ABCDEFGHIJKLMNOP
1
TimestampSubmitter NameSubmitter's Email AddressDocument NameVolumeLine NumberSection NumberIssueProposed ChangePriorityResolution?
2
8/12/2020 7:21:14Thomas SchwereAI Workflow for Imaging (AIW-I)Volume 179250.4.1.6Wouldn't it be helpful to somehow link the UPS subtasks to their main UPS task? Without this information it's not possible to dig into the details about the subtasks that have been created while performing the main UPS task. The only information that's available would be the workitem codes if the performer additionally annotates the workitem codes of the UPS subtasks in the performed workitem codes of the UPS main task.Add UPS linking/grouping information in the UPS (potentially requiring a DICOM CP). The same issue was also discussed couple of times already in the IHE-RO domain, i.e. how to group multiple UPS tasks/subtasks that need to be performed together in a single treatment session/patient visit at the delivery device. Medium
3
3/21/2021 2:22:34Christopher J Roth*White Paper: AI Interoperability in Imaging-comments due 2021-04-11My name as writtenChange my name to Christopher J Roth in the author list (there is a Chris Roth, radiologist at Thomas Jefferson and we get mixed up often). Really I'm just putting this in because I wanted to work through the comment form at least once as I'm sending it far and wide. Low
4
3/22/2021 4:01:04Koen Vergote*White Paper: AI Interoperability in Imaging-comments due 2021-04-113222.1.4This is perhaps another task that could be considered in the list:
* removing normal anatomical structures from an image, in order to improve conspicuity of abnormal structures (e.g. bone removal on x-rays, vessel suppression on CT chest exams)		See aboveMedium
5
3/22/2021 5:01:33Koen Vergote*White Paper: AI Interoperability in Imaging-comments due 2021-04-1115823.4.5Figure 3.4.5-1: I believe in many scenarios the PACS may act as the AI orchestrator, routing studies to the connected AI algorithms.I would propose to see another option in this figure, where modality sends to PACS, and PACS communicates with AI... this may be even be the preferred workflow in many cases.Medium
6
3/22/2021 11:49:51Felipe Campos Kitamuta*White Paper: AI Interoperability in Imaging-comments due 2021-04-11105-1153.6 - 3.7It seems there is no item describing how to communicate with the end user how to use the AI results. I think between sections 3.6 and 3.7 there should be an item describing how to communicate with the end user how to use the AI results. Maybe this was covered somewhere else in the document and I missed it.Medium
7
3/23/2021 9:41:27Daniel Rubin*White Paper: AI Interoperability in Imaging-comments due 2021-04-112530: Encoding annotation data elements3.8.4Encoding qualitative image findings in form of coded concepts and other items are proposed to be encoded in Comprehensive 3D SR Storage. Recently DICOM-SR TID 1500 was introduced into the DICOM standard precisely to encode this information, and this is the preferred format for storing this info for interoperability. Other encoding formats are also discussed in this section that can encode this information, and to maximize interoperability, a single format (TID 1500) should be advocated.I propose that qualitative image findings in the form of coded concepts, as well as coordinate-based ROIs should be encoded using DICOM-SR TID 1500. This format should be advocated as the preferred format for encoding this information to promote interoperability. Annotations that are informally recorded using textual labels and NIFTI should be explicitly discouraged.High
8
3/24/2021 15:06:06Julian Marsahll*White Paper: AI Interoperability in Imaging-comments due 2021-04-1113412.1.5Your vision is a bit simplistic, sort of at the "thing" level. Need to expand to include combinations of things.Include examples like: position of the end of a tube relative to some anatomic marker. Comparison of locations of findings in multiple images (e.g. a lesion in a LCC mammo compared to same lesion in an LMLO mammo), plus the obvious extension to temporal comparison.Medium
9
3/24/2021 15:08:23Julian Marshall*White Paper: AI Interoperability in Imaging-comments due 2021-04-1113652.1.5Your example on registering images is too simple.Incorporate elastic registration -- the registration of images may require complex geometric transformations, especially in the case of malleable organs (such as the breast).Medium
10
3/24/2021 15:15:02Julian Marshall*White Paper: AI Interoperability in Imaging-comments due 2021-04-1113392.1.5You are missing a concept. AI can / might determine whether the image is "appropriate" for the algorithm.Add a discussion around AI that predetermines whether or not the image or study is appropriate for the algorithm or a different algorithm. A simple example is breast AI that is not trained to process images with an implant in the view. An implant prescreen AI could identify whether or not a breast implant is present in the image as a triage for whether or not the image is appropriate for a cancer detection AI, for instance.Medium
11
3/24/2021 15:17:28Julian Marshall*White Paper: AI Interoperability in Imaging-comments due 2021-04-1114512.1.9Sentence is poorly worded "• Assessing the performance of individual imaging equipment to proactively monitor against machine breakdown."Sentence is poorly worded "• Assessing the performance of individual imaging equipment to predict possible near-term machine breakdown and trigger proactive preventive maintenance activities."Medium
12
3/24/2021 15:26:34Julian Marshall*White Paper: AI Interoperability in Imaging-comments due 2021-04-11112053.2.8A poor quality image should not necessarily be considered "bad data"Because the range of image qualities is broad, even in a heavily regulated imaging space such as mammography, it may be unwise to consider a poor quality image "bad data". Instead, the database should reflect the clinical goal. Is the goal to differentiate bad vs good images quality? Or is the goal to operate on a wide range of clinical images, in which case there should be a normal distribution of image qualities in the training (and testing) data sets.Medium
13
3/24/2021 15:38:31Julian Marshall*White Paper: AI Interoperability in Imaging-comments due 2021-04-11113133.3.3Identification of the model following training is not mentionedEarlier in the document you talked extensively about identifying and controlled the provenance of data sets. But it is equally true with the algorithm itself. AS SOON AS a model is trained, the model, source code, parameters, weights and any other controlled factors should be locked down and controlled. They should be collectively named, dated, and a very unique version number applied. Any change any of the controlled factors must result in a different version number being applied.High
14
3/24/2021 15:44:21Julian Marshall*White Paper: AI Interoperability in Imaging-comments due 2021-04-11113473.3.4You talk only about the appropriateness of the validation datasetsThe model that is to be tested, incorporating all of its associated controlled factors, should be identified by name and version (or however it can be uniquely identified). Similarly, the test data set should be similarly identified. It is the combination of model ID and test data set ID that leads to a specific set of results (which by inference have a unique model ID / test data set ID combination). And yes, rerunning the test should net the same result for AI.High
15
3/24/2021 15:46:35Julian Marshall*White Paper: AI Interoperability in Imaging-comments due 2021-04-11113653.3.4This bullet is counter to my prior comment and will lead to confusion.If the model is "from scratch", "transfer learning" or "fine-tuning / localization" trained, I would submit that in fact those result in different models. Once again, each should be identified and version controlled to avoid ambiguity.High
16
3/24/2021 22:19:33Tomo Araki (IHE-J)*Contrast Administration Management (CAM)- comments due 2021-03-25Volume27804.Y2.4.1.2 Message SemanticsPlease specify the tree structure of the sequence tag.
We would like to be clear about which tags in the sequence tags we should to target		Description of sequence tag tree structure. (target tag number)Low
17
3/25/2021 3:49:05Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Rev. 1.0 Draft PC165Open Issues, Q5Is Storage Commitment essentially needed, since IAASRs are no documents needed for diagnose finding. So they are different from image type DICOM Objects. Well one may discuss this from a technical point of view.Low
