When the GLM for a group of subjects is computed using mri_glmfit, a residual FWHM value is computed and stored in the file 'FWHM.dat'.
When running a subsequent mri_glmfit-sim, using a pre-cached Monte-Carlo Z (mc-z) simulation for clusterwise correction for multiple comparisons, there appear to be separate simulations for a range of FWHM values stored (for each hemisphere, lh and rh) in /freesurfer/average/mult-comp-cor/fsaverage/?h/cortex/ from 01 to 30 (1 to 30 mm? - fwhm01, fwhm02, etc).
My question is, if the value in FWHM.dat for the original mri_glmfit for a hemisphere is say, 15.294 mm, does mri_glmfit-sim (mc-z) use interpolated data from fwhm15/ and fwhm16/ in its calculations, or does it choose the nearest value (in this case, fwhm15)?
Thanks, Chris Hyatt
Olin Neuropsychiatry Research Center Institute of Living