That can happen because the null distribution of the cluster sizes change when you change the threshold. So at a threshold of 1.3, you might have a slightly larger cluster (which improves the cluster p-value), but you pay a penalty for lowering the threshold (which hurts the cluster p-value). In this case, the penalty is worse than the benefit.

On 1/25/2021 9:24 PM, Nasiriavanaki, Zahra wrote:

I just wanted to add that the attached maps in the previous email, was made by the tksurfer command below. I also added the mri_glmfit command.

mri_glmfit --y cespct.nii.gz --osgm --glmdir main.wls --nii.gz --mask ../mask.nii.gz  --surface fsaverage ${space:(-2)} --cortex --fwhm 5 --no-prune

tksurfer fsaverage lh inflated -ov cache.th13.abs.sig.masked.mgh -fminmax 1.3 7

I appreciate if you could let me know your comments.

Thanks a lot!

*Zahra (Mona) Nasiriavanaki, M.D.* Postdoctoral Research Fellow Department of Psychiatry Massachusetts General Hospital 149 13th Street, 149-2615 Boston, MA, USA, 02129

...

On Jan 22, 2021, at 6:28 PM, Nasiriavanaki, Zahra <ZNASIRIAVANAKI@mgh.harvard.edu mailto:ZNASIRIAVANAKI@mgh.harvard.edu> wrote:

Hi Freesurferers

I am running mri_glmfit-sim to do multiple comparison correction with the command below. It goes through without any error. However when I view the corrected maps in tksurfer, I see some clusters in corrected map with p=2 and p=3, but I don’t see them in corrected map with p=1.3 (images attached). The contrast of interest is the same between the maps. The viewing threshold is 1.3 in p=1.3 maps and is 2 and 3 in the p=2 and p=3 maps respectively. I’m confused. Could you please comment on that?

mri_glmfit-sim --glmdir main.wls --cache 1.3 abs --3spaces mri_glmfit-sim --glmdir main.wls --cache 2 abs —3spaces

Thanks a lot Mona

<PastedGraphic-1.tiff>

*Zahra (Mona) Nasiriavanaki, M.D.* Postdoctoral Research Fellow Department of Psychiatry Massachusetts General Hospital 149 13th Street, 149-2615 Boston, MA, USA, 02129

---

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