac.uk/motorcontrol/imaging/suit.htm for each of your individual T1 files and mask these regions, then forward these adjusted T1 files into your VBM preprocessing pipeline. However, to make this meaningful you should also mask cerebellar regions in the tissue probability maps provided with SPM. In principle, this should also be possible with SUIT, but I think it's tuned to work on T1 images, so you might need to adjust some parameters.
Best
Helmut
|