Dear SPMer's,
A quick search shows smoothing has been discussed ad nauseum on this
list but I have a question I can't find an answer to.
I'm trying to preprocess some event-related fMRI data and I've noticed
the standard procedure seems to include two smoothing steps - once
during realignment where the default setting for fMRI data is a 5 mm
kernel and then again as an entirely separate step after normalisation
where the rule-of-thumb is a 6 to 8 mm kernel.
Is it optimal or even necessary to smooth twice?
And if my original voxels are 2 x 2 x 3 with a 1 mm gap, is it better
to smooth isometrically or keep the same voxel proportions?
Ta,
Rich.
|