Left-right flipping possibility?

[afni-dglen]

This repository has an impressive number of atlases. I took a quick look at two of these atlases that were of particular interest, and it appears that both may be flipped left-right relative to an MNI template. I looked at the Glasser and Brodmann atlases. It's not clear which MNI template is being used, so I can't be completely sure. Still neither match the 2009c asymmetric nor the original MNI152 templates. The usual petalia with the left hemisphere slightly more posterior and the right hemisphere slightly anterior seems to be backwards here. Both qform and sform codes (value=4 for MNI) and values for the sform_rows seem to be set in the NIFTI header, so the problem is not a missing header but possibly an incorrect one. I will note that we have what I think is an improved version of the MNI transformed Glasser atlas in the AFNI distribution, and soon a better Brodmann atlas too.

[frankyeh]

The aal_space-MNI152NLin6_res files are also flipped, as the left region all show MNI X value > 0.

[afni-dglen]

I'm going to bump this issue. It's been more than 4 years, but many of these atlases still appear to be left-right flipped relative to the template. I have put as an example the Glasser atlas over the MNI 2006 nonlinear template from this repository. The template provided with FSL seems identical to this one too. You may note the atlas is a little to the left of the underlying template and off in other more subtle ways. Left-right flipping the data of the atlas seems to fix this, as shown in the central image. The image on the right is the Glasser HCP atlas distributed with AFNI over the MNI 2009c asymmetric template. That atlas was transformed and projected in an obviously different way, but you can see the closer match to the anatomy. Note, the Glasser atlas seems to be updated relative to the one 4 years ago, but the problems are essentially the same even with the newer version shown here.