Hello all,
I'm trying to run TFCE on a seed-based resting-state functional connectivity analysis while controlling for the influence of two other seed regions. I was able to run TFCE for a single seed region (bivariate model) in randomise by concatenating each subject's first-level connectivity map for that seed into a single 4D nifti file. Is it possible to add additional seeds to this model? I have first-level connectivity maps for each seed region separately, but I don't have first-level maps that already accounts for the unique contribution of seed A while controlling for seeds B and C.
One option that occurred to me is including the concatenated 4D nifti files for seeds B and C as voxelwise EVs. I've done a low-iteration test of this using the command:
randomise -i seedA_firstlevels_concat.nii.gz -o vwise_test -d vwise_test.mat -t vwise_test.con -m GM_mask.nii.gz -n 100 -T -D --vxl=3,4 --vxf=seedB_firstlevels_concat.nii.gz,seedC_firstlevels_concat.nii.gz --debug
The .mat and .con files were created in FEAT with the specifications:
EV1=connectivity of seed A
EV2=binary confound variable
EV3(vox)=seedB_firstlevels_concat.nii.gz
EV4(vox)=seedC_firstlevels_concat.nii.gz
Positive connectivity contrast=1,0,0,0
Negative connectivity contrast=-1,0,0,0
However, the output is very different from what I get when I run a multivariate analysis via the SPM-based tool CONN, so I suspect this approach isn't valid. Any advice would be greatly appreciated!
Thank you very much,
-Ely
|