Complete MarsAtlas

[UlrikeS91]

Overview: https://docs.google.com/presentation/d/1BgQfFLI6he-001FhLZ_Cfj1SgGcQhJdhgIWm1hmhgUQ/edit?usp=sharing (NOTE: names in jsonlds not identical to the ones in overview!)

PR #257 is related to this (at_id remains the same either way)

General:

• 3 BAVs:
  • 1st from 2016 - 'cortical model': parcellation scheme with 41 cortical PEVs (and 7 PEs representing lobes/regions for hierarchical grouping)
  • 2nd from 2017 - 'subcortical extension': 1st parcellation scheme extended with 7 subcortical PEVs (no additional PEs)
  • 3rd from 2018 - 'Colin27-MNI': parcellation scheme (cortical + subcortical) applied to the Colin27 average brain in the MNI space
• BA(V) names, IDs & descriptions:
  • fullName: Marseille Atlas
  • shortName: MarsAtlas
  • abbreviation: MarsAtlas
  • versionIDs: 'cortical model', 'subcortical extension', 'Colin27-MNI'
  • description: The MarsAtlas is a model of cortical and/or subcortical parcellations and can be applied to any human brain volume files following T1 imaging. The model is implemented in the FreeSurfer software, and cortical and/or subcortical segmentations can be obtained through specific pipelines. When the MarsAtlas is applied to commonly used coordinate framework with a standard reference dataset in a common coordinate space, the output may act as a reference brain atlas.
  • versionInnovations: describe version specific details
• PE(V)s:
  • all cortical PEVs + hierarchical grouping terms + all subcortical PEVs = 41 + 7 + 7
  • hierarchical terms represent lobes/larger cortical regions (e.g., occipital lobe)
  • subcortical regions do not have additional hierarchical terms (also: 'brain' does not exist as the highest level like some of the other atlases have)
  • cortical PEVs and their corresponding PEs have same hierarchical PEs as parents, subcortical never have parents
  • some PEs have related UBERON terms (NOTE: linked UBERON term may not exist in repo yet, but I checked the PRs for Add all UBERONParcellations from UBERON Minimal Nervous System subset #133 and they are included there)
  • all PEVs have AtlasAnnotations, each has two - one per hemisphere with corresponding internal ID; 'Colin27-MNI' has additionally type 'annotationMask' (when parcellation schemes are applied, they always result in surfaces first but can be turned into masks (as was done for the 'Colin27-MNI' version), therefore we cannot state one or the other for the PEVs of the parcellation schemes)
  • versionInnovations:
    • 1st set of cortical PEVs: This is the first version of this cortical parcellation area.
    • 2nd set of cortical PEVs: This is identical to/unchanged compared to its previous version.
    • 1st set of subcortical PEVs: This is the first version of this subcortical parcellation area.
    • Colin27-MNI set of PEVs: This is the delineated parcellation area in Colin27-MNI following the MarsAtlas parcellation scheme.

Questions/Noteworthy:

• see New BrainAtlas: MarsAtlas #22: @AhmetNSimsek provided some content there
  • we may want to add the colors for the 'Colin27-MNI' version (if those are the ones originally provided by the authors)
  • they are organizing the PEVs differently though (one annotation per PEV, thus, one PEV for left and one for right with a PE representing the generally term as parent for the two)
• BAV/digitalIDs are the DOIs of the related publication (meaning the Colin27-MNI version does not have one), there is nothing else available; is that ok?
• copyrights exists, should we add it?
• could add qualitativeRelationAssessments between cortical PEVs of the 1st and 2nd version (aka 'identical'); would this be useful?
• not sure about the 'howToCite' descirpitons for each BAV, the one for BA was impossible to write; what should I do here?
• none of the BAV have licenses, what should we do about that?
• BAV releaseDates are the publication dates of the related articles, ok?

[lzehl]

@UlrikeS91 do you really think it is worth separating the MNI-Colin27 version from the pure parcellations? They actually do not change through this and the MNI-Colin27 application would give the other two versions more substance. Chronologically speaking you are of course right.

[UlrikeS91]

@UlrikeS91 do you really think it is worth separating the MNI-Colin27 version from the pure parcellations? They actually do not change through this and the MNI-Colin27 application would give the other two versions more substance. Chronologically speaking you are of course right.

I don't think I understand what you mean. All three versions are attached to the version-independent atlas representation. The subcortical extension is a new version of the cortical model and the MNI-Colin27 application is a bit on their sideline. We could link it as either alternative or new version of the subcortical extension, but neither seemed very accurate.