[FIX] Consistently refer to Neuromag/Elekta/MEGIN

[sappelhoff]

• closes [BUG] inconsistent specification of Elekta/Neuromag and Neuromag/Elekta/Megin #1309

Thanks for your report @robertoostenveld -- I have looked through the spec to see where else we are inconsistently referring to that system that has been owned by three manufacturers.

In this draft PR I have naively changed the naming to Neuromag/Elekta/MEGIN, however:

1. If we want to go through with this, some metadata field values have to be deprecated. For example ElektaNeuromag is currently a restricted keyword for the MEGCoordinateSystem metadata. We can change it to be NeuromagElektaMegin, and deprecate EletaNeuromag, that is, print a warning when somebody uses that "old" term and recommend they use the new one instead.
2. if this MEG system changes "owners" in the future again, e.g., to company XYZ, are we then calling it Neuromag/Elekta/MEGIN/XYZ?

This PR can remain a draft until we've discussed these issues.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 87.97%. Comparing base (70d7750) to head (8f21acb).

❗ Current head 8f21acb differs from pull request most recent head 3b12da1

Please upload reports for the commit 3b12da1 to get more accurate results.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1310      +/-   ##
==========================================
- Coverage   88.04%   87.97%   -0.07%     
==========================================
  Files          16       16              
  Lines        1380     1356      -24     
==========================================
- Hits         1215     1193      -22     
+ Misses        165      163       -2     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[effigies]

Does it make sense to just declare that these are synonyms that SHOULD be used consistently throughout a dataset, and then we use one consistently throughout the spec (linking to the bit about synonyms)?

[sappelhoff]

@robertoostenveld what do you think about the two points mentioned in my OP, and about Chris' point?

Perhaps @ftadel and @Moo-Marc can offer additional opinions?

[robertoostenveld]

I like the idea of considering them as synonyms, that also aligns with (non-technical) use of these in regular conversations.

With FieldTrip I also suffered from this (also for other companies), which I why I at a certain point adopted the strategy to refer to them as "old/new/newer/evennewer", as that at least puts them in a consistent order. But considering them as synonyms is better, and aligns better with the actual market situation. E.g. Elekta (the company) might not like it nowadays that we still refer to their name in relation to MEG systems that are currently not manufactured by them any more (but by MEGIN), whereas labs that purchased their system a few years back will still have the old brand stickers on the scanner and probably prefer to refer to that system as their "Elekta" system, and not the name it would have had - had they bought it recently.

[robertoostenveld]

another similar update would be desired for the MEG system that was initially designed at the Kanazawa Institute of Technology (KIT), later produced and supported by Yokogawa and that now continues at the Ricoh company (we have a copying machine of them, not a MEG system). That should be considered the KIT/Yokogawa/Ricoh system

[Moo-Marc]

+1 for synonyms. Could this also apply to keywords? Instead of renaming ElektaNeuromag, could we have Neuromag, Elekta and Megin as synonym keywords?

Even CTF has renamed itself at least 4 times, though people kept calling them CTF, hence its return: CTF, VSM, MISL, CTF MISLP, CTF Neuro Innovations

[sappelhoff]

I decided to go the deprecation route here. Let's discuss the "synonym" idea again when one of our supported manufacturers actually changes their name. The changes here are "bug fixes".

Ready for review / merge @effigies @robertoostenveld @Moo-Marc @Remi-Gau

[effigies]

If we're going to deprecate, then we should have checks to warn curators to normalize. It seems this shows up in 7 metadata fields:

$ git grep -B12 _MEGCoordSys | grep \ name:
src/schema/objects/metadata.yaml-  name: AnatomicalLandmarkCoordinateSystem
src/schema/objects/metadata.yaml-  name: DigitizedHeadPointsCoordinateSystem
src/schema/objects/metadata.yaml-  name: EEGCoordinateSystem
src/schema/objects/metadata.yaml-  name: FiducialsCoordinateSystem
src/schema/objects/metadata.yaml-  name: HeadCoilCoordinateSystem
src/schema/objects/metadata.yaml-  name: MEGCoordinateSystem
src/schema/objects/metadata.yaml-  name: NIRSCoordinateSystem

Could make a check in src/schema/rules/checks/deprecations.yaml:

AnatomicalLandmarkCoordinateSystemDeprecation:
  issue:
    code: ELEKTA_NEUROMAG_DEPRECATED
    message: |
        "ElektaNeuromag" as a coordinate system is deprecated.
        Use "NeuromagElektaMEGIN" instead.
    level: warning
  selectors:
    - sidecar.AnatomicalLandmarkCoordinateSystem
  checks:
    - sidecar.AnatomicalLandmarkCoordinateSystem != "ElektaNeuromag"

DigitizedHeadPointsCoordinateSystemDeprecation:
  $ref: rules.checks.deprecations.AnatomicalLandmarkCoordinateSystemDeprecation
  selectors:
    - sidecar.DigitizedHeadPointsCoordinateSystem
  checks:
    - sidecar.DigitizedHeadPointsCoordinateSystem != "ElektaNeuromag"

...

[sappelhoff]

Thanks for the suggestions @effigies I agree and have created a PR for this, using your code. Please feel free to adjust if that's not what you had in mind. #1848

[robertoostenveld]

looks good to me.

[effigies]

If the checks will be in another PR, then this seems good to go.