Re: [Freesurfer] Minimum n for glmfit maps?

Hi Adam, it is necessary that the DOF be greater than 0, which means that the number of subjects must be greater than the number of regressors. For, OSGM, the number of regressors is 1, so you can run it with 2. But being able to run it does not mean that you will get good results out of it. The number of subjects you need to see something that looks "normal" depends on how big your effect is and how big your noise is (if there's no effect, then an infinite number of subjects will not help). doug

Adam Nitenson wrote:

Hi Freesurfers,

 Is there a minimum number of subjects that should be used in

isxconcat groups and subsequent glmfit analysis? I recently generated activation maps from a group with only 2 subjects, but the resulting image (too large to attach) is very strage. The activation patterns are "whispy", almost looking like someone just doodled red and blue over parts of the brain, sometimes forming hollow circles. Is this just a consequence of only using 2 subjects or is there another problem I'm missing. I find it strange that both of the subjects' individual maps have rather robust (negative) activation)...much more than the group map would suggest. Another group of three subjects is also a bit strange (not nearly as bad as 2) and it doesn't seem until I reach a group of 4 or 5 that the patterns appear "normal".

For the record here are the commands I used:

isxconcat-sess -sf /cluster/roffman/users/Stable5_PerRun/Subject_Files/Apr2011/2_FB_DEP_TT_1T3T5T7T -d /cluster/roffman/users/Stable5_PerRun -analysis SIRP_Stable5 -c Cond1v3 -c Cond1vFix -c Cond3vFix -c Cond1v5 -c Cond1v7 -c Cond5vFix -c Cond7vFix -hemis -o /cluster/roffman/users/Stable5_PerRun/April_2011/2_FB_DEP_TT_1T3T5T7T

mri_glmfit --y /cluster/roffman/users/Stable5_PerRun/April_2011/2_FB_DEP_TT_1T3T5T7T/SIRP_Stable5/Cond1v7/lh.ces.nii --fsgd /cluster/roffman/users/fsgd/April_2011/2_FB_DEP_TT_1T3T5T7T.fsgd --osgm --glmdir /cluster/roffman/users/Stable5_PerRun/April_2011/2_FB_DEP_TT_1T3T5T7T/SIRP_Stable5/osgm/LH1v7 --surf fsaverage lh --fwhm 4.6

Thanks!

Adam Nitenson, B.S. Brain Genomics Lab Massachusetts General Hospital nitenson@nmr.mgh.harvard.edu Phone: 617-643-3215 _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer