3dclustsim for ROI

Dear experts,

I have a question related to 3dclustsim correction for a ROI analysis, I hope you may be able to advise me.
Next to a whole brain analyses, we have some hypotheses related to specific regions, so we plan to run some tests within (quite an extensive) ROI.

Is it necessary to apply the ROI mask to both 3dFWHMx and 3dClustSim commands? Or could I use the same acf values and/or cluster thresholds as for the whole brain analysis?

Thank you,
Kind regards, Rozemarijn

Hi Rozemarijn,

That might be a touchy subject these days, and one can imagine arguing either way. And along those lines, hopefully it would not matter much. I would be more inclined to estimate it at the whole brain level, which should be more robust, rather than making it dependent on the exact ROI chosen. You could use the values afni_proc.py already outputs. But again, one could argue the other way.

  • rick