Hi, Ian-
So, I assume your afni_proc.py (AP) command uses:
-align_unifize_epi local \
is that right? We recommend that now in most cases (at least for scans of humans) to deal with brightness inhomogeneity in the EPIs. Actually, the reason we don't necessarily recommend it in animal scans at present is because the skulls/nonbrain material tend to be so much bigger and more difficult to remove, and the inhomogeneity doesn't interact so well with that. Given the difficulty of auto-skullstripping here, I wonder if that kind of interaction is actually the problem (given discussion about trying 3dSkullStrip). But, for the moment, I still recommend keeping that option on, because of the inhomogeneity of the EPI.
What does your "align_opts_aea ..." option look like? In this case, your initial EPI-anatomical alignment is quite close to start (the 'vorig' QC block image where obliquity is applied in the APQC HTML is the appropriate one to look at for judging that)---so, you probably shouldn't need "-giant_move" or anything like that there. I might just start with:
-align_opts_aea -cost lpc+ZZ \
-check_flip \
If lpc+ZZ is not working well (which surprises me, but MRI datasets caaaan be surprising), then I would next try "-cost nmi".
So, please try adjusting nmi, and then it would be helpful to know your other relevant EPI-anatomical AP options (or, perhaps just plopping the whole AP command for simplicity).
--pt