Dear AFNI experts,
I was trying to do an ROI analysis by correlating beta coefficient differences with behavioral performance. My question is when creating ROIs with 3dUndump based on activation map, which master dataset should be used? Since my data were wrapped into the Talairach space with TT_N27 template (before 3dDeconvolution), I tried using TT_N27 as the master dataset. I also tried using individual statistical data as the master dataset, though I expected the ROIs generated from these two methods would be the same, I found the averaged beta coefficients (calculated by 3dmaskave) were actually different (though not very much different) from each other, which means the ROIs were not identical. I am wondering which method is the good way to go, using the TT_N27 template or individual statistical dataset? I pasted two example scripts I used to create ROIs below, the only difference is the master dataset:
echo “2 -64 38” | 3dUndump -prefix precuneus_R1 -srad 5 -orient LPI -master ./TT_N27+tlrc. -xyz -
echo “2 -64 38” | 3dUndump -prefix precuneus_R1 -srad 5 -orient LPI -master ./stat.s1+tlrc. -xyz -
In addition, I would like to ask a basic question about percent signal change. I have followed the standard processing pipelines in AFNI, which means scaling the data before individual subject analysis, then could I interpret the beta weights as percent signal change? Say, the beta weights for one regressor is 0.5, could I say the percent signal change for that condition is 0.5%?
Thank you very much for your help!