18
3/25/2021 4:57:36Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC165Open Issues, Q10Selecting by technologist (name) is definitly one feature used in QA software, so this should be solved.IAASR uses EV(121008, DCM, "Person Observer Name") within TID 1003. This ist shown in part 17 example MMMM. TID 1003 can define the EV(121010, DCM, "Person Observer's Role in the Organisation") by CID 7452 to be a "Radiologic Technologist" (SCT code 159016003) or "Lead Radiologic Technologist" (DCM code 128674). Furthermore EV(121010, DCM, "Person Observer's Role in this Procedure") can define by CID 7453 the observer to be "Performing" (SCT code 121094) or "Assisting" (DCM code 121099). This seems to be sufficiant and and could be mapped into the SR Document General Module -> Partipiciant Sequence (0040,A07A) with Observer Type (0040,A084) "Person" (PSN) to "Person Name" (0040,A084) when setting the "Organizational Role Code Sequence" (0044,010A) according to the role values above.High
19
3/25/2021 7:01:03Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC240Volume 1, X CAM ProfileIt should be stated out here that CAM dose not only support administrations by power injectors but also inhalated and/or swallowed agents which may be self administered. This is stated out in the Closed Issues section, last Question on page 9, bottom " Q. Which imaging agent sdministrations are addresses by the profile? A. Primarily Injected contrast agents (... others are not prohiited)", but I assume this will be removed in the final document.Add a second Note which replicates Q&A of page 9 bottom.Medium
20
3/25/2021 7:17:33Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PCBetween 230 an 235Appendix D - GlosseryRow "Contrast Administration" Column "Definition": "... contrast agent by an infusion device in the context ..." sonmehow stands in contradiction with last Q&A of closed issues on page 9, whereas inhalted and/or swallowed agents where also covered. Remove words: "... by an infusion device ... "High
21
3/25/2021 7:38:51Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC275 belowTable X.1-1: CAM Profile - Actors and TrasactionsRow "Infusion Manager", 2nd sub-row, "Storage Commitment" has Optionality "R"It might be possible to relax to "O"Medium
22
3/25/2021 7:46:36Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC282 / 282X.1.1.1 Infusion ManagerAgain Problem with specializing on power injectorsChange sentence the following way: "Infiusin Managers output contrast information (IAASR instances) on behalf of the administration process with which they are associated. This may but must not be a power injector."Medium
23
3/25/2021 7:52:44Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC306X.1.1.1 Infusion ManagerThis adresses the use case of how patient information has to be input to the Infusion Manager. Since CAM level 1 will not address the problem of getting patient data from e.g. an MWL, this seem to be not special protocol case but defines a system property. At this level, the user is also able to input the corresponding information, since there are no special IAASR attributes for phantom/calibration purposes.Remove sentenceHigh
24
3/25/2021 7:52:50Antje Schroeder *Contrast Administration Management (CAM)- comments due 2021-03-251415X.4.1.4Later in X.4.2 , you require the images and the IAASR to be in the same study. Nevertheless how do you anticipate Study Instance UIDs being coordinated between the modality and the Infusion Manager, when there is no Modality WorklistMedium
25
3/25/2021 7:54:06Antje Schroeder *Contrast Administration Management (CAM)- comments due 2021-03-2526664.Y1.4.1.2According to Section x.4.1.1 (around line 346) you are stating that the Planned IAASRs are not covered in this profile yet. However you state above that the Sender shall support one of the two SOP Calasses. So theoretically you could create a sender, that does not really work iwith the profile. Please ensure consistency
Low
26
3/25/2021 7:55:56Antje Schroeder *Contrast Administration Management (CAM)- comments due 2021-03-2527217214.Y1.4.1.3Which UID, do you refer to (130196, DCM, Imaging Agent Adminstration Performed Step UID)Add appropriate concept nameLow
27
3/25/2021 8:00:15Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC325X.2 CAM Actor OptionsTable X.2-1 Row "Infusion Manager", Column "Option Name" says "No options defined". Question: wouldn't it be necessary to be consitent to have the option "Storage Committment"?---Medium
28
3/25/2021 12:29:15Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC360-364X.4.1.3 Analysis & ReportingFrom the view point of quality management two key grouping modes were not addressed:
- operational unit
- radiology technologist		Add "operational unit, radiology technologist, " after "modality, " in line 361 and line 363High
29
3/25/2021 12:38:34Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC371-373X.4.1.3 Analysis & ReportingGiven example "(e.g., were there any differences between the procedure as scheduled and the procedure as performed, was the difference appropriate or inevitable, and was it appropriately approved)" sounds good, but cannot really work without processing of an IAASR plan, which defines the procedure scheduled. But definition of this workflow was defined out of scope of this document.Remove this exampleHigh
30
3/25/2021 13:59:22Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC449-450X.4.1.6 Relating Performed Contrast Administration and ImagesMake sentence more specific:Change sentence to: "Contrast Information Consumers are recommended to preferentially use contrast information from Performed IAASR instances and not contrast/bolus modules."High
31
3/25/2021 14:06:37Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC468-474X.4.2.1.1 Uneventful Case DescriptionAgain this describes a workflow which is not adressed by CAM Level 1 but Level 2. Permanently switching this might confuse the first time ingenuous reader.Disclaim from referencing to the worklist and leave open how one will get this information.Low
32
3/25/2021 14:12:19Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC519X.4.2.2.1 Extravasation Case DescriptionMake sentence more specific with respect how to build the " ... single IAASR instance ...".Add "with two steps" after " ... single IAASR instance ...".High
33
3/25/2021 14:44:50Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC548-565X.4.2.4 Use Case #4: Repeat Prior Protocol CaseTwo major issues:
1. This case describes a situation in which the Infusion Manager also gets the role of an Infusion Consumer. This was not described elsewhere and specifically not in section X.1, Figure X.1-1 CAM Actor Diagram.
2. The case of retrieving of information from a Performed IAASR from the Image Manager/Archive and loading it back into the Injector Device needs a similar but different protocol "compilation" process as "compiling" an IAASR Plan into an executable injector protocol. But there is a difference. Whereas the Plan IAASR has the attribute EV(130445, DCM, "Imaging Agent Adminstration Step Sequence Number") this is not part of a Performed IAASR. So the Infusion Manager needs some tricky (vendor specific) logic to compile a perfomed IAASR back into a protocol for injection.
3. This use case contradicts with the statement in paragraph lines 577-579. Due to inter vendor specific incompatibilities it might lead to patient security issues.
4. The idea behind seems to be to generate comparable injection situation.		1st choice: Remove section with use case #4 and defere topic until describing Level 2.
2nd choice: Try to describe process by using IAASR Plans. Indeed this needs to redifine Roles of Infusion Manager and/or to (re)define the role of a "Contrast Information Consumer".		High
34
3/25/2021 14:48:51Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC588-590X.6 CAM Cross-Profile ConciderationsIAASR does not only describe infusion and agents but also consumeables, which are relevant to Charge Posting. This is not adressed.Add "and consumables" after "... imaging agent usage"Low
35
3/25/2021 15:00:21Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC595-600X.6 CAM Cross-Profile Considerations -> PDI - Portable Data for ImagingDefinitely true what is said, but this view to CAM needs to have the same functionality as a viewer in order to combine images and IAASRs (in this special case even automatically and not user guided). The paragraph starting at line 414 defers this to "Phase 2".

