Using perm will certainly avoid a lot of questions. You should still use the --2spaces. You changed the threshold from 2.3 to 1.3. This threshold is arbitrary, you don't necessarily need/want a lower threshold for perm. I've been doing some experiments lately, and a threshold of 2.3 with >5mm of smoothing gives about 5-10% false positive rates (when expecting 5%) on thickness. So it is not nearly as bad as the fMRI results from Eklund. Note that if your cluster p-value is much less than .05, then you're probably in good shape.
On 10/24/16 9:49 PM, neuroimage analyst wrote:
Hi FreeSurfer experts,
I am currently performing some cortical thickness comparisons in my group of subjects and using mri_glmfit-sim --glmdir my_dir --cache 2.3010 --cwpalthresh 0.05 --2spaces found some clusters that have significantly different cortical thickness values,
Given Eklund et al PNAS paper, do you suggest I should rather use mri_glmfit-sim --glmdir my_dir --sim perm 5000 1.3 perm.pos.13 --sim-sign pos
in order to convince that the results are not due to false positive?
Thanks
Regards
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer