Meeting minutes
Intros
Lukman_Ismalia: Johns Hopkins, CS background PHD, Machine Learning in France. Post doc school of med. Int in FAIR. Medical biobank. Int in DICOM and FAIR.
Josh_Moore: In Leiden, trying to gather people around DICOM. Responsible for microspopy images. Have been pushing toward semantic tech.
DICOM
erich: Started a COVID image data lake. Wrote software for 20B triple store. Did my own mapping of DICOM to RDF. It worked. For multiple reasons, that work will continue. But would rather do community effort on DICOM. Only one proj seemed to still be active. Curious to see how FHIR was handling imaging data.
… Also reached out to David Cluney, who said their's no official casting to RDF.
josh: Two groups int: German equiv of NIH. His employer might call in from the train.
… Another: EU horizon project, funded for next 30 months, in Torino. Want to bring them all together if possible.
… How official can we make on eof the castings?
erich: Current appproach I was working on was SHACL. Seemed to be a good way to do it, could restrict some of the mappings.
… Where are you with the RME version of the formats?
josh: OWL hand written translatoin from noria and rican went nowhere. Want an auto-generated representation.
… Want a formal accepted representation.
erich: Hard becaus DICOM's normative docs are XML, w updates every quarter.
… Python lib pulls it all out, but it hasn't been run in years.
josh: LMK if I can help add interest and pressure.
… OME schema is in XSD, using pydantic for converting to semantic formats.
erich; Where is DICOM at in terms of use in FHIR?
dbooth: Two potential routes: DICOM to RDF directly, or DICOM in FHIR, whch would give RDF for free.
erich: But how much of DICOM would be in FHIR?
eric: Does FHIR rep of DICOM have all of the attributes?
erich: Havne't seen.
rob: Need to talke to FHIR imaging folks.
… .FIrst sentence in the groups description mentions DICOM.
josh: Anohter reuquirement: In icroscopy want to convince vendors to build on a schema to add their own private tags. Hope there's a way to share the meaning of those.
dbooth: FHIR extensions could hold the private tags.
… Sounds like the FHIR route would be best.
erich: In DICOM community, is there a procedure for a vendor to register tags, to prevent clashes
josh: I think there's a lot of flexibility in the private tags.
erich: I don't necessarily need a complete mapping, but want completenexs for future uses.
josh: Needs to automated conversion.
rob: Imaging group: https://
ACTION: Erich to reach out to imaging group
ADJOURNED