[Sry I named this Level 2 throughout my comments so far.]		DOn't know. Remove PDI?Medium
36
3/25/2021 15:05:41Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC601-606X.6 CAM Cross-Profile Considerations -> TCE - Teaching Files and Clinical Trial ExportTrue, but this this needs also an automatic matching strategiy als for the PDI issue in sction X.6. Same arguments here.Same solution as for the PDI issue in section X.6.Medium
37
3/25/2021 15:26:04Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC640-6454.Y1.2, Table 4.Y1.2-1: Actor RolesAs Infusion Managers act as Senders it seems odd that they may send Planned IAASRs. For all explanations so far in the document Infusion Managers act mainly als data collection and report generator instances.
It would be plausible that the will also act as a kind of Consumer for plans (this is not stated out directly anywhere, one can assume, when reading 4.Y3 Retrieve Contrast Information).

Ok, I know what ist meant, but the problem is, that there is nothing said about the role of IAASR Plans and some roles as Infusion Planners and Infusion Plan Consumers.		Define roles Infusion Planner and Infusion Plan Consumer. Perhaps the idea of a Contrast Information Provider is more generic.High
38
3/25/2021 15:35:45Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC6704.Y1.4.1.2 Message SemanticsDescribe also how an Infusion Planner will behave.Insert the following sentence: "An Infusion Planner shall integrate all steps of an Administration plan and theire relevant details in a given order into one IAASR Plan."High
39
3/25/2021 15:44:53Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC6954.Y1.4.1.2 Message Semantics -> Table 4.Y1.4.1.2-3 Contrast Administration AttributesRow "Performed Procedure Code Sequence", Column "Requierements", last sentence: Shall this be a functional description of the flow of information? If yes it leads to a necessary information flow from the Modality to the Injector. This must not be mandatory. It may be one way of inplementation.Remove sentence.High
40
3/25/2021 15:49:01Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC6954.Y1.4.1.2 Message Semantics -> Table 4.Y1.4.1.2-3: Contrast Administration AttributesRow "Referenced Request Sentence...", Column "Attribute Name": Typo with the ">" ???UncertainHigh
41
3/25/2021 15:51:55Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC6974.Y1.4.1.2 Message SemanticsReference to RAD TF-2 seems to be wrongI found: 4.6.4.1.2.3.4High
42
3/25/2021 15:55:32Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC7044.Y1.4.1.2 Message SemanticsTID 11020 is wrong. This is only the root.Better use TID 11004High
43
3/25/2021 16:03:48Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 Draft PC720-7214.Y1.4.1.3 Expected ActionsThis is a very nesty action for a consumer since it needs to make a "full table scan" for all IAASRs of the Image Manager/Archive. We should avoid this (otherwise we don't have to think about performance of C-FIND requests and smart indexing the IAASRs by DICOM tags).Bettes state out something like "Avoid fragmentation of IAASRs." or "There may be a aggregation instance / proxy for IAASRs" (oh, yes here we could use storage committment).High
44
3/25/2021 16:52:57Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC782-7854.Y2.4.1.2 Message Semantics -> Table 4.Y2.4.1.2-2 IAASR Instance Specific Query Matching and Return KeysSome additional index attributes were needed to:
1. Fast filter for a Readiologic Technologist.
2. Fast filter for a specific examination unit (room).
3. Fast filter for a modality type, where an injector is associated with. 		1. Person Name (0040,A084) in the Participiant Sequence (0040,A07A) of the SR Document General Module. In the order of the table shall have: O, R+, O, R+. Fill from TID 1003, row 1 EV(121008, DCM, "Person Observer Name")

There seems to be the need to define a new term TECHNOLOGIST or AUTHOR in the DICOM standard, since they cannot be found in Part 3 Section C.17.2.5. See also issue open issue Q10. -->> LEADS TO DICOM CP

2. Station Name (0008,1010) in the General Equipment Module. In the order of the table shall have O, R+, O, R+. Fill from TID 1004, row 10 EV(121013, DCM, "Device Observer Name").

