Stream: implementers
Topic: DICOM Code system
Grahame Grieve (Mar 02 2017 at 00:49):
From the beginning, the DICOM code system has had the canonical URL http://nema.org/dicom/dicm, which is actually a typo (so I discovered when investigating).
It should actually be http://nema.org/dicom/dcm
though something else is possible. I'm talking to DICOM about what URL it should be
but does anyone thing we can not or should not change it at this point?
Elliot Silver (Mar 03 2017 at 02:05):
I have no concerns with changing the URL.
But, I do have a concern about the interplay of FHIR releases with releases of external code systems and value sets. My specific concern is with respect to DICOM codes, but I expect this will hit other code systems as well.
FHIR documents a DICOM code system at http://build.fhir.org/codesystem-dicom-dcim.html (or wherever it will move to when published). This currently points to the always-most-recent version of the DICOM terminology. Similarly, the value set described at http://build.fhir.org/valueset-dicom-cid29.html, says it is defined in terms of http://medical.nema.org/medical/dicom/current/output/html/part16.html#sect_CID_29, which is the always-most-recent version of that value set.
The FHIR pages listed above contain the values in the code system and the value set, respectively.
DICOM publishes updates approximately 5 times a year. FHIR expects to publish, lets say, every 18 months. That means there could be about 8 updates to DICOM vocabulary between FHIR publications. As far as I am aware, DICOM does not rev changes to the code scheme, as it is currently at 01. However, they do version each change of the value set, with the revision being the change date.
What is the intent of FHIR with respect to tracking changes to the code system and value sets between FHIR releases? Is the “official” FHIR code system/value set the one published at the time FHIR is published? Is it the latest DICOM published version? Is it some fixed release of DICOM (2017a, 2016d, etc.)? It seems wrong to point to something that can change, but then include a specific version of the content in the code system/value set expansions.
Grahame Grieve (Mar 03 2017 at 02:16):
this is why it is better for DICOM to publish the code system and the value sets; this means that DICOM is in control of the versioning, life cycle, and releases.
Grahame Grieve (Mar 03 2017 at 02:16):
well, at least the code system.
Grahame Grieve (Mar 03 2017 at 02:18):
we would make the value sets version-less, but since the only apparatus is to define the value sets by enumeration, that doesn't buy us very much. That really means that it would be best for DICOM to publish the value sets, and we'll just bind to them.
Grahame Grieve (Mar 03 2017 at 02:18):
So far, we've got agreement from NEMA/DICOM to put up the redirect; it would be even better if dicom also hosted the code system and the value sets. But given the time line, that might be too much this time around, and we can aim for that next time
Elliot Silver (Mar 03 2017 at 03:34):
So, my guess is that from DICOM's point of view, they are "hosting" the code system and value sets.
Grahame Grieve (Mar 03 2017 at 04:10):
wekll, they're not. but I think you mean, it's what should happen
Elliot Silver (Mar 03 2017 at 18:25):
No, what I mean is that DICOM is publising the code system and value sets, in the way that meets their needs. I expect other SDOs are doing the same. What is unclear to me is how FHIR would like to align with updates, and what FHIR's expectation of what an SDO's hosting of code system and value sets would mean.
(I'm also unsure if the redirect NEMA put in place is actually to what FHIR wants.)
Grahame Grieve (Mar 03 2017 at 18:49):
well, here is what I'd like:
- DICOM publishes the code system and any of the value sets they want to change in an ongoing fashion
- each of them has a canonical URL that points to a DICOM maintained web site (e.g. nema.org)
- when a user (web browser, accept type = text/html) accesses the canonical URL, they get some useful presentation of the content of the code system or value set (a bridging page that explains language differences,(e.g. value set vs concept group), a link to the correct that place in the DICOM standard, and manual download of the relevant resource)
- when an app (accept type = application/fhir+xml|json) access the canonical URL, they get back the resource directly
Grahame Grieve (Mar 03 2017 at 18:49):
- we just reference the value sets by canonical URL in the resource definitions
Elliot Silver (Mar 03 2017 at 18:57):
That's what I was afraid of. FHIR would like another SDO to present their content in a FHIR-friendly way (bridging page, multiple ways to download a FHIR resource, etc.). I expect DICOM won't mind doing this, but is work that is being imposed on another organization.
Grahame Grieve (Mar 03 2017 at 20:12):
well, right now, we're doing that, but it has to be at our life cycle. You can't have your cake and eat it too
Grahame Grieve (Mar 03 2017 at 20:12):
the II committee can keep the current build as up to date as it likes though
Grahame Grieve (Mar 03 2017 at 20:13):
the real question is, 'whose content is it'?
Last updated: Apr 12 2022 at 19:14 UTC
