New version of @animal_warper

Dear AFNI experts,
I just tested the new version of animal_warper. Everything works fine, but I just find something not normal when the bain mask is produced at the end. I guess it is based on the skullstriped anatomical image in the template space that is sent back into the subject space? It seems that the mask is cut when compared to the previous version. In consequence, in macaque, I am losing a few parts of the cerebellum and of the inferior temporal cortex.
Any logical reason for that? Is there a way to correct it?
Thank you!

Salut, Clem-

Could you please post your version number of AFNI, and the exact command usage of @animal_warper? (Esp. for the latter, there are multiple ways masking can be done-- by automasking, or by template mask mapping.)

Also, can you please check what version of the NMT template you are (presumably) using? That would be provided in a README.txt file in your directory downloaded with @Install_NMT.


Salut pt!
The problem occurred when changing from
Version AFNI_20.1.06 ‘Otho’
Version AFNI_20.2.12 ‘Aulus Vitellius’

the version of the nmt is 1.2
and my command was:

@animal_warper -input /Oliver2019_02_02/anatT1_deob.nii.gz -base /nmt_1.2/NMT_SS.nii.gz -atlas /nmt_1.2/D99_in_NMT_with_WM.nii.gz  -outdir 
/7_animal_warper/Oliver2019_02_02 -ok_to_exist

the original anatomical image was previously centred to 0 based on the same images “deoblique” and “skullstriped”.
I am not sure if I understand correctly the command -center_out native, but I also tried with this added to the @animal_warper thinking that it might change something since my anatomical image was already pre-centred. Apparently not =)

Thanks again pt!

Well, that seems like a fine form of the command (and it is the most recent form, indeed), nothing jumps out at me as troubling…

Would you mind uploading that dset, and I will take a look?

A couple of side suggestions, and you might already be aware of these, but:

  • there is a much newer version of the NMT available (NMT v2), as described here:
    … and described in more detail here:
    … where the latter contains direct comparisons showing the improvements in structure/detailing with NMT v2; it also notes that NMT v1.3 (basically, the NMT v1.2 template remade using updated approaches that were developed+applied when making NMT v2) might be a preferable choice to NMT v1.2?

  • The NMT v1.2 and v1.3 only come in 0.250 mm isotropic voxel size. That is tiny! And means that looots of time and memory get used when processing it. There are “lower res” versions of the NMT v2 having 0.5 m isotropic voxels, which are still detailed, but involve less processing (about 8x smaller dsets, memory and spacewise, because voxels are 3D). Basically, for most practical applications, I think the 0.5 mm “low res” versions make a lot of sense to use.

  • The NMT v2 comes with the CHARM (Cortical Hierarchy Atlas for the Rhesus Macaque), also described on an AFNI webpage:
    … and in the above bioRxiv draft. There are 6 levels of the atlas, at increasingly higher spatial resolution. You should read the paper for more details, but it has some practical benefits over the D99 when working at lower resolutions that most practical datasets have, I think.

Sooo, those are a few plugs for altering the base template in your study, which may or may not be practicable at this point. But let me know if you would like me to take a look at the alignment stuff in more detail, with the dataset itself.


Hi pt,
Thank you for the update!
I was not aware of this new version. The new atlases that come with it are going to be extremely useful! This paper is really interesting!
I send to you one of the anatomical images that doesn’t work. I think that it might be interesting to see why it used to work with my previous version of animal_waper and now it doesn’t? On my side, I will try the coregistration with the new version of the NMT.
Again, congratulations for this article, it will be extremely useful to the primate research community!