image orientation - DICOM

This is a discussion on image orientation - DICOM ; Is there an unambiguous definition within DICOM or IHE that where is the image orientation defined for display? Should the modality already rotate and/or flip images and store them as such OR should the PACS do this (which I do ...

+ Reply to Thread
Results 1 to 2 of 2

Thread: image orientation

  1. image orientation

    Is there an unambiguous definition within DICOM or IHE that where is
    the image orientation defined for display? Should the modality already
    rotate and/or flip images and store them as such OR should the PACS do
    this (which I do doubt) OR is it the viewing software that is
    responsible? Or, is the responsibility shared between the different
    actors? I'd appreciate comments on this. VPT


  2. Orientation - display or modality responsibility ?, was Re: imageorientation

    Hi

    vpt296@yahoo.com wrote:
    > Is there an unambiguous definition within DICOM or IHE that where is
    > the image orientation defined for display? Should the modality already
    > rotate and/or flip images and store them as such OR should the PACS do
    > this (which I do doubt) OR is it the viewing software that is
    > responsible? Or, is the responsibility shared between the different
    > actors? I'd appreciate comments on this. VPT


    Ideally, since the viewing software is where the hanging
    protocol is applied, and orienting to the radiologist's
    preference is a function of the hanging protocol, then
    it is the responsibility of the viewing software.

    To do this presupposes that the modality has encoded
    sufficient information to drive the hanging protocol rules.

    For cross-sectional modalities like CT and MR, the Image
    Orientation (Patient) direction cosines provide
    the information, and the viewing software should rotate/flip
    accordingly, should the hanging protocol require it. Most of
    the time this is not done because the modality already encodes
    the pixel data in the most common orientation, since these
    acquisitions are fairly consistent both in terms of how they
    are performed and how they are expected to be viewed (with
    direct prone coronal sinus CT being an exception, for example).

    For projection radiography, the information is encoded in
    Patient Orientation, but being type 2 in CR objects this is
    not always populated, in which case the viewing software is
    unable to determine what to do and is left at the mercy of
    however the cassette was positioned by the tech or whether
    or not the modality provided an interface for the operator
    to flip/rotate into the "correct" orientation, or to apply
    rules based on its knowledge of the body part and/or
    possible analysis of the image content. For this reason
    we made Patient Orientation type 1 in DX objects to make
    downstream application of hanging protocols more effective.

    Finally, there is always the possibility of using an
    operator (at the modality or a QC station) to orient the
    images and save a presentation state recording the rotate
    and/or flip to address this problem, the existence of which
    the viewing software may need to take into account.

    The PACS (or the Image Manager/Archive component of the
    PACS) has a largely passive role in this regard, and one
    would not expect it to be changing orientation.

    The only mention of this in IHE that I am aware of is in
    the Mammography profile, where it is explicitly stated
    that an Image Display shall rotate/flip as necessary to
    match the expected display orientation regardless of the
    actual orientation of the images. Specifically:

    "4.16.4.2.2.1.1.2 Image Orientation and Justification

    Image Display actors shall not assume that the pixel data
    is encoded with an orientation that is suitable for direct
    display to the user without flipping or rotating into the
    correct orientation.

    The Image Display actor shall use the values of Image Laterality
    (0020,0062), View Code Sequence (0054,0220), View Modifier Code
    Sequence (0054,0222) and Patient Orientation (0020,0020) to
    display images according to the preferred hanging protocol of
    the current user, rather than depend on descriptive attributes
    such as Series Description (0008,103E).

    The Image Display shall allow the user to select or configure
    hanging protocols such that given a set of images containing
    these attributes, the placement of images relative to one another,
    the required orientation of the images, the display of current
    and prior images, and the sequence of layouts displayed can
    be defined."

    Once we get some experience with this sort of thing in IHE
    with mammography, I would imagine that we could extend these
    concepts from the mammography-specific profile to more general
    content and display behavior profiles for all projection and
    cross-sectional images. Until recently (with the NM Display,
    Mammography and Fusion profiles), IHE has largely ignored the
    payload of images and behavior of displays and focused mostly
    on the work flow. It will be interesting to hear the feedback
    from the mammography display vendors preparing for the January
    2007 Connectathon, as well as from those that choose not to
    participate for one reason or another.

    David

+ Reply to Thread