DICOM tags yaml vs md

[CPernet]

a general comment related to #1021

• metadata.yaml is, in my mind, the BIDS set of definitions and therefore independent of DICOM
• xxx.md is modality specific and can point to DICOM tags

key thing, a DICOM tag can be ambiguous with slightly different usage according to the modality and that's why i do not think they can be unambiguouly attached to a BIDS metadata definition and consequently, as a rule, the metadata.yaml has no reference to DICOM and tables generation from each modality can add arguments referring to specific DICOM

[tsalo]

As a rule of thumb, we have generally included the DICOM tags in the object definitions by default, and then moved them to tables if and when we came across cases where the metadata field applied outside of MRI. That said, I don't think anyone is heavily invested in that approach, and just having the DICOM tag in the table seems fine to me.

[yarikoptic]

@CPernet can you give an example here for me or point to more specific case if it is still present at all?

In

• [ENH] Add explicit wording on DICOM terms correspondence #1450 (comment)
  I mention that we have "corresponds to" and "based on" - are you talking about those "based on" since "corresponds to" should I think be "identical" in value.

[CPernet]

For almost all PET BIDS we added 'Corresponds to DICOM Tags' --> https://bids-specification.readthedocs.io/en/stable/modality-specific-files/positron-emission-tomography.html#radiochemistry

[CPernet]

and yes sometimes 'Corresponds to DICOM Tags' means strictly identical over times you can see on the PET spec is a combo of 2 tags or conversely some tags get split into 2 BIDS keys (eg DICOM would report a filter name a parameters together)