The fwhm.dat is computed from your data and reflects all the spatial correlation in your data, not just the smoothing you imposed. It is usually the case that there is some inherent smoothness and so the total smoothness is more than what you apply. This may be due to an outlier or just that some areas don't fit the GLM that well (which will create correlation). It could be that one or more subjects have a global thickness that tends to be less than the rest of the subjects. Some modalities (eg, gyrification) also tend to have very high inherent spatial correlation.

doug

On 8/18/14 11:42 AM, Vy Dinh wrote:

Dear Surfers,

While running a mri_glmfit-sim (monte carlo simulation) for roi analyses generated using mri_glmfit, I encountered the following error code:

ERROR: cannot find /Applications/MRI/freesurfer_5.30_x86_64/average/mult-comp-cor/fsaverage/lh/cortex/fwhm31/abs/th13/mc-z.csd

I checked the fwh.dat file within my glm/ directories and realize that there is a discrepancy between the fwhm I set for the analyses and the fwhm in this file (which explains why I encountered the error above.)

Can someone explain why there is a large discrepancy for these roi analyses (which is specified using the --label option)? What kinds of things can I do to run the monte carlo simulation appropriately?

Thank you, Vy Dinh MD Candidate, Class of 2017 University of Wisconsin School of Medicine and Public Health

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer