Atlas download for anatomical ROI extraction

AFNI Message Board
1

23.3.02
Hello, all
I cannot get afni to recognize a downloaded atlas through the -whereami command. I downloaded AAL3_dist directly from the url into my ~/abin directory. I extracted the contents of this folder into the ~/abin folder. I used @Atlasize in the command line:

image
image1092×245 12.1 KB

When I use whereami -show_atlases, the downloaded AAL3 atlas does appear. However, I cannot pull anatomical ROIs from this atlas with whereami -mask_atlas_region AAL3v1_1mm:155 -prefix L_dACC. Here are screenshots showing these two outputs:

image
image1117×532 49.7 KB

image
image1112×127 6.53 KB

When I open afni, a warning pops up that the plugins it is loading are all outdated. In my hours of trying to download this atlas, I made a lot of changes to AFNI_PLUGINPATH via @AfniEnv -set. I'm sure my own ill-informed choices created the following warning:

image
image1084×555 16.7 KB

Two issues I'd like to resolve:

  1. Update my plugins
    1b. Have afni atlases + AAL3 atlas appear in Plugin > Draw dataset
  2. Extract regions as they are anatomically defined in AAL3 with whereami command

Thanks so much. Been working on this over the weekend and I am at a complete standstill in my analytic pipeline because of something I think is relatively simple. sigh.

2

I wasn't sure, so I tried doing that kind of whereami behavior with the Brodmann_pijn_afni.nii.gz atlas.

If I run 3dinfo -labeltable Brodmann_pijn_afni.nii.gz, I see a 2-column list of a bunch of region values (integers) and the corresponding region label (strings/words/descriptors). One of these is the the region labelled "ctx-rh-BA11", which corresponds to the region of number 110.

If I try asking whereami to create an output dataset by specifying region value, I get the same error you do:

$ whereami -mask_atlas_region Brodmann_Pijn_AFNI:110 -prefix TEST1
++ Input coordinates orientation set by default rules to RAI
No exact match for 110
   I don't even have good suggestions.
   Try to be more explicit.

However, if I try to do so by using the region label, then life is good:

$ whereami -mask_atlas_region Brodmann_Pijn_AFNI:ctx-rh-BA11 -prefix TEST2
++ Input coordinates orientation set by default rules to RAI
Best match for ctx-rh-BA11:
   ctx-rh-BA11 (code 110)

So, I suspect if you query by the region label associated with region value "155" in your atlas, things should work.

I'm not sure about the plugins aspect. It might help if you copy+pasted the text of what you have changed in your AFNI environment variable setup in ~/.afnirc. Or if you want to display what your current AFNI_PLUGINPATH env var value is currently, what is the output of:

afni -q -VAFNI_PLUGINPATH=

?

--pt

3

Hi ptaylor,
Thanks for your insights on the mask issue! Very helpful.
Here is the output you requested. There is also a screenshot of what I can see when I right-click in the viewer window, and we are missing some essential features. These features are missing for every user on our lab server

**tur50860@cla19097**:**/data/projects/STUDIES/SEAT/fMRI/Masks/Anatomical**$ afni -q -VAFNI_PLUGINPATH=

/home/local/TU/tur50860/abin

image

4

I can't make out the issue in your last post. A couple notes on this:

  1. The AAL3 atlas you downloaded from our site was already "atlasized", so you don't need to do it again. Setting AFNI_ATLAS_SUPP_DIR in the .afnirc file to point to the path where the atlas and the CustomAtlases.niml file are sitting should be sufficient to use it similarly to any other atlas in AFNI.
    https://afni.nimh.nih.gov/pub/dist/atlases/aal/AAL3_dist.tgz

  2. The AAL3.1 1mm atlas is very similar to the MacroLabel atlas from the Eickhoff-Zilles Anatomy toolbox. That's in our standard distribution as theMNI_caez_ml_18+tlrc. See the image below for comparison (AAL3.1 1mm left, MacroLabel on right). Mostly the subcortical regions have been more finely delineated, and the cortical regions have been more smoothly defined in the 1mm version of the AAL3 atlas. All versions are made for the MNI N27 template (not the MNI 2009c template). Even there, you will notice the atlas (old or new) does not cover the template or goes too far past the brain. Many regions seem to be defined in axial planes as seen by the distinct planar delineations in coronal and sagittal views with a few regions that have coronal plane delineations (See second image below). There are also a number of holes and disconnected "lost clusters" of regions, separated from the larger part of the region. In a quick review of that, I found 48 lost clusters.

AAL3.1_1mm vs MacroLabel
image

AAL3.1_1mm (axial and sagittal views)
image

1 Like