C-FIND Hierarchical Search Method doubt - DICOM

This is a discussion on C-FIND Hierarchical Search Method doubt - DICOM ; Hello, We have developed a Q/R SCU client that is working correctly with two different PACS of two multinational providers (we have adecuades responses). We use de Study Root Information Model and the SERIES Query Level. We contruct C-FIND request ...

+ Reply to Thread
Results 1 to 4 of 4

Thread: C-FIND Hierarchical Search Method doubt

  1. C-FIND Hierarchical Search Method doubt

    Hello,

    We have developed a Q/R SCU client that is working correctly with two
    different PACS of two multinational providers (we have adecuades
    responses).

    We use de Study Root Information Model and the SERIES Query Level. We
    contruct C-FIND request with those tags:

    Study Date (from study level) = ""
    Study Time (from study level)= ""
    Patients ID (from study level)= ""
    Patients Name (from study level)= ""
    Study Description (from study level)= ""
    Modality (from series level)= ""
    Accession Number = "VALUE"
    Study Instance UID (from study level)= ""
    Query Retrieve Level = "SERIES"
    PerformedStationAET (from study level, i guess)= ""
    SeriesInstanceUID (from series level)= ""

    Is this request allowed?, because I'm afraid that this is the same
    case that http://groups.google.es/group/comp.p...d36c057a30fabe,
    is this true?

    Thanks in advance,
    Ramón


  2. Re: C-FIND Hierarchical Search Method doubt

    On Tue, 24 Jun 2008 14:53:30 -0700 (PDT), xose

    >Hello,
    >
    >We have developed a Q/R SCU client that is working correctly with two
    >different PACS of two multinational providers (we have adecuades
    >responses).
    >
    >We use de Study Root Information Model and the SERIES Query Level. We
    >contruct C-FIND request with those tags:
    > ...
    >Is this request allowed?


    With relational search it might work BUT with the std. query/retrieve
    model not!

    Why?

    Because you can only search on one level in one query.
    That means you can search on
    patient-level
    OR
    on study-level providing ALL identifying tags from the level above
    (patient-id)
    OR
    on series-level providing ALL identifying tags from the level above
    (patient-id + study-instance-uid)
    OR
    on object-level providing ALL identifying tags from the level above
    (patient-id + study-instance-uid + series-instance-uid)

    That means if want to find all image for a patient with name 'Hugo'
    you have to

    Query on patient-level for patients with name='Hugo'
    You get 0..n patient-id as result

    foreach patient-id found you have to
    Query on study-level for all studies providing the found patient-id
    You get 0..n study-instance-uids as result

    foreach study-instance-uid found you have to
    Query on series-level for all series providing the found patient-id
    PLUS study-instance-uid
    You get 0..n series-instance-uids as result

    foreach series-instance-uid found you have to
    Query on object-level for all objects providing the found patient-id
    PLUS study-instance-uid PLUS series-instance-uid
    You get 0..n sop-instance-uids as result

    thats it.

    Sometimes you are able to start the query on patient-study-root which
    might provide a shortcut




  3. Re: C-FIND Hierarchical Search Method doubt

    If I look through the set of tags for which you expect/want to get
    return keys, the only ones that are at the series level are modality
    (0008,0060) and Series Instance UID (0020,000E). What it appears you
    are doing is looking to build a list work item entries sorted or
    filtered by modality, and include in that worklist, links enabled
    retrieval by series (or at least have the C-MOVE operation to retrieve
    the series in the study individually by UID). Perhaps you only want
    the count of the series in the study and saw asking for the series
    instance UIDs as a way to obtain it?

    In any event, suggest you can get most of what you may want doing a
    query at the study level and using some alternative tags
    (0008,0061) Modalities in Study, where it is supported, will give you
    a potentially multi-valued result listing all the modalities in the
    study. Some QR SCPs are stupid about this, simply giving you a listing
    of the (0008,0060) tags for all the series in the Study -- e.g. for an
    MR study with 8 series you might get the value MR\MR\MR\MR\MR\MR\MR
    \MR. For a lot of studies you will get MR\PR or CR\MR or CT\PR as
    some modalities are now sending the window/level settings done by the
    tech at the modality as presentation states. Similarly, ultrasound
    exams may come back with US\SR because the study includes structured
    reports for some of the measurements performed in the study.

    Not all Q/R SCPs support (0008,0061), in which case you could drop
    into a loop to query each study at the series level to obtain the list
    of modalities. However, there is an awful lot of support for querying
    the tag (0008,0060) on QR SCPs out there - even though the SCPs may
    not offficially support a relational query (or negotiate it).

    As for the list of Series UIDs - would question why you want these
    UIDs. If you're simply looking for series count, you could query for
    (0020,1206) Number of Study Related Series. Again, not all SCPs
    support but a suprising number do. You can get the number or series,
    and the number of images in the study total (0020,1208) if you want to
    gauge the size of the study. Detail like the series UID or series
    description (useful for deciding which series you really want to pull,
    if you're going to support user selection of a specific series for
    download/display).




  4. Re: C-FIND Hierarchical Search Method doubt

    Thank you both of you for your kindly answers...

    We will think about remake our C-FIND library...


    On 25 jun, 21:24, goodalle wrote:
    > If I look through the set of tags for which you expect/want to get
    > return keys, the only ones that are at the series level are modality
    > (0008,0060) and Series Instance UID (0020,000E). What it appears you
    > are doing is looking to build a list work item entries sorted or
    > filtered by modality, and include in that worklist, links enabled
    > retrieval by series (or at least have the C-MOVE operation to retrieve
    > the series in the study individually by UID). Perhaps you only want
    > the count of the series in the study and saw asking for the series
    > instance UIDs as a way to obtain it?
    >
    > In any event, suggest you can get most of what you may want doing a
    > query at the study level and using some alternative tags
    > (0008,0061) Modalities in Study, where it is supported, will give you
    > a potentially multi-valued result listing all the modalities in the
    > study. Some QR SCPs are stupid about this, simply giving you a listing
    > of the (0008,0060) tags for all the series in the Study -- e.g. for an
    > MR study with 8 series you might get the value MR\MR\MR\MR\MR\MR\MR
    > \MR. For a lot of studies you will get MR\PR or CR\MR or CT\PR as
    > some modalities are now sending the window/level settings done by the
    > tech at the modality as presentation states. Similarly, ultrasound
    > exams may come back with US\SR because the study includes structured
    > reports for some of the measurements performed in the study.
    >
    > Not all Q/R SCPs support (0008,0061), in which case you could drop
    > into a loop to query each study at the series level to obtain the list
    > of modalities. However, there is an awful lot of support for querying
    > the tag (0008,0060) on QR SCPs out there - even though the SCPs may
    > not offficially support a relational query (or negotiate it).
    >
    > As for the list of Series UIDs - would question why you want these
    > UIDs. If you're simply looking for series count, you could query for
    > (0020,1206) Number of Study Related Series. Again, not all SCPs
    > support but a suprising number do. You can get the number or series,
    > and the number of images in the study total (0020,1208) if you want to
    > gauge the size of the study. Detail like the series UID or series
    > description (useful for deciding which series you really want to pull,
    > if you're going to support user selection of a specific series for
    > download/display).



+ Reply to Thread