To get the t-score, compute the square root of the F, like

fscalc F.mgh sqrt -o t.mgh

You can then get the clusters with

mri_surfcluster --in t.mgh --thmin t-threshold --sign sign --no-adjust --sum sumfile --ocn ocn.mgh

ocn (output cluster number) will be a map with the custer number as the vertex value, you can then run

mri_segstats --i t.mgh --seg ocn.mgh --exludeid 0 --accumulate --o sumoft.dat

somoft.dat will be the sum of ts for each cluster

Hi Doug et al,

I would like to get a different cluster summary stat out of mri_glmfit-sim. Rather than size, I'd like to use the "sum of t-scores" (after Maris & Oostenwald 2007, for example). That is, set a cluster inclusion threshold, and then for each cluster of adjacent vertices that clear the threshold, sum their t-scores. For the permutation significance testing, then, the largest such clusterwise sum would be the statistic used to generate the null distribution... I'm happy to write my own scripts, but can't figure out where to try to pull this number out.

Thanks!

Abby Noyce

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