I don't categorize myself as an expert, but I had the same question a
few years ago.
The smoothness estimates in SPM.xVol.FWHM are in units of VOXELS, so
you need to multiply by the voxel dimensions to get them in mm.
Something like this:
M = SPM.xVol.M;
VOX = sqrt(diag(M(1:3,1:3)'*M(1:3,1:3)))';
FWHM = SPM.xVol.FWHM;
The calculations of smoothness are carried out by spm_est_smoothness
and are based on Gaussian random field theory. They will not match the
3dFWHMx results exactly, but should be close. In the May 2011 spm
archives there is a comparison of 3DFWHMx smoothness and the result of
SPM.xVol.FWHM, and they got the results below.
Very best wishes!
FWHMx = 9.82782
FWHMy = 10.0744
FWHMz = 9.31861
FWHMx = 10.2832
FWHMy = 10.3741
FWHMz = 9.4135
On 8/27/13, Umi V. <[log in to unmask]> wrote:
> Dear Experts,
> I am a relatively new user of SPM and have been unable to find a clear
> answer to the question of what is contained in the SPM.xVol.FWHM file. I
> would like to obtain the intrinsic smoothness of the data (not the
> statistical map) to feed into 3dClustSim of the AFNI package. I know for
> AFNI users, this is usually the output of 3dFWHMx, which gives the average
> smoothness of the residual time series across all participants. When I look
> at the SPM.xVol.FWHM at the group-level for a specific contrast, is this
> number the SPM equivalent to AFNI's FWHM output? If not, what exactly do the
> numbers in the file represent? I tried running 3dFWHMx on each type of map
> (con, spmT, ResMS) and was not able to replicate the values contained in
> Any help you can provide is highly appreciated!