3. Modality (0008,0060) but the one in the SR Document Series cannot be used. -->> LEADS TO DICOM CP		High
45
3/25/2021 17:02:59Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC8114.Y2.4.1.2.1 Filtering StrategiesNote -> 2nd sentence: "... to confirm which IAASR instances correspond ...". This does not describe the two level process in detail. Replace with: "... to first confirm which IAASR instances correspond to which images/modalities and second which step(s) within a IAASR correspond to which images/modalities."Medium
46
3/25/2021 17:14:10Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC8424.Y2.4.1.2.1 Filtering StrategiesWhat is an "Instance Attribute"? ---Low
47
3/25/2021 17:23:08Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC9134.Y3.4.1.2 Message SemanticsC-GET missed here. If an Injector Device incorporates an Infusion Manager. C-GET seems to be more appropriate, since it's slimmer implementation (no MOVE SCP needed).Specify C-GET also. Change 4.Y3.4.1.3 "Expected Actions" and 4.Y3.4.2.1 "Trigger Events" appropriately.High
48
3/25/2021 17:31:27Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC925-9264.Y3.4.2.2 Message SemanticsWhat happens, if the Requester requests multiple Documents and they were of different SOP Classes ---Low
49
3/25/2021 17:35:27Uwe Tronnier*Contrast Administration Management (CAM)- comments due 2021-03-25Revision 1.0 - Draft PC939-9454.Y3.4.2.3 Expected ActionsSame issue as in line 720. Aggregate IAASRs at a global level seem to be very time consuming.Avoid fragmentationMedium
50
4/10/2021 16:10:33Eugene Igras*White Paper: AI Interoperability in Imaging-comments due 2021-04-111. Consider expanding the scope of the document to include the concept of Explainable AI (XAI) beyond the use case described in 3.6.3 Perform Inference.

Rationale:
The healthcare sector relies heavily on information technology, and increasingly on AI solutions. As adoption of AI becomes more mainstream, a concern has emerged about the extent to which AI systems make complex computations and arrive at conclusions without the users being able to objectively understand how, or why and verify the results. The traditional ‘black box’ model of AI is an example of an application that is not transparent regarding the assumptions, computations, and decision-making.

References (examples):
 Explainability for artificial intelligence in healthcare: a multidisciplinary perspective. https://bmcmedinformdecismak.biomedcentral.com/track/pdf/10.1186/s12911-020-01332-6.pdf
 Explainable AI meets Healthcare: A Study on Heart Disease Dataset.
https://arxiv.org/pdf/2011.03195.pdf
 Explainable AI in Healthcare
https://ieeexplore.ieee.org/document/9139655
 Explainable AI in Healthcare and Medicine: Building a Culture of Transparency and Accountability.
https://www.springer.com/gp/book/9783030533519

2. Consider expanding the document to include methods of measuring the interoperability of systems to accommodate interoperations between/among AI and non-AI solutions.

Rationale:
While the document covers the interoperability topics extensively, it focuses primarily on the technical interoperability aspects (which loosely correspond to the foundational and structural interoperability levels, as described at the HIMSS portal https://www.himss.org/resources/interoperability-healthcare. Consider expanding its scope to (a) include the remaining interoperability levels and, (b) exploring the application of interoperability frameworks and measurement methods used in other sectors.

References (examples):
 Interoperability in Healthcare
https://www.himss.org/resources/interoperability-healthcare
 Determining the Measures of Success for Interoperability
https://www.himss.org/resources/determining-measures-success-interoperability
 A framework for semantic interoperability in healthcare: a service-oriented architecture based on health informatics standards
https://pubmed.ncbi.nlm.nih.gov/18487823/
 Quality framework for semantic interoperability in health informatics: definition and implementation
https://core.ac.uk/download/pdf/79547379.pdf
 An Approach for Enterprise Interoperability Measurement http://ceur-ws.org/Vol-341/paper1.pdf
 Interoperability Measurement
https://scholar.afit.edu/etd/2643
Medium
51
4/12/2021 3:37:27Collège des Enseignants de Radiologie de France *White Paper: AI Interoperability in Imaging-comments due 2021-04-112.1 Applications of AI in ImagingAdditional type of AI in Imaging (can be include may be in 2.1.8 Population Health)Global information on an individual patient's health for prevention (free insight):

there is a lot of additional information which can be found on images beyond the disease or clinical question, and allow identification of risk factors which may benefit from prevention (pulmonary emphysema, osteoporosis, sarcopenia/denutrition, vascular calcifications, dental infection, kidney size, etc...). AI could quantify automatically and deliver a global health report in addition to the radiologist's report.

