New BrainAtlas: MarsAtlas

[AhmetNSimsek]

siibra team created MarsAtlas based on http://meca-brain.org/software/marsatlas/. The color map and the subcortical structures are taken from http://meca-brain.org/software/marsatlas-subcortical/. Could we validate and integrate it to openMINDS?

• https://github.com/FZJ-INM1-BDA/siibra-configurations/blob/master/maps/colin27-marsatlas-labelled.json
• https://github.com/FZJ-INM1-BDA/siibra-configurations/blob/master/parcellations/marsatlas.json

Publication: MarsAtlas: A cortical parcellation atlas for functional mapping

[UlrikeS91]

Thank you for the request, @AhmetNSimsek. It would be very interesting to integrate this into openMINDS.

I had a look at the resources you linked here and additional ones that I found online. To me, it seems that the MarsAtlas is a model/a parcellation scheme. It is not a complete brain atlas in itself. We have registered parcellation schemes (i.e., Brodman) in openMINDS, so we should be able to do this in a similar way for the MarsAtlas.

I am curious about the resources that you linked and how these have been produced. colin27-marsatlas-labelled.json seems to be a representation of the MarsAtlas parcellation scheme in the MNI Colin 27 space. Has the MarsAtlas model been applied to the template data to produce the volume (https://data-proxy.ebrains.eu/api/v1/buckets/reference-atlas-data/temp/MarsAtlas/colin27_MNI_MarsAtlas.nii)? Or has a pre-existing volume been transform into the MNI space?

[lzehl]

@UlrikeS91 thanks for pushing this. @AhmetNSimsek as discussed MarsAtlas is a deterministic brain atlas (normally autogenerated on a subject space). For openMINDS we will register MarsAtlas with Colin27 as common coordinate space. @AhmetNSimsek what is the status? do you need help? we can meet to further discuss.

[AhmetNSimsek]

Thank you @UlrikeS91 and @lzehl. Excuse me for returning to you so late.

To me, it seems that the MarsAtlas is a model/a parcellation scheme. It is not a complete brain atlas in itself

I am not sure how do you differentiate a complete atlas and a parcellation scheme. Could you elaborate?

colin27-marsatlas-labelled.json seems to be a representation of the MarsAtlas parcellation scheme in the MNI Colin 27 space

The jsons I linked are configuration files made for siibra-toolsuite. We create such files for parcellations, atlases, features, spaces and more. So these files are generated from original source material to be in the format siibra can digest.

https://data-proxy.ebrains.eu/api/v1/buckets/reference-atlas-data/temp/MarsAtlas/colin27_MNI_MarsAtlas.nii

The nifti file you are referring to is available at http://meca-brain.org/software/marsatlas-colin27. We just stored it in a data-proxy due to the way it was stored (technical limitations).

For openMINDS we will register MarsAtlas with Colin27 as common coordinate space.

Maybe we should have a short meeting to discuss this because from what I understand, it is not a common coordinate space.

EDIT: Perhaps MarsAtlas is not a good starting point due to these ambiguities. We could take a look at another parcellations that are supported in siibra but not registered in openMINDS yet.

[lzehl]

To clarify a bit:

Brain atlases according to the Atlas Ontology Model (AtOM) must have:

• annotations (with defined criteria on how they were generated)
• terminology (naming the annotations)
• coordinate space
• reference data (brain) in the coordinate space

The MarsAtlas fulfills all these requirements:

• annotations are automatically created based on sulci and gyri patterns
• terminology is defined for the resulting annotations
• coordinate space is defined, usual use case: patient space
• reference data (brain) is defined, usual use case: patient brain

The usual use case, would require us to register one MarsAtlas version for each patient. However, as a generic example MarsAtlas provides one application to the coordinate space and reference data of the Colin27 reference brain.

[lzehl]

So for the atlas registration in openMINDS we only register the Colin27 case

[AhmetNSimsek]

Thank you for the clarification and discussion on Zoom, @lzehl. Based on our conversation, it is easier for me to continue in the following order: 1) Convert the AAL atlas from openMINDS to siibra configuration. (tracked here). I will create a script as generic as possible for this purpose. 2) Create a PR to openMINDS for MarsAtlas on Colin27. 3) Recreate the siibra configuration for MarsAtlas using the script from 1.