Hello,

1) I am running into several problems with tractography group comparison between patients and controls. My original aim was to find group differences between tracts running from the amygdala to frontal cortex. Using randomise, my initial results showed that controls appear to have a more robust tract between amygdala and frontal cortex that appears to pass through the thalamus (p<0.05). In light of this finding, I ran tractography from amygdala to thalamus and from thalamus to frontal cortex separately. Strangely, neither of these reach statistical significance. Essentially, if group difference from A-->C is significant and appears to pass through B, how is it possible that neither A-->B nor B-->C are significant?

2) I've consistently had the problem of probtrackx tracking into non-sensical areas of the brain. To remedy this, I've tried using elaborate exclusion masks that essentially only allow probtrackx to track through tunnels where known tracts exist. I've also used simpler exclusion masks that simply mask out the contralateral brain and brainstem. Are either of these legitimate methods? Would it be more correct to:
A) Run probtrackx without exclusion masks, then take the intersection of all_FA_skeletonised and each individual subject's tractography, then use randomise to compare these intersections between groups, or
B) Run probtrackx without exclusion masks, then use randomise to compare between groups, then take the intersection of all_FA_skeletonised and the output of randomise (tbss_tfce_corrp_tstat1/2)?

Thank you so much for your time.

Best,
Zak