On a population scale, these data may indicate a population's global health and allow projections on the number of expected heart attacks, dialysis, bone fractures, etc...
Medium
52
4/12/2021 3:42:29Collège des Enseignants de Radiologie de France (CERF)*White Paper: AI Interoperability in Imaging-comments due 2021-04-11470 - 4732.2.1Removal - Radiologist must be strictly defined Radiologist: Provides differential diagnoses on imaging studies and perform image-guided
therapeutic procedures. Increasingly asked to integrate AI into image interpretation tasks.
Remove :
Note:
Use cases that reference “Radiologist” are often also applicable to other clinicians who use
imaging.		High
53
4/12/2021 3:45:35Collège des Enseignants de Radiologie de France (CERF)*White Paper: AI Interoperability in Imaging-comments due 2021-04-113.5.3AddWe should include in the validation by user the analysis of the indirect effect of the AI on patients not concerned by the algorithm used. Indeed, there could be a bias with patients diagnosed or triaged by AI algorithms being given priority over other patients, simply for "opportunistic" reasons (attention drawn to them by alerts of the AI software, or pleasure or confort of using a diagnostic aid), possibly diminishing the latter patients' chance for timely management.Medium
54
4/13/2021 2:00:23Andy Wilson*White Paper: AI Interoperability in Imaging-comments due 2021-04-1113893.4.1,3.4.4,4.7.1"Docker file" is quite a loose term"OCI-complaint container image" is a more accurate term to use. OCI containers can be run in a range of runtime systems, of which one is Docker (others include Podman, Kubernetes, etc.). See https://opencontainers.org/ for more information.Medium
55
4/13/2021 12:34:43Graham King*White Paper: AI Interoperability in Imaging-comments due 2021-04-11126143.8.5 TransformsSpatial registration is mentioned once in 3.2.3 but not mentioned here. It is an appropriate use case for DataSet Assembly. At present, a lot of imaging AI algorithms are having to do this as part of their preprocessing before data enters the models ; it would be more efficient for the industry as a whole for this to be incorporated earlier in the pipeline and performed by a common actor prior to data entering models. Add Spatial Registration as an appropriate type of transformMedium
56
4/13/2021 15:45:49Sally Baxter*White Paper: AI Interoperability in Imaging-comments due 2021-04-11Revision 1.0 - Public Comment260-2812.1.1 Ordering and SchedulingThe section on Ordering and Scheduling is specifically focused on imaging orders, which is appropriate given the focus on radiology. However, for clinical practices such as ophthalmology, AI could also be used for automating or suggesting medication orders (e.g. AREDS2 vitamin supplementation for a patient without an existing medication order who meets eligibility criteria based on retinal imaging), procedure orders (i.e. intravitreal injection), or automating follow-up orders (e.g. order for follow-up appointment in 1 year for diabetic patient who requires annual screening examinations). Upon further review of the white paper, this is mentioned in section 2.1.7 but should be linked to this section for clarity.
Addition of a bullet point to section 2.1.1 to state, “Suggesting orders related to care or treatment plans, as detailed in section 2.1.7 Patient Management and Treatment Planning.”
Low
57
4/13/2021 15:46:46Sally Baxter, MD*White Paper: AI Interoperability in Imaging-comments due 2021-04-11Revision 1.0 - Public CommentN/AN/A“Procedures” in this white paper generally appear to refer to imaging procedures, rather than interventional procedures. In specialties such as ophthalmology, “procedures” often refer to interventions (e.g. injections, lasers, surgery, etc.) rather than imaging. It may be useful to distinguish these entities given that they entail distinct workflows.
Use “imaging procedures” to indicate imaging, and perhaps a distinct term such as “interventional procedures” to distinguish these from imaging procedures. There are a few instances of “interventional imaging procedures” included in the white paper, but perhaps it would be worth making this broader to “interventional procedures,” since specialties may be engaged in interventional procedures that are not imaging-based per se.
Medium
58
4/13/2021 15:47:34Sally Baxter, MD*White Paper: AI Interoperability in Imaging-comments due 2021-04-11Revision 1.0 - Public Comment430-4412.1.8 Population HealthRecommend adding the use of AI for automating bulk orders for cohorts of patients.Adding a bullet point stating “Facilitating or automating bulk orders for aggregate patient populations (e.g. yearly retinal imaging for patients with diabetes).”Medium
59
4/13/2021 15:48:27Sally Baxter, MD on behalf of April Maa, MD*White Paper: AI Interoperability in Imaging-comments due 2021-04-11I do not have specific changes or comments and I focused on Section 2. I thought the list and descriptions of uses described in section 2 were very well done and comprehensive. They are general enough that I think ophthalmology could also utilize AI in the same way for telehealth, images, etc.

