Multiple Accession Numbers in Image IOD - DICOM

This is a discussion on Multiple Accession Numbers in Image IOD - DICOM ; I am working on an algorithm with matches RIS data to studies in my existing image server. I would like to rewrite my image SOP instances with the reconciled data to export to a cleaner database (migration and data cleansing). ...

+ Reply to Thread
Results 1 to 3 of 3

Thread: Multiple Accession Numbers in Image IOD

  1. Multiple Accession Numbers in Image IOD

    I am working on an algorithm with matches RIS data to studies in my
    existing image server. I would like to rewrite my image SOP instances
    with the reconciled data to export to a cleaner database (migration and
    data cleansing). However, many imaging studies represent more than one
    requested RIS study (eg CT Abdomen/CT Pelvis.) Even after reviewing
    Annex J of Part 17, it is still not clear to me the best way to
    represent multiple requested studies (from the RIS) by their accession
    numbers.

    I notice the MPPS instance has a Scheduled Step Attributes sequence
    which can contain a accession number, but this is not supposed to be
    part of an image SOP instance, is it?

    The Request Attributes sequence can be in the image SOP instance, but
    it only contains Requested Procedure ID and Requested Procedure Step
    ID, not specifically an accession number. Would mapping the accession
    number to the Requested Procedure ID be the most
    acceptable/standardized way of handling this?

    Any thoughts? Thanks.

    Cliff Sweet


  2. Re: Multiple Accession Numbers in Image IOD


    cliff wrote:
    > I am working on an algorithm with matches RIS data to studies in my
    > existing image server. I would like to rewrite my image SOP

    instances
    > with the reconciled data to export to a cleaner database (migration

    and
    > data cleansing). However, many imaging studies represent more than

    one
    > requested RIS study (eg CT Abdomen/CT Pelvis.) Even after reviewing
    > Annex J of Part 17, it is still not clear to me the best way to
    > represent multiple requested studies (from the RIS) by their

    accession
    > numbers.
    >
    > I notice the MPPS instance has a Scheduled Step Attributes sequence
    > which can contain a accession number, but this is not supposed to be
    > part of an image SOP instance, is it?
    >
    > The Request Attributes sequence can be in the image SOP instance, but
    > it only contains Requested Procedure ID and Requested Procedure Step
    > ID, not specifically an accession number. Would mapping the

    accession
    > number to the Requested Procedure ID be the most
    > acceptable/standardized way of handling this?
    >
    > Any thoughts? Thanks.
    >
    > Cliff Sweet



    The Accession Number dicom attribute is only a container to convey an
    identifier, not an information entity into itself. Under the rules of
    the IHE Schedule Workflow, Group Procedure Case (ie. more than one
    entry from the MWL is being satisfied by the acquisition) the Accession
    Number attribute is left empty, signaling the AE receiving the images
    to look into the Request Attributes Sequence, and use the Requested
    Procedure ID attribute values from different sequence items to get the
    RIS procedure identifiers.

    This should imply the same value is in fact normally encoded into both
    the Accession Number and Requested Procedure Id. Look in the IHE
    Technical Framework. There will be a mapping from the IHE standard HL7
    to the DICOM attributes of the MWL. Its one of the ORC segment fields.

    I'm sure they would really have liked to do away with the Accession
    Number altogether, but its usage is so ingrained it will be hard to get
    rid of.

    Given that a lot of modalities dont support generating the Request
    Attributes Sequence you're still going to find a lot of these group
    cases are simply not linked to all of the RIS entries. In some cases
    the RIS may know/have multiple procedures in their database linked to
    the same DICOM study, but the same is not neccessarily true for the
    PACS (hence the fact that you're writing this piece of code). In an
    implemention my company did, we built extensive heuristics process
    lists of RIS procedures and lists of PACS studies to identify
    correlations. Things like - in the ris procedures with the same
    modality, different body part, with acquisition date/time very close
    together, only one of the set has a corresponding entry in the PACS,
    all ris procedures reported = pretty good probability all the ris
    procedures should be linked to the same dicom study. This works
    especially well when you also note that certain RIS procedures are
    typically members of group case, acquisition set --ie. CT Chest Abdomen
    Pelvis, CT Head, Neck, Spine, MR Head Neck Spine, CR Right Wrist and
    Left Wrist, CR Right Ankle and Left Ankle, etc. A lot of times the RIS
    procedures are in fact a single order entry code, but the RIS expands
    them into a multiple MWL entries through group procedure logic. A lot
    can be learned by simply looking through the procedure table on the
    RIS, and adapt your algorithm appropriately


  3. Re: Multiple Accession Numbers in Image IOD

    Thanks, Eric. I'll map the accession number in to the procedure id in
    the request attributes sequence. I'll pick one accession number to
    fill tag 0008,0050 because my current image server like it best that
    way...


+ Reply to Thread