I have DTI data from two groups: 32 subjects in healthy controls and 31 subjects a patient population. My data was acquired with Siemens 3.0T Trio scanner, 71 directions, and 2x2x2 mm resolution. The group differences revealed seemingly good results, with overlap of skeleton regions with what others have reported in the literature as well as some novel regions. 

However, when I run tbss_sym FA to test for symmetrical differences between left and right separately for each group, I see many voxels reporting as significantly different in both directions (left > right and right > left) for both groups (controls and patients, separately). Not only are there many voxels significantly different, but the corrp files have huge significance (ranges up 0.999899). This does not seem believable and something seems to be wrong. 

I saw on some older forum posts that people had similar concerns but were not able to get to the bottom of this. One thing I can think of is that I did not adjust for any regressors (age, gender, etc.), as I had done in my group contrasts for patients vs. controls. However, I did not think to do this as each subject is matched up to himself and left is subtracted from right for each subject, correct?

I think I remember reading that others thought the issue might be due to scanner inhomogeneities in the x-direction. If this was true, we were thinking to maybe place a couple of subjects on their belly and test to see if that effect is there but inversed now.  

Any insights on this issue would be greatly appreciated. I am happy to provide more info, as will likely be needed, as well. 

Thanks,

-Ricky