Hi FSL group,
I recently performed a VBM analysis between 2 groups and a continuous covariate using randomise and TFCE cluster correction. The interaction was significant for Group 1 - Group 2. However, I'm having a bit of trouble interpreting these results. I take it to mean that Group 1 has larger brain volumes in relation to the covariate. But is the correlation positive or negative? I think that graphing the results would help in analyzing the interpretation but I'm unsure how to do this. I made anatomical ROIs of the significant voxels in the tfce_corrp_tstat.nii.gz file and then used fslstats to get the volume of the significant voxels within the ROIs. But this doesn't differentiate between Group 1 and 2. How can I accomplish this?
And is graphing the interaction not the suggested route to go? Instead, should I look at the correlation within each group and the covariate to see which regions are significant?
Best,
Dana
|