Multilevel Q/R

[dsnishimura]

Hello everyone!

I need to retrieve specific series from specific patients' CT studies. I am currently able to retrieve the whole study but i would like to retrieve only the series based on a specific description.

In summary: study level query by patientID --> series level query by description (0008,103E) --> c-get only the series

I could not find any examples of something similar. Can someone help me?

[medihack]

Your summary scenario looks quite ok to me. Find the StudyInstanceUID and SeriesInstanceUID using C-FIND queries, and then fetch the series using C-GET (if your PACS supports it). There are easy to follow examples on the pynetdicom homepage. You can also take a look at the core of our ADIT project where we do something similar (but it may be unnecessary complex for your use case). Hope this helps.

[dsnishimura]

I am still confused about how do I fetch the results from the first C-FIND to use them in the C-GET query. I am able to display the results but not use them in the rest of the code.

[mrbean-bremen]

As @medihack has pointed out, the example already shows how to get a single series, using SERIES Q/R level. You just need to plug in your own values for PatientID, StudyInstanceUID and SeriesInstanceUID from your query. Which point there is not clear to you?

[dsnishimura]

Iniatially, I was only given the PatientID and the description, not the StudyInstanceUID, nor the SeriesInstanceUID. I was able to find the StudyInstanceUID through C-FIND at PATIENT level, but how do I store that information to use in the next level query, considering I need to repeat this process for all the patients?

From what I understand I need a C-FIND at PATIENT level to fetch the StudyInstanceUID , then another C-FIND at STUDY level to fetch the SeriesInstanceUID containing the description I need. Finally a C-GET to retrieve the series. I just don´t understand how do I associate this cascade.

[mrbean-bremen]

This sounds more like a question about your application logic than about pynetdicom. As the SCU implementations are basically just function calls (as opposed to SCP implementations that use an event loop), I see no real problem here. Just parametrize your C-GET calls with the needed values and call them where needed (e.g. after the needed information has been retrieved).

[dsnishimura]

@medihack Thak you very much!!!! The part in your project where you dictfy the responses helped a lot with my ETL pipepline. It was thanks to that that I was able to extract the correct studies. You were a life saver!! Btw, sorry for the late response I've been caught up in some other projects here at the hospital and was not able to give much attention to this one for a while.

[edmcdonagh]

Once you have your study instance UID, use this to build a series level query: ``` d2 = Dataset() d2.StudyInstanceUID = rsp.study_instance_uid d2.QueryRetrieveLevel = "SERIES" d2.SeriesDescription = "" d2.SeriesNumber = "" d2.SeriesInstanceUID = "" d2.Modality = "" d2.NumberOfSeriesRelatedInstances = "" d2.StationName = "" d2.SpecificCharacterSet = "" d2.SeriesTime = "" ``` customising the fields that you'd like to be populated, always ensuring that the mandatory fields at each level are included. And bearing in mind that the optional fields might be returned empty or missing - the DICOM conformance statement of the PACS will detail what it will respond with. Series description will not necessarily be returned. (Don't pay any attention to the D2 naming, I'm copying and pasting code on my phone) You can see what is required at each level here https://www.medicalconnections.co.uk/kb/Query-Parameters/ Finally, be aware that C-GET isn't widely used and often isn't implemented. Instead it is common to use C-MOVE, which is actually a request for the remote system to instigate a C-STORE back to you, with the only information you can give it being your AE Title, which means the remote system needs to be configured to know that your AE Title is associated with your IP address and port that you are running a store SCP on. I hope that gets you a little closer? Kind regards Ed

[dsnishimura]

You have just clarified a lot of misunderstandings I had. Thank you!!!