Stream: implementers
Topic: Study instance UID in Rad. Observation
Simone Heckmann (Aug 15 2016 at 08:29):
Where would I put a Study instance UID in a radiological Observation resource?
I have plain text, which is the focus of the Observation, so I'd chose value[x]=string. But I also want to add the StudyInstanceUID to the resource. Is there a recommendation where to put this?
Simone Heckmann (Aug 15 2016 at 08:47):
We're currently discussing, using a DiagnosticReport with reference to both an Observation, holding the text result and a ImagingStudy holding the SIU. But it's a bit difficult/overkill to create an ImagingStudy Resource, when all I have is a SIU...
Grahame Grieve (Aug 15 2016 at 09:27):
it's kind of a black art, matching study id to other things. In your case, what does it identify?
Simone Heckmann (Aug 15 2016 at 09:29):
Actually, it's a parameter we'd need to call the (non-fhir) PACS webviewer to display the pics associated with the Observation.
Grahame Grieve (Aug 15 2016 at 09:29):
valueAttachment?
Simone Heckmann (Aug 15 2016 at 09:30):
nope. I already wasted my choice of values for the plain text of the observation.
Simone Heckmann (Aug 15 2016 at 09:31):
Observation.component, maybe?
Grahame Grieve (Aug 15 2016 at 09:32):
this is starting to sound like an extension then
Simone Heckmann (Aug 15 2016 at 09:34):
Yes. Quite a corner case, I guess. In a (more) perfect FHIR world, the PACS would "speak" FHIR and we'd just reference the appropriate ImagingStudy resource :)
Grahame Grieve (Aug 15 2016 at 09:58):
one day....
Simone Heckmann (Aug 15 2016 at 10:01):
Way to go :)
Eric Haas (Aug 15 2016 at 16:44):
@Simone Heckmann Can you share an example or two? I am looking for more imaging stuff to put in the DiagnosticReport examples
Elliot Silver (Aug 15 2016 at 17:32):
@Simone Heckmann Agree, it does sound a little heavy weight. Although, it should be possible to fill out a skeleton ImagingStudy . Can you file a tracker item, or ballot comment, and II WG can look at this? Thanks.
Simone Heckmann (Aug 16 2016 at 08:15):
Yes, creating an ImagingStudy is definitely possible and also necessary in case you have a 1..n relationship between Observation and SIU. But since I know nothing more of the Study than it's identifier, I'm stumbeling over some 1..1 attributes (e.g. numberOfSeries).
For cases in which we have a 1..1 relationship betwees Observation and SUI (or in most cases: Observation and AccessionNumber) I wonder if these could simply be considered as Identifiers for the Observation from the PACS's namespace?
As said: the main reason why we want to save these is that we need them to construct the URLs to access the Studies via a non-fhir REST service (DICOM viewer).
Elliot Silver (Aug 17 2016 at 21:06):
@Simone Heckmann I just noticed those attributes too. They should probably have a min cardinality of zero. Submitted #GF10463.
By the way, to construct the URL, you may need more than the studyUID. For example, you may need to know which server you can get these particular images from. The baseLocation element in ImagingStudy is meant to address this. If the viewer you're invoking isn't IHE IID compliant, use the extensible baseLocation.type coding, to put in a value to identify "Simone's viewer" URLs.
I'd be interested to hear more about your use cases and what's needed to make them work better.
Elliot Silver (Aug 17 2016 at 21:06):
@Simone Heckmann I just noticed those attributes too. It seems to be an oversight, and they should probably have min cardinality 0. I've filed #GF10463.
I think ImagingStudy is what you want to use. In many cases having the Study UID isn't enough-- you'll need to know which server has that particular study. The baseLocation is designed exactly to support invoking a viewer on the study. (If your viewer isn't an IHE IID-compliant viewer, you can extend the baseLocation value set to include "Simone's viewer".)
I'd be interested in hearing more about your use cases and what's needed to support them.
Simone Heckmann (Aug 22 2016 at 14:43):
Well, I think my use case just blew up. Turned out, that the Accession number can be used just as well as a parameter for accessing the pacs viewer and that the accession number also happens to be the local identifier of the DiagnosticReport that RIS sends to HIS. So...case closed.
+1 for the Change request on the cardinalities of ImagingStudy, though.
Last updated: Apr 12 2022 at 19:14 UTC
