Your data don't necessarily need to be registered to MNI space -- they could
be averaged to a group mean, or they could be Talairached. Your data do need
to be in some sort of stereotactic space for any sort of group analysis,
however. Otherwise your results are garbage. MELODIC will assume that each
voxel represents approximately the same area of the brain in each subject. You
will be violating this assumption if you skip registration entirely.
Pretty much the only time you can run MELODIC on unregistered data is if you
are analyzing a single scan from a single subject. Even in this case, it would
be prudent to do motion correction (i.e. coregistration) before running
MELODIC.
On Thursday, April 05, 2012 10:49:24 you wrote:
> Dear All,
>
> Out of curiosity, how would the MELODIC analysis be affected if my data was
> not registered to another template and not MNI space and I skipped the
> registration ? Thank you.
>
> dominic
|