Rad Tech Minutes 2023-11-13-17
Monday, November 13, 2023
Attendees: Andrei Leontiv (Visage), Wim Corbijn, (Philips), Chris Melo, (Philips), Ana Kostadinovska (Philips), Steve Nichols (GE), Antje Schroeder (Siemens), John Paulette (Rad AI), Kinson Ho (Tempus), Kevin O'Donnell (Canon), Sean Doyle (Hyperfine), Natalia Farish (ACR), Brian Bialecki (ACR), Chris Roth M.D., David Kwan (Insygnia), Lynn Felhofer (RAD TPM), TeRhonda McGee (RSNA), Jaime Dulkowski (RSNA), Chris Carr (RSNA)
Guests: Rob Ferre M.D., Rachel Liu M.D. (ACEP), Kevin Little (OSU)
- A1: 8:45am-9:00am Administrative Time
- Introductions (Name and IHE Member Organization)
- IHE Radiology Technical Committee Roster
- IHE International Member Organizations
- IHE Patent Disclosure Statement
- Current Patent Disclosure Statements
- S0: 9:00am-9:30am: Agenda work plan review
- S1: 9:30am-10:30am: ACEP POCUS update (Steve Nichols)
- See these slides: ACEP EBIW POCUS Workflow Update
- This working group will continue to meet weekly; currently working on use cases, will especially like to consult with IHE as Vol 2 transactions will be identified to be used in the workflow. Steve and Kevin will continue to keep IHE RAD updated on progress. We can anticipate scheduling consultation tcons in the future with IHE RAD Tech.
- S2: 10:45am-12:15pm: Maintenance
- See full list of Maintenance topics
- From Top Ten list:
- CP-RAD-475 Extensions to RAD-107 for IMR (Kinson) - we reviewed and updated the content. This CP is approved as Completed and ready for ballot, pending updates agreed to during discussion
- CP-RAD-523-03 Clarify display of DBT images (Steve) - we updated the text to clarify the requirements in a different section. This CP is approved as Completed and ready for ballot,
- CP-RAD-501 WIA – RAD-129 fix in Expected Actions (Steve) - we reviewed and updated the content. Steve will make the recommended edit to the Enterprise Identity Table in RAD-14; we will review those updates this week to prepare this for ballot.
- Other CPs ready for committee review:
- CP-RAD-479-01 - RAD-60 - update reason codes reference to DICOM (Lynn) - reviewed and approved as Completed pending update to rationale.
- A2: 1:15pm-1:30pm: Maintenance
- Other CPs ready for committee review:
- CP-RAD-463 - Add RDSR Display option in REM (Kevin) - status update: This is now separated into 2 CPs, including CP-RAD-525 Add EMR Extract option in REM. These 2 CPs remain Assigned & Kevin will bring back to RAD Tech for review in the future.
- CP-RAD-496-01 - Missing arrow in SWF.b actor/transaction diagram (Lynn) - reviewed, updated, and approved as Completed.
- Other CPs ready for committee review:
- S3: 1:30pm-3:00pm: Imaging Diagnostic Report (IDR)
- Discussion of IDR use cases
- During initial profile development, the most recent version of IDR will be kept in this folder in google drive. (Note: google drive is used for doc storage, NOT for editing)
- Later, it is anticipated that this content will become a FHIR IG developed in this Github repository at https://github.com/IHE/RAD.IDR
- S4: 3:15pm-4:45pm: Reject Analysis (RAM)
- During profile development, the most recent version of RAM will be kept in this folder in google drive[
- A3: 4:45pm-5:00pm: Administrative Time
- A1: 8:45am-9:00am Administrative Time
Tuesday, November 14, 2023
Attendees: Andrei Leontiv, David Kwan, Wim Corbijn, Ana Kostadinovska, Brian Bialecki, Sean Doyle, Antje Schroeder, Wim Corbijn, Kinson Ho, Steve Nichols, Kevin O'Donnell, John Paulett, Yasunari Shiokawa-san (Salt, IHE-J), Jaime Dulkowski, TeRhonda McGee, Lynn Felhofer
Guests: David Goyard, Stephane, Nicolas Dehouck (Incepto), Natalia Farish, Kevin Little
- A1: 8:45am-9:00am Administrative Time
- S1: 9:00am-10:30am: Appropriateness of TID 1500 for AI Results (Antje, Sean)
- Refer to this presentation with notes from Antje
- Sean pointed out the issue encoding with being unable to encode classification (i.e. confidence, certainty...) of findings within TID 1500. Oncology has use cases that can inform this work for us. Kevin noted that AI cannot necessarily apply a certainty value to a finding, but AI can report how close the current result is to its training set. The AIR profile has open issue #34 with an initial write-up on this topic. We will likely need to convene a working group to address this.
- Brian questioned whether using TID 1500 is the right solution; perhaps consider using FHIR R5 Observation / ImagingSelection, instead. There was some discussion of this future direction, but Andrei asked that that discussion of this future consideration be deferred and that we spend the time i this session specifically on how to get guidance for current challenges on encoding AI results in TID 1500.
- David G. reported that they have chosen DICOM SR for encoding AI Results because a goal is to have the AI output in the PACS. The problem they have encountered that the PACS will not accept and enable viewing and editing of the SR content. Brian noted that an additional challenge is when there are many AI-related SRs in the PACS, which are relevant for the radiologist.
- Sean reported that the main challenge workflow issue
- Antje recommended perhaps making the TID 4xxx (Mammo CAD SR) templates more general.
- Kevin pointed to the AI Result Tree encoding specification and examples in the AIR Profile
- Encoding of results status -- Kevin pointed to Open Issue #32 in AIR.