The only general comment I had was the following about AI test sets. I strongly believe this should be specifically mentioned. Any data set used for training, testing, and validation, should be representative from an age, race, gender perspective, of the population that the AI will be used on. For example, for dermatology, rashes show up differently in darker skinned individuals. Therefore, an AI program trained predominantly on light skinned people may not work for darker skinned individuals. Therefore, to really ensure applicability and equality of the AI program for all patients, the data sets used must be as diverse as the patient population. For eye images, we should have informatics and fundus images from a diverse ethnic background as well as different age groups, gender, rural vs urban, household income, other social factors because these factors may influence outcomes. If we are going to use AI as part of a predictive model, then these non-medical, social, factors may play a role.
Medium
60
5/2/2021 15:23:58BERTINIWhite Paper: AI Interoperability in ImagingRevision 1.0 – Public CommentGeneralThe number of data (examination) used to create a model should be known by the final user (for reliability)Additional informationMedium
61
5/2/2021 15:54:02BERTINI CristinaWhite Paper: AI Interoperability in ImagingRevision 1.0 – Public Comment7703.1.3 Contribute DataThe country generating the data set should be specified (some pathologies are more frequent in some regions or countries)Add informationMedium
62
5/2/2021 16:02:39BERTINI CristinaWhite Paper: AI Interoperability in ImagingRevision 1.0 – Public Comment5502.2.2 SystemsThe Imaging modality should also specify the Dicom IOD used. For example, Enhanced CT ou Enhanced MR don't give the same possibilities of the other CT and MR IODAdditional informationMedium
63
4/7/2022 10:59:03Neil Tenenholtzneil.tenenholtz@gmail.com*AI Results - Extensions (AIR+)-comments due March 29, 2022L4846.5.1Acyclic directed graph, while not incorrect, is not the common term of art."acyclic directed graph" --> "directed acyclic graph" (https://en.wikipedia.org/wiki/Directed_acyclic_graph)Low
64
4/7/2022 11:07:38Alexander Goelalex@junipercds.com*Interactive Multimedia Report (IMR)-comments due 2022-03-18https://profiles.ihe.net/RAD/IMR/volume-1.html#1xx24-external-iid-image-display-retrieve-optionhttps://profiles.ihe.net/RAD/IMR/volume-1.html#1xx24-external-iid-image-display-retrieve-optionhttps://profiles.ihe.net/RAD/IMR/volume-1.html#1xx24-external-iid-image-display-retrieve-optionThe use case 3 link is broken: [Use Case #3] (#xx423-use-case-3-consume-and-interact-with-multimedia-report-by-report-reader-with-integrated-invoker-image-display)make this a link to the use caseHigh
65
4/7/2022 11:11:37Alex Goelalex@junipercds.com*Interactive Multimedia Report (IMR)-comments due 2022-03-18https://profiles.ihe.net/RAD/IMR/volume-1.html#1xx25-dicom-instance-retrieve-optionhttps://profiles.ihe.net/RAD/IMR/volume-1.html#1xx25-dicom-instance-retrieve-optionhttps://profiles.ihe.net/RAD/IMR/volume-1.html#1xx25-dicom-instance-retrieve-optionLink to Profile. Link to the IHE Profile where these transactions are fromLink to Profile. Link to the IHE Profile where these transactions are fromLow
66
4/7/2022 11:15:34Alex Goelalex@junipercds.com*Interactive Multimedia Report (IMR)-comments due 2022-03-18https://profiles.ihe.net/RAD/IMR/volume-1.html#1xx51-security-considerations-for-actorshttps://profiles.ihe.net/RAD/IMR/volume-1.html#1xx51-security-considerations-for-actorshttps://profiles.ihe.net/RAD/IMR/volume-1.html#1xx51-security-considerations-for-actorsLink to ITI ATNA profile Add link to referenced ITI ATNA profile Low
67
4/7/2022 11:16:35Alex Goel*Interactive Multimedia Report (IMR)-comments due 2022-03-18https://profiles.ihe.net/RAD/IMR/volume-1.html#1xx6-imr-cross-profile-considerationshttps://profiles.ihe.net/RAD/IMR/volume-1.html#1xx6-imr-cross-profile-considerationshttps://profiles.ihe.net/RAD/IMR/volume-1.html#1xx6-imr-cross-profile-considerationsLink to cross profiles Link to the profiles pdfs or IGs as applicable to make it easier for implementers to find information Low
68
4/7/2022 11:19:33Alex Goelalex@junipercds.com*Interactive Multimedia Report (IMR)-comments due 2022-03-18https://profiles.ihe.net/RAD/IMR/RAD-Y1.html#24y13-referenced-standardshttps://profiles.ihe.net/RAD/IMR/RAD-Y1.html#24y13-referenced-standardshttps://profiles.ihe.net/RAD/IMR/RAD-Y1.html#24y13-referenced-standardsDICOM Missing from referenced standards add link to DICOM spec (is it necessary since it's only referenced through WADO?)Low
69
4/7/2022 11:23:05Alex Goelalex@junipercds.com*Interactive Multimedia Report (IMR)-comments due 2022-03-18https://profiles.ihe.net/RAD/IMR/RAD-Y1.html#24y1412-message-semanticshttps://profiles.ihe.net/RAD/IMR/RAD-Y1.html#24y1412-message-semanticshttps://profiles.ihe.net/RAD/IMR/RAD-Y1.html#24y1412-message-semanticsTransaction should be submited to [base] end point not bundlehttps://hl7.org/fhir/http.html#transaction Transaction should be submitted to the base endpoint not to base/bundle - most servers will save the whole bundle if submitted to base/bundle, but will only process the transaction if bundle is submitted to base High
70
4/7/2022 12:00:27Alex Goelalex@junipercds.com*Interactive Multimedia Report (IMR)-comments due 2022-03-18https://profiles.ihe.net/RAD/IMR/RAD-Y1.html#24y1422-message-semanticshttps://profiles.ihe.net/RAD/IMR/RAD-Y1.html#24y1422-message-semanticshttps://profiles.ihe.net/RAD/IMR/RAD-Y1.html#24y1422-message-semanticsNo mention of failure errors (400s) Mention how to handle 400s. 200 and 300s are clearly described, but error handling is not. Medium
71
4/7/2022 12:01:58Alex Goelalex@junipercds.com*Interactive Multimedia Report (IMR)-comments due 2022-03-18https://profiles.ihe.net/RAD/IMR/RAD-Y1.html#24y15-security-considerationshttps://profiles.ihe.net/RAD/IMR/RAD-Y1.html#24y15-security-considerationshttps://profiles.ihe.net/RAD/IMR/RAD-Y1.html#24y15-security-considerationsUser AuthenticationShould IUA be mentioned to authorize the transaction between 2 FHIR Servers? A link to the spec would be sufficientLow
72
4/7/2022 12:11:45Alex Goelalex@junipercds.com*Interactive Multimedia Report (IMR)-comments due 2022-03-18https://profiles.ihe.net/RAD/IMR/StructureDefinition-imr-servicerequest.htmlhttps://profiles.ihe.net/RAD/IMR/StructureDefinition-imr-servicerequest.htmlhttps://profiles.ihe.net/RAD/IMR/StructureDefinition-imr-servicerequest.htmlAdd description of this ServiceRequest ProfileAdd a description of this FHIR profile in particular, though all profiles would benefit from descriptions. Some text about when what order this ServiceRequest represents would be helpfulLow
73
4/15/2022 18:46:08Ryan Yoderryoder@epic.com*Interactive Multimedia Report (IMR)-comments due 2022-03-1811:XX.1-1The Actors and Transactions table currently lists RAD-Y5 as required, but not all Report Readers will be able to display images natively.Update the Actors and Transactions table so the Report Reader is able to support either RAD-Y5 or RAD-106 (as the initiator).Medium
74
4/15/2022 18:47:24Ryan Yoderryoder@epic.com*Interactive Multimedia Report (IMR)-comments due 2022-03-1811:XX.4.2.4.1A Report Reader may be able to display current and prior/comparison studies but also utilizes an Image Display Invoker because it is not an Image Display itself (see Use Case 3). This scenario is not represented in Use Case 4.The wording should be updated from so that after this sentence: “When the Radiologist clicks on the links, the Report Reader triggers the viewport in the Image Display currently showing the prior study to show the specific image in which the measurements are derived from.” Is a sentence like this: “If the IMR Report Reader does not natively support image viewing capabilities, then the Image Display Invoker will invoke the Image Display with the prior study’s image.”Medium
75
4/15/2022 18:47:34Ryan Yoderryoder@epic.com*Interactive Multimedia Report (IMR)-comments due 2022-03-18GeneralGeneralGeneralIs the IMR Profile expected to have the capability to fully replace transmission of IMR reports via other methods like HL7v2 interfaces? Or is this intended to be supplementary?Low
76
4/18/2022 15:54:04Ryan Yoderryoder@epic.com*Interactive Multimedia Report (IMR)-comments due 2022-03-1811:XX.2.4This section mentions that an actor can utilize Invoke Image Display [RAD-106] to pull images up on an external image display. There is also mention of a future state where CP-RAD-474 will enable "retrieve Display of Series Images functionality", but are there plans for triggering the display of a specific image within a series?Provide some additional information for CP-RAD-474. Primary concern here is that the External IID Image Display Retrieve Option will not enable invocation of specific images within a series rather than just displaying an entire series. Section 2:4.Y1.4.1.2.3.1 specifies that the series instance UID and the SOP Instance UID will be accessible via the Observation.component.valueString - is this part of what CP-RAD-474 will use/are there plans to add specifications that enable the invocation of images within a series using these identifiers? Medium
