Hi Rick -

First, thanks for taking the time to make such a thorough and thoughtful response.

First things first: the transformation I used to translate my MNI results into TLRC space was:

3dWarp -mni2tta -prefix tlrc_stats_nii.gz mni_stats.nii.gz

I did this to the stats and also the reference anatomical, to maintain concordance with the reported results - the idea was that whatever got done to the stats would also get done to the anatomical.

Reading your response, though, I realize that I'm confused with regard to nomenclature. It seems like there are a few of issues in play:

1. The particular brain that is used as the 'ideal' brain, and to which the subjects' brains (and functionals) are aligned/warped.
2. The coordinate system that is used to navigate that brain, and relative to which landmarks are defined.
3. The atlas, which maps the coordinates from the space in #2 to names or features.

Even after some research I don't feel confident in prying the first two things apart. According to my current understanding I'd call the first the 'template brain' and the second the 'space.' If that's right, then it would seem that you could align any subject's brain to any template brain, and as long as that template had a well-defined mapping to the space you were interested in, then if you had an atlas for that space, then you'd be able to reconstruct areas of anatomical interest.

If that's right, it would be a revision to what I was thinking when I wrote my original message; I assumed that if you wanted to be in TLRC space, and use some of the TLRC atlases, then you'd have to align your subjects to that very atypical brain, which, since it's atypical, would lead to needless distortions. But that reading of that situation no longer seems to be correct.

Anyway, if I understand correctly, if I want to use some of the atlases only available in TLRC, the best thing would be to re-process the data; choosing a more typical brain in TLRC space as the template. Presumably TT_avg152T1 would be acceptable for that purpose. Following this process would alleviate issues like those pointed out here:

[imaging.mrc-cbu.cam.ac.uk]

in which the temporal lobes in the MRI brain lie outside the coordinate range of the temporal lobes in the TLRC template. This wouldn't be a problem because the anatomy would have been squashed to conform to the TLRC template's geography; and so the coordinates of the temporal lobes would (in the transformed image) be in-bounds; and atlas references to the anatomy would be appropriate.

If I've made any egregious mistakes I'd appreciate you pointing them out. Otherwise, thanks again for taking the time to lay out the situation.