Making ROIs (masks) in MNI space

Hi everyone!

First time posting. I have what feels like a fairly basic question but I can’t find simple answers to.

I have fMRI data in MNI space (which I like because I prefer MNI to Talaraich). However the only way I know how to make Regions of Interest /masks from an atlas was to use the TT_Daemon atlas already in afni.
My goal is to either a) learn how to transform masks from TT to MNI or b) Find another atlas in MNI space already for me to get my masks from.
Although I have used afni so far I am open to non-afni solutions.

For a) my thought was to find a TT template and MNI template brain and transform TT to MNI to get a transformation matrix. I found “MNI_avg152T1+tlrc.BRIK” and “TT_N27+tlrc.BRIK” in /opt/afni/. I used @auto_tlrc -base MNI_avg152T1+tlrc.BRIK.gz -input TT_N27+tlrc.BRIK.gz -suffix .trans to attempt this. I am not sure if it worked but I don’t see a transformation matrix or at least I am not sure how to tell what a transformation matrix looks like.
Also, once I do get a transformation matrix, how am I to use that to transform my ROIs to get the same transformation?

For b), I have looked online but haven’t found anything. Anyone have tips on how to find this?

I know I am a bit of a novice so please let me know if you need more info to answer my question. Thanks so much for any help you can offer.


You could transform the TLRC versions too, but it’s not necessary. The MNI_ANAT versions are distributed with AFNI. That space is a slightly shifted version of MNI space with the (0,0,0) coordinate at the anterior commissure. These were made available by the Eickhoff,Zilles group in their Anatomy Toolbox. You could shift these over with the 3dWarp command or download them separately. The MNI versions of atlases are available here:

You can learn more about the way we deal with atlases, templates and spaces from this class documentation:

Okay Thank you very much for the atlas.

The MNI atlas should be exactly what I need. I’ll try implementing it and then update on my progress!

Hi Daniel

Unfortunately the atlases by the Zilles group does not have ROIs for the thalamic subregions that I need.

So, I am trying to warp the TT_Daemon atlases into MNI space to get MNI ROIs that way.

My plan is to warp from the afni directory the TT_avg152T1+tlrc.BRIK.gz to the MNI_avg152T1+tlrc.BRIK.gz template. Then I will use the resulting transformation matrix to transform the atlas. However I am getting an error I was hoping you could help me with.

My command was set up like this:
3dAllineate -input /opt/afni/TT_avg152T1+tlrc.BRIK.gz -prefix warped152_mniHopefully -1Dparam_save ParametersSaved.1D -master /opt/afni/MNI_avg152T1+tlrc.BRIK.gz

But the error says: ** FATAL ERROR: No base dataset AND source dataset has only 1 sub-brick
I figure there is something special I have to do to let it know it is anatomical to anatomical and not epis, but I can’t find from the documentation how?

I tried using the .py script in case but it also doesn’t work. -anat /opt/afni/MNI_avg152T1+tlrc.BRIK.gz -epi /opt/afni/TT_avg152T1+tlrc.BRIK.gz -epi_base 1
AttributeError: ‘NoneType’ object has no attribute ‘input’

Thanks for any tips you can offer!


So I am trying out more things with the script. I have tried 3dcopying the anatomy files from /opt/afni/ to my working directory. I have also tried adding -cost lpa (which supposedly estimates transforms better when going from anatomy to anatomy).

It’s running now and the result is… “Finished alignment successfully”!!!

Great, thank you so much for the help and all the hard work you guys must have put into afni, documentation, and monitoring this message board! We really appreciate it.


Depending on which regions are of interest, you might want to take a look at this preliminary version of the sub-cortical atlas from the Keuken group. I have made an AFNI compatible version you can download at the link below to try. Regions like the STN (sub-thalamic nuclei) are included.

Keuken atlas in suma

Get that file and expand it with “tar xzvf …” Then set the environment variable so AFNI’s whereami knows it exist and uses it.

@AfniEnv -set AFNI_SUPP_ATLAS_DIR $outatlasdir

I’ve put surfaces there along with the “original” maximum probability maps and modally smoothed versions for each age group. These are my own modifications of the Keuken atlas, and I have yet to speak with the authors about this. That atlas and other related sub-cortical atlases are available in their original form here:

In the past, I’ve discussed a variety of methods for the transformation between Talairach space and MNI space. There are lots of options, and none of them are really “right” especially regarding the Talairach daemon atlas. There is no corresponding MRI template for the Talairach daemon, so these are all very rough approximations. Adding on to these difficulties of transformations and templates, the Talairach atlas is roughly sliced by coronal slices, and the atlas has a number of errors of inconsistencies across slices in 3D.,70173,80922#msg-80922