Hmm... it seems those contingency tables only show the associations of each subject with those 6 factors, and they do not include the most informative ones about the those 6 factors: Scanner, AA, Male, BinCan, BinTC, BinDrug.
I suspect that the problem is that you don't have all the combinations available among those 6 factors. 3dMVM is picky in the sense in the sense that it requires all the combinations available with at least 2 samples (e.g., subjects). To get around the issue, try dummy code those 6 factors as 1 and -1 (assuming there are only 2 levels for each factor), and see if 3dMVM would accept it.
Gang
