Hi John, I would recommend that you use permutation. See https://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial/MultipleComparisonsV6.0... In terms of the threshold, "not going higher than .01" means using between 0 and .01 (eg, don't use .05). For permutation, you can use whatever thershold you want.
On 1/23/19 8:46 AM, john Anderson wrote:
External Email - Use Caution
Hi Dr Greve, I would like to correct surface based analyses of PET data for multiple comparisons. I ran group comparisons in three spaces. left and right hemispheres and subcortical. I used the method --cache in mri_glmfit-sim to correct the analyses in eft and right hemispheres. I used the method --grf in mri_glmfit-sim to correct the analyses in subcortical regions. My questions are:
In one of your kind responses to a collegue here https://mail.nmr.mgh.harvard.edu/pipermail//freesurfer/2014-January/035650.h... you had suggested "If you want to use GRF, you should use smooth by twice the native voxel size. I would not go higher than .01 on the threshold". Would you mind clarify what it means don;t go higher than .01 is this the cwp threshold? also smoothing by twice the native voxel is this means fwhm=2mm?
The methods --cache and --grf are giving me the expected results.. when I should consider using simulation --perm ?
Thanks for any clarification John
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer