Study Instance and SOP Instance UIDs - DICOM

This is a discussion on Study Instance and SOP Instance UIDs - DICOM ; OK, I just want to make sure I have the correct termonology here: A Study Instance UID is a unique number that is assigned to every image in a study. So if a CT had 250 images, all of the ...

+ Reply to Thread
Results 1 to 3 of 3

Thread: Study Instance and SOP Instance UIDs

  1. Study Instance and SOP Instance UIDs

    OK, I just want to make sure I have the correct termonology here:

    A Study Instance UID is a unique number that is assigned to every
    image in a study. So if a CT had 250 images, all of the images would
    have the same Study Instance UID.

    A SOP Instance UID is a unique number that is assigned to every image
    in a study, but each image in a study has a different SOP Instance
    UID. So if a CT had 250 images, each image would have a SOP Instance
    UID, but there would be 250 different ones.

    With me so far?

    I am having trouble with a certain PACS vendor that sends images to my
    PACS.

    They automatically forward/export all of their images to me. This is
    done instantly after they receive the images. However, sometimes they
    need to make changes to these studies (usually because the modalities
    do not provide enough information.)

    If they make any changes to a study, the Study Instance UID changes.
    They then send the study with the correct information to me.

    My PACS sees this new study coming in, and because it doesn't match
    any existing Study UIDs, adds it to the worklist.

    I now have two studies on my PACS that are essentially the same.

    But this isn't the problem. The problem is that, even though the
    Study Instance UID change, the SOP Instance UIDs all stayed the same.

    My PACS cannot handle two studies having the same SOP Instance UIDs.
    None of the images will show up, in either study. I have spoken to
    them about this, they claim that according to the DICOM standards, if
    you change the Study UID, you have to also change all of the SOP
    Instance UIDs. I am trying to convince the other PACS of this but I
    am having trouble finding any proper documentation for this.

    Does anyone have any advice on what I should do about this?


  2. Re: Study Instance and SOP Instance UIDs

    Hi,

    As far as I understand it UIDs are made up of a sequence of numeric
    components separated by dots (so not a single number).

    They have to globally and uniquely identify an object (where an object
    can for example be a study/series or SOP instance).
    Which means that two different objects, even if they are not of the same
    type or do not appear in the same context, can not have the same UID.

    e.g. having two different SOP instances (even if they are in different
    studies) with the same UID should not be legal.

    Regards,
    Nils Gladitz

    yaksha42@gmail.com wrote:
    > OK, I just want to make sure I have the correct termonology here:
    >
    > A Study Instance UID is a unique number that is assigned to every
    > image in a study. So if a CT had 250 images, all of the images would
    > have the same Study Instance UID.
    >
    > A SOP Instance UID is a unique number that is assigned to every image
    > in a study, but each image in a study has a different SOP Instance
    > UID. So if a CT had 250 images, each image would have a SOP Instance
    > UID, but there would be 250 different ones.
    >
    > With me so far?
    >
    > I am having trouble with a certain PACS vendor that sends images to my
    > PACS.
    >
    > They automatically forward/export all of their images to me. This is
    > done instantly after they receive the images. However, sometimes they
    > need to make changes to these studies (usually because the modalities
    > do not provide enough information.)
    >
    > If they make any changes to a study, the Study Instance UID changes.
    > They then send the study with the correct information to me.
    >
    > My PACS sees this new study coming in, and because it doesn't match
    > any existing Study UIDs, adds it to the worklist.
    >
    > I now have two studies on my PACS that are essentially the same.
    >
    > But this isn't the problem. The problem is that, even though the
    > Study Instance UID change, the SOP Instance UIDs all stayed the same.
    >
    > My PACS cannot handle two studies having the same SOP Instance UIDs.
    > None of the images will show up, in either study. I have spoken to
    > them about this, they claim that according to the DICOM standards, if
    > you change the Study UID, you have to also change all of the SOP
    > Instance UIDs. I am trying to convince the other PACS of this but I
    > am having trouble finding any proper documentation for this.
    >
    > Does anyone have any advice on what I should do about this?
    >


  3. Re: Study Instance and SOP Instance UIDs

    yaksh,

    Your understanding is correct. A SOP Instance UIDs must be globally
    unique, not just unique within a Series Instance UID or a Study
    Instance UID. Similarly, a Series Instance UID must be globally
    unique, not just unique within a Study Instance UID. In other words,
    a SOP Instance UID cannot "belong" to more that one Series Instance
    UID or Study Instance UID, and a Series Instance UID cannot "belong"
    to more than one Study Instance UID.

    This is based on the relationship of the UIDs, and it doesn't matter
    if it's really copies of "the same" image.

    If your PACS changes the Study Instance UID, then they must also
    generate new Series and SOP Instance UIDs.

    I need to leave it to the more knowledgeable DICOM folks here to point
    to the chapter and verse in the DICOM spec.

    Other than DICOM legalese, here are some things that might help
    convince your PACS vendor:
    * There are places where composite SOP instances (DICOMese for
    "images or other objects like structured reports, presentation states,
    etc.) are identified ONLY by their SOP Instance UID, without providing
    the Series or Study Instance UID. One example is Storage Commitment,
    there are others.
    * When the try to work with other DICOM vendors, they are going to
    run into one brick wall after another if they try to do this. They
    are in for a world of hurt.

    Frankly, for a long time I also interpreted the rules the way that
    your PACS did, and found the specification of this requirement
    vanishingly subtle (and I've been working with the spec for over 10
    years). Today, our PACS (like most) follows this rule to the letter.

    Good luck!
    Dan


+ Reply to Thread