Hi all,
Please can someone confirm whether or not SPM (5 or any earlier
version) takes into account non-stationary smoothness or roughness of
residuals?
A couple of off-list replies I had to my message regarding FWE for VBM
(http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0611&L=spm&P=31993 )
seem to suggest that SPM might use a local estimate of roughness in
order to perform the FWE Random Field correction.
Now, it's true that spm_est_smoothness does estimate roughness; an
image of the local resels-per-voxel measure is stored as RPV.img and a
volume-handle to this is stored in SPM.xVol.VRpv.
However, neither this image, nor the handle to it, appear to be
references anywhere else in the SPM5 codebase -- at least not as far
as I can tell with e.g.
# find spm5/ -name '*.m' -exec grep -H -e SPM.xVol.VRpv '{}' ';'
So I have two questions: am I right in thinking RPV is ignored, in
favour of a stationary/global FWHM estimate? And if so, why?
Many thanks,
Ged.
|