77
4/21/2022 9:23:47Kinson Hokinson.ho@arterys.com*AI Results - Extensions (AIR+)-comments due March 29, 2022A.3GeneralA.3Suggest to add a tree diagram at the beginning to show the hierarchy of the outline of the contents in the result tree. This will help people reviewing the details later.Low
78
4/21/2022 9:26:10Kinson Hokinson.ho@arterys.com*AI Results - Extensions (AIR+)-comments due March 29, 2022Open Issue #1Open Issue #1Agree that Result Tree Option should be required for Image Display.Low
79
4/21/2022 9:34:29Kinson Hokinson.ho@arterys.com*AI Results - Extensions (AIR+)-comments due March 29, 2022Open Issue #2Open Issue #2Supporting longitudinal analysis is much more involved. The system needs to keep track of prior analysis, able to determine relationship between studies and be able to correlate the new finding to the corresponding old finding. I suggest that you may describe this in the concept section, but not a required feature.Medium
80
4/21/2022 9:42:01Kinson Hokinson.ho@arterys.com*AI Results - Extensions (AIR+)-comments due March 29, 2022Open Issue #4Open Issue #4SOP Class UID is preferred. Matching on Template ID is not well supported, making it difficult to quickly find all the root results. However, since not all Evidence Creators create Root Result, Image Displays still need to query for all AIR objects and root result to catch all cases. So it may not help query at all.Medium
81
4/21/2022 9:45:23Kinson Hokinson.ho@arterys.com*AI Results - Extensions (AIR+)-comments due March 29, 2022Open Issue #6Open Issue #6Workflow reasons should be recorded separately out of the object. The system is likely to have existing audit logs or application logs that track progress status. Low
82
4/21/2022 9:51:54Kinson Hokinson.ho@arterys.com*AI Results - Extensions (AIR+)-comments due March 29, 2022Open Issue #7Open Issue #7Accept / reject of the AI results by radiologist is an internal process of the system. Quite often the AI model may detects many findings but the radiologist will only review the top 3-5. Usually it is not required to record the explicit actions of whether the rad accept or reject or ignore any AI model findings. This will add extra load to the radiologist review flow and force them to review everything and make decision, which is not what happening is practice.Medium
83
4/21/2022 9:57:59Kinson Hokinson.ho@arterys.com*AI Results - Extensions (AIR+)-comments due March 29, 2022Open Issue #10Open Issue #10Isn't this already supported by the Imaging Document Consumer?Low
84
4/21/2022 10:15:00Kinson Hokinson.ho@arterys.com*AI Results - Extensions (AIR+)-comments due March 29, 2022Open Issue #9Open Issue #9We found showing a fine grain value (e.g. score) is confusing to the user. The value does not have well-defined meaning and the user interprets the value linearly and literally which is not accurate. So we only show 'certain' and 'uncertain' instead. With this said, this can be an implementation details of the Image Display. The AI Model is free to provide whatever details it identified in the AI results.Medium
85
4/21/2022 10:18:37Kinson Hokinson.ho@arterys.com*AI Results - Extensions (AIR+)-comments due March 29, 2022Open Issue #11Open Issue #11I think this can be treated the same way by the PACS for any late objects. Usually there are mechanisms to alert about this internally in the PACS.Low
86
4/21/2022 10:27:32Kinson Hokinson.ho@arterys.com*AI Results - Extensions (AIR+)-comments due March 29, 202234613Missing a close parenthesis that match the open parenthesis before 'showing'.Low
87
6/23/2022 8:45:12Krister Valtonen (Sectra, Sweden)krister.valtonen@sectra.comAI Results (AIR)Revision 1.1 – Trial Implementation1405Table 4.137-1The attribute (0008,103F) "Series Description Code Sequence" is placed on "General Analysis Result Specific - Instance Level". However, this is a typical Series Level attribute and should reside on that level.Put (0008,103F) "Series Description Code Sequence" on "General Analysis Result Specific - Series Level".MediumCP submitted
88
8/9/2022 8:24:02Krister Valtonen, Sectrakrister.valtonen@sectra.comWeb-based Image Access110304.129.4.1.3The specification says "In addition, the Responder shall support Issuer of Patient ID (0010,0021) and Issuer of Accession 1030 Number Sequence (0008,0051) as both matching and return keys.". No issues for the base level attributes, however (0010,0021) Issuser of Patient ID is an optional return key within the (0010,1002) Other Patient IDs Sequence for the RAD-14 Query Images transaction and it would be logical to require (0010,0021) Issuer of Patient ID as a return key (but not matching key) within that sequence as well.
In the same manner (0008,0051) Issuer of Accession Number Sequence is present in the (0040,A370) Referenced Request Sequence used in transactions RAD-26 and RAD-30 and should be a required return key in that sequence.
Section 4.129.4.1.2 needs adjustment in the same manner.		See issue, above.MediumCP submitted
89
11/18/2022 11:25:23Jörg Riesmeierdicom@jriesmeier.comAI Results (AIR)321526.5.3.9Revision 1.2 of the document states that the Template ID "IHERADAIR2" could be used to identify the SR Template that was used in the Content Template Sequence (0040,A504). This is not correct since this SR Template does not have a CONTAINER Root Node (in contrast to "IHERADAIR1"). See DICOM PS3.3 Section C.18.8 and C.18.8.1.2 for details.Remove the following part from the sentence: or "IHERADAIR2"Medium
90
11/18/2022 11:32:28Jörg Riesmeierdicom@jriesmeier.comAI Results (AIR)321816.5.3.9.2In Revision 1.2 of the document, the "Content Item Descriptions" table below Table TID IHERADAIR2 states for Row 3, 4, 5 "Purpose of reference shall not be present." This makes no sense to me since the referenced Content Items actually use the Template parameter "$Concept" for the Concept Name, which is not empty when invoked from TID IHERADAIR1 or IHERADAIR2.Remove the row for "Row 3, 5, 6" from the "Content Item Descriptions" table.Medium
91
11/18/2022 11:37:06Jörg Riesmeierdicom@jriesmeier.comAI Results (AIR)321826.5.3.9.2The description for Row 5 and 6 in the "Content Item Descriptions" table below Table TID IHERADAIR2 mixes the description for an IMAGE and a WAVEFORM Content Item. The first paragraph applies to Row 5 and the second paragraph to Row 6 only.Split the description in "Row 5, 6" into two rows: one for "Row 5" and one for "Row 6".Low
92
3/29/2023 11:13:05Eric Martinericmartin@siemens-healthineers.com*Integrated Reporting Applications (IRA)-comments due 2023-04-16Volume 11.XX.4.1.1Minor changes recommended to bulleted list in this section.* Participating ... <- no change
* Subscribers do not communicate ... <- no change
* The Hub only communicates with authenticated Subscribers <- minor wording change recommended
*Subscribers can configure their subscription request to limit what types of events the Hub forwards to them. <- moved up two bullets and minor wording change recommended
* When Subscribers generate data that should be made available to other applications, or perform actions of which other applications should be aware, they publish it by sending an event request with the relevant details to the Hub <- minor wording change recommended
* The Hub forwards accepted event requests from a Subscriber to other Subscribers subscribed to that type of event <- minor wording change recommended
* Subscribers react to events ... <- no change
* It is not necessary (nor possible) for Subscribers to be aware of what other Subscribers (if any) are receiving an event they requested be forwarded by the Hub nor how other Subscribers react to the event <- minor wording change recommended
* The Hub maintains the current state of content (if any) associated with all open contexts <- minor wording change recommended
* Subscribers can request the current ... <- no change
* The Hub can simultaneously manage multiple groups of Subscribers and their associated data in different sessions <- no change
* Each session is identified by a unique “topic ID” <- no change
* The Subscriber which opens a context typically is responsible for closing that context and is informally referred to as the Driving Application. A Driving Application (or other Subscribers) may launch other applications, providing them with the address of the Hub and the topic ID so they can join the same session. <- medium wording change recommended		Low
93
3/29/2023 13:28:55Eric Martinericmartin@siemens-healthineers.com*Integrated Reporting Applications (IRA)-comments due 2023-04-16Volume 11:XX.4.1.4Typo Recommendations For example, DiagnosticReport-open represents an event that an application opens a study for reporting. -> For example, DiagnosticReport-open represents an event that an application has opened a study for reporting

