Cluster correction after 3dTcorr1d and 3dMVM

AFNI Message Board
1

Hi experts,

I am new to AFNI and could use a bit of help. I am interested in examining how my variables of interest relate to amygdala resting-state connectivity. I am using preprocessed normalized resting-state fmri data.

In order to calculate amygdala connectivity maps, I have ran 3dTcorr1d on my preprocessed individual datasets. I have then used 3dMVM to assess the associations between my variables of interest and amygdala connectivity at the group level. I would now like to perform cluster correction on my group level results, but am not sure how to go about this. If I’m correct, both 3dTcorr1d and 3dMVM do not output residuals, which would be necessary to run 3dFWHMx to acquire the ACF model parameters. Or am I missing something? Should I choose different programs?

Any guidance would be much appreciated!

Thank you in advance.

Sandra

2

Sandra,

Since this is resting-state data, you can run 3dFWHMx -acf on your preprocessed individual datasets, use those ACF parameters with 3dClustSim.

3

Thank you Gang!

I ran 3dFWHMx -acf on my preprocessed resting-state datasets and averaged those outputs to feed into 3dClustSim. However, for some of my participants I get much higher b parameters than others (e.g. 64.8471 vs 4.74161). Do those values make sense? What would cause the big difference?

I also need really large clusters in order to get significant results (NN1, pthr = .001, athr = .05, cluster size > 500). I think that has to do with the large averaged b parameter. Is that correct?

Thank you!

Sandra

4

Hi Sandra,

Make sure the 3dFWHMx is run with a brain mask. If you show a command, we can take a look.

  • rick