Howdy -

I'm trying to track down a weird issue. I have a number of anatomical masks that I've generated through wfu pickatlas. These masks default to Talairach space. We've converted our functionals to MNI space using uber_subject, as you can see from the following 3dinfo report:

#

$ 3dinfo all_runs.tp119_day1+tlrc. | grep -i 'template'

++ 3dinfo: AFNI version=AFNI_2011_12_21_1014 (Mar 1 2013) [64-bit]
Template Space: MNI

#

I noticed, however, that the masks we're using, which I thought I'd converted to MNI space using 3dresample, have header info claiming that they're still in Talairach space:

#

$ 3dinfo b38_afni.nii | grep -i 'template'

++ 3dinfo: AFNI version=AFNI_2011_12_21_1014 (Mar 1 2013) [64-bit]
Template Space: TLRC

#

This was exciting news, because if I'd screwed that up, then some very mysterious things would have a plausible explanation. So I tried to move these masks into MNI space to test the idea:

#

$ 3dresample -master all_runs.tp119_day1+tlrc. -inset b38_afni.nii -prefix b38_mni.nii
$ 3dinfo b38_mni.nii | grep -i template

++ 3dinfo: AFNI version=AFNI_2011_12_21_1014 (Mar 1 2013) [64-bit]
Template Space: TLRC

#

However, as you can see, it didn't work -- even after resampling using a -master image that reports being in MNI space, the resampled image reports as being in TLRC space.

I'm not sure what to make of this. When you look at these masks as overlays in AFNI all seems fine, but then, I know that AFNI will juggle things behind the scenes to put underlays and overlays in the same space. One might assume that MNI and TLRC are synonyms for each other in the same spirit as when AFNI names files as +tlrc even when the data in the file is not in Talairach space, except that here the header actually _does_ make the distinction between TLRC and MNI, as reported in 3dinfo.

So what's going on? Seems like two possibilities:

#1: everything really is in MNI space, but for some reason AFNI doesn't alter the header from TLRC --> MNI during resampling because it thinks TLRC is synonomous with MNI for some reason.

#2: 3dresample is not actually transforming one space to another, as I thought it was, and instead is only converting the voxel size.

Can anyone shed some light on this?

Thanks very much,

Shane