It is the responsibility of any applications that are interested in such events to subscribe them. -> It is the responsibility of any applications that are interested in such events to subscribe to them.

Low
94
3/29/2023 13:36:05Eric Martinericmartin@siemens-healthineers.com*Integrated Reporting Applications (IRA)-comments due 2023-04-16Volume 11:XX.4.1.5I believe the example on timing is covering only two extremes and a "right-sized" recommendation would be helpful.In the paragraph starting "On the other hand," change:
"it is not necessary to wait until" to "it is not appropriate to wait until"

After the paragraph starting "On the other hand," add something like:

A reasonable approach could be for an application to acquire a complete measurement and perhaps some measurement characteristics, then send an event request containing this information to the Hub.		Low
95
3/29/2023 13:40:22Eric Martinericmartin@siemens-healthineers.com*Integrated Reporting Applications (IRA)-comments due 2023-04-16Volume 11:XX.4.1.6Typo Suggestions"has the knowledge of an event has happened" -> "has the knowledge that an event has happened"

"reacts to the event and performed" -> "reacts to the event and performs"

"This business logic may be automatic or requires additional user input." -> "This business logic may be automatic or require additional user input."		Low
96
3/29/2023 16:04:55Eric Martinericmartin@siemens-healthineers.com*Integrated Reporting Applications (IRA)-comments due 2023-04-16Volume 11:XX.4.1.7Consistency of resource.id approachesIn response to a FHIRcast ballot comment (JIRA ticket FHIR-36910) a PR (https://github.com/HL7/fhircast-docs/pull/483) was created that provides guidance on the creation of resource id and its reuse if the resource is added to another information sharing session. I believe the approach of:

nnn-nn-nn or Observation/nnn-nn-nn or http://myserver.com/Observation/nnn-nn-nn (as the fullUrl in the entry of the update Bundle)

interchangeably causes no issues; however, we should discuss to ensure everything is consistent. We may also wish to tweak the wording (or add some words) to this section based on that discussion. 		Medium
97
3/30/2023 8:31:13Eric Martinericmartin@siemens-healthineers.com*Integrated Reporting Applications (IRA)-comments due 2023-04-16Volume 11:XX:4.1.11Additional clarification.In the FHIRcast specification there are words that events following the pattern like *-open, *-close, *-update should be handled even if they are not listed in the event catalog. Specifically:

"FHIRcast supports all events that follow this format. The most common events definitions have been provided in the event catalog."

Hence I believe an additional sentence may be useful at the end of the first paragraph, something like:

For example, an Image Manager may setup a separate advanced visualization session with an Evidence Creator and uses ImagingStudy-* events for communication. The FHIRcast specification indicates that [FHIRresource]-(open | close | update | select) events should be supported even for FHIR resources not included in the specification's event library.		Low
98
3/30/2023 13:05:28Eric Martinericmartin@siemens-healthineers.com*Integrated Reporting Applications (IRA)-comments due 2023-04-16Volume 11:XX.4.2.1.1"Manual" Population of Report versus "Automatic"Bullet:
* Radiologist selects some of the measurements made and uses voice commands to auto-populate the report with the selected measurements

This is certainly possible and will of course work fine. Current implementations of which I am aware automatically have Image Display send a measurement when it reaches a certain state of information (e.g., a VOI is measured and then labeled as "Lung Nodule 1") and automatically sends (DiagnosticReport-update) to the Report Creator. The Report Creator then decides if and how to display this information in the report with the Radiologist accepting or rejecting the information in the Report Creator.

I don't know if the profile should get to the level of detail to mention that the Basic Reporting flow could work either way, mention both while recommending a particular approach, or simply leave as currently specified.
Medium
99
3/30/2023 13:11:16Eric Martinericmartin@siemens-healthineers.com*Integrated Reporting Applications (IRA)-comments due 2023-04-16Volume 11:XX.4.2.1.2.1Typo RecommendationChange:

"started and terminated, or it can be put in focus and minimize when not needed but keep running in the background for efficiency, or a combination"

to:

"started and terminated, or it can be put in focus and minimized when not needed but kept running in the background for efficiency, or any combination thereof"		Low
100
3/30/2023 15:25:49Eric Martinericmartin@siemens-healthineers.com*Integrated Reporting Applications (IRA)-comments due 2023-04-16Volume 11:XX.4.2.1.2.1.4Typo Recommendations"(e.g., user clicks on the measurement in a report in the Report Creator triggers the Image Display to bring the corresponding images to focus)"

to

"(e.g., user clicks on a measurement in a report in the Report Creator which triggers the Image Display to bring the corresponding images to focus)"

"highlight to the user what are selected so that the user can perform some actions"

to

"highlight to the user what is selected so that the user can perform an appropriate action"		Low