Not sure. It might be more accurate. Once you have the FWHM you want, you can run mri_glmfit-sim with --fwhm-override
On 07/17/2017 05:43 PM, Jordi Pegueroles wrote:
Dear FS experts,
I was wondering if someone had performed multiple comparisons (ie monte-carlo) with the LME analyses. I have found a thread (https://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg40195.html) where Doug recommends to run a GLM fit with the same data and then overwrite the sig.mgh file with the LME one.
My question is if is not more accurate to estimate the residuals from the LME model, then compute the relative smooth of the data with the mris_fwhm command and pass it as an argument in the mri_surfcluster command (mri_surfcluster --in $direct1/sig.mgh --csd /Applications/freesurfer/average/mult-comp-cor/fsaverage/${hemi}/cortex/fwhm${rfwhm}/abs/th13/mc-z.csd --cwsig $direct1/mc.abs.th13.sig.cluster.mgh --vwsig $direct1/abs.th13.sig.vertex.mgh --cwpvalthresh 0.05 --surf white).
Thanks in advance.