3dROIMaker optimal number of ROIs

AFNI Message Board
1

Hi everyone,

I was wondering if there was a way to determine the optimal number of ROI’s one can achieve from performing 3dROIMaker?

There are many arguments that come with the 3dROIMaker command. And while they are useful, they can be overwhelming in the sense that I’m not sure how to get that golden number of ROI’s that are optimal for each network?

Playing around with the different options for the arguments, made me realize I don’t know how one would achieve the “optimal” number of ROIs.

What I have is this:

3dROIMaker -inset network0006.nii.gz -thresh 5.4 -prefix network6_thresh5.4 -volthr 130 -inflate 2 -trim_off_wm -wm_skel JHU-ICBM-FA-4mm.nii.gz -mask MELODIC.gica/mask.nii.gz -skel_thr 0.2 -nifti

But so far it’s been a trial and error, paired with visual inspection. However, I’m not sure if thats reproducible for others who want to perform the same steps I did.

Would you happen to have any suggestions?

2

Hi, Sondos-

Well, the program is mainly meant to implement values you have from elsewhere-- threshold values from statistical hypotheses/testing, volume thresholds from cluster-based calculations, etc. I don’t see any inherent optimality available for choosing a number of ROIs-- I would think that most of the parameters should be given by other data or prior assumptions.

–pt