- Examples:
- Sean pointed out that good examples are really valuable. He suggested a Github repository of examples.
- Antje said targeted examples that demonstrate key features (AI outputs / primitives) would be good to produce.
- Post discussion actions to address. Create ad-hoc working groups to address these topics.:
- Looking into alternatives for derived measurements (temporal, spacial)
- Flavors of certainty
- Workflow (accept...reject...). Perhaps keep a library of patterns of usage of AI algorithms.
- Be sure to engage with the AIGI task force in Europe (Marc Kammerer and others)
- Bring this topic up at the RAD meeting at RSNA. For future RAD Tech meetings, include an update from these working groups as a sub-topic for maintenance.
- S2: 10:45am-12:15pm: Imaging Diagnostic Report (IDR)
- Discuss and add details to the Concepts section in Vol e
- Discussion of HL7 V2 export option (see notes in IDR open issue #2). We will provide guidance in the profile in a hybrid environment section in Vol 1 concepts, but not a named option.
- A2: 1:00pm-1:30pm: Administrative Time
- Connectathon-related...
- Summary of RAD testing at 2023 Connectathon
- 2024 Connectathons: Feb 19-23 @ RSNA Headquarters in Oak Brook and June 3-7 in Trieste, Italy
- Connectathon-related...
- S3: 1:30pm-3:00pm: Reject Analysis
- Discussion of Open Issues in the document
- Work on actor requirements
- S4: 3:15pm-4:45pm: Imaging Diagnostic Report (IDR)
- A3: 4:45pm-5:00pm: Administrative Time
- S3: 1:30pm-3:00pm: Reject Analysis
Wednesday, November 15, 2023
Attendees: John Paulett, Kevin O'Donnell, Steve Nichols, Ana Kostadinovska, Wim Corbijn, Antje Schroeder, Kinson Ho, David Kwan, Kevin O'Donnell, Lynn Felhofer, Yasunari Shiokawa-san (Salt), Jaime Dulkowski, TeRhonda McGee, Chris Carr
Guests: Jason Malobicky (Rad AI), Natalia Farish, Charles Parisot, Kevin Little Clinical consult guests: Raym Geis, David Koff, Chris Roth, Seth Berkowitz, Adam Flanders, Tarik Alkasab
- A1: 8:45am-9:00am Administrative Time
- S1: 9:00am-10:30am: Imaging Diagnostic Report (IDR)
- Discussion of Vol 1 concepts with focus on the structure of the FHIR resources
- S2: 10:45am-12:45pm: Maintenance
- Review New Submitted CPs
- Discussion of the WIA/WIC option for XCA-I (Steve, Charles)
- Review / update Assessment of RAD profiles for promotion or retirement
- TO DO for RAD Tech -- communicate candidate FT profiles to national deployment committees as a 'last call' for CPs
- Include review of (old) criteria on the wiki for Final Text Process and retirement
- Discussion -- how to help new audience consumer RAD profiles. WIKI content is too technical for this audience. C
- If time, see Thurs maintenance topics
- Review New Submitted CPs
- A2: 1:00pm-1:30pm: Administrative Time
- RAD Retrospective Action Item discussion, including Kinson's lessons learned / IRA estimates, etc...
- S3: 1:30pm-2:45pm: Imaging Diagnostic Report (IDR)
- Working through use cases
- S4: 3:00pm-4:00pm: Imaging Diagnostic Report (IDR) - Clinical Consult
- S5: 4:00pm-4:30pm: Checkpoint assessment: Reject Analysis
- Review of effort estimate sheet
- End-of-meeting checklist -- see notes that follow:
Profile Name: Reject Analysis
- Describe gaps in Use Case coverage
- None really.
- Review ALL "uncertainty points" in the evaluation. Is there a resolution plan for each?
- Yes.
- Do the effort points in the evaluation still seem right?
- Feels good.
- Did the Breakdown of Tasks accurately reflect the work? What extra tasks arose?
- Matched well. Added consideration of un-reject; Split Reject and Quality into two use cases
- Describe unresolved technical issues/tasks
- Nothing major
- Describe potential practical issues
- Seems OK
- Review the open issue list. Does it feel complete
- Yes
- Which open issues feel most risky; what other risks exist?
- Have tentative or closed resolutions for all
- How is the work fitting in the allocated bandwidth? (Time to spare? Just right? Things were left undone?)
- Time to spare this meeting
- How does the scope feel? (Room to expand? Just right? Pretty ambitious?)
- Feels right.
- If you had to reduce scope, what would you drop?
- Other modality quality code lists (will take what we can get with not too much fuss)
- Have the promised resources manifested?
- Yes
- What tasks would benefit from additional expertise?
- Other modality codes - Kevin will check with relevant AAPM Cmtes, Kevin O will check with DICOM WG (and maybe MITA)
- What vendors are engaged for each actor? Record how many.
- Reporter - Canon, Siemens,
- Analyzer - GE
- IM/IA - Visage, GE, Philips
- Was the profile where it needed to be at the start of the Kickoff meeting (See "Kickoff Meeting" above), if not what was the gap
- Basically
- Was the profile where it needed to be at the end of the Kickoff meeting, if not what was the gap
- Basically
- How many tcons would you like between now and the PC Prep Meeting?
- Pretty good shape. Any issues can probably be brought to the Feb meeting.
Thursday, November 16, 2023
Attendees:
Guests:
- A1: 8:45am-9:00am Administrative Time
- Photo of (most of) the group
- A1: 8:45am-9:00am Administrative Time