Mammo Hanging Protocol problem in combination with PresentationStates - DICOM

This is a discussion on Mammo Hanging Protocol problem in combination with PresentationStates - DICOM ; My colleagues responsible for the implementation of DICOM Hanging Protocols in our Image Display stumbled over an issue concerning GSPS in combination with HP. It popped up at the European IHE Connectathon this year. The problem occurs if the HP ...

+ Reply to Thread
Results 1 to 3 of 3

Thread: Mammo Hanging Protocol problem in combination with PresentationStates

  1. Mammo Hanging Protocol problem in combination with PresentationStates

    My colleagues responsible for the implementation of DICOM Hanging
    Protocols in our Image Display stumbled over an issue concerning GSPS
    in combination with HP. It popped up at the European IHE Connectathon
    this year.

    The problem occurs if the HP and the GSPS contain contradictory
    specifications. The Mammography HP requires some of the images to be
    flipped. However, some studies contained presentation states with
    DisplayedAreaTopLeftHandCorner and DisplayedAreaBottomRightHandCorner
    specified. According to our interpretation the presentation state is
    'stronger' and the images were not flipped and thus the hanging did
    not meet the user expectations.

    Even if the presentation state would luckily define the spatial
    transformation as required for the Mammography case, we would run into
    issues if the HP should support ventral and dorsal mode. As this would
    require each images to be displayed plain and flipped once.

    From 'Supplement 60: Hanging Protocols' page 59, C.X.3.1.4 we conclude
    that the presentation intent with respect to the spatial
    transformation should be taken from the presentation state, if
    present. So the Display Set Patient Orientation (0072,0700) would be
    ignored.

    In chapter W.4.2.1.4.1.3.1 it is proposed to choose the 'best
    matching' GSPS for the chosen HP. However, the problem is that there
    is 'just' one GSPS.

    So if we privilege the PS, we might result in wrong hangings. The
    other way around we are violating the DICOM standard and the PS will
    not be shown initially.

    One solution would be to define, that the value of Display Set Patient
    Orientation (0072,0700) in the Hanging Protocol shall override any
    such intent in associated Presentation States, as defined for HP
    Presentation Intent Attribute Show Grayscale Inverted (0072,0706).

    A more comprehensive solution would be to define an additional flag
    attribute (e.g. "Override Display Set Patient Orientation in
    Presentation State") which controls, if the Patient Orientation
    defined in the Hanging Protocol or that specified in an associated
    Presentation State shall be applied.

    Such flags may also make sense for other HP Presentation Intent
    Attributes, in particular for VOI Type (0072,0702), Show Image True
    Size Flag (0072,0710) and Show Grayscale Inverted (0072,0706).

    Regards,
    Gunter Zeilinger
    Agfa Healthcare
    Vienna/Europe

  2. Re: Mammo Hanging Protocol problem in combination with PresentationStates

    Hi Gunter

    I think that you are looking at this from the wrong perspective.

    Whether or not to use the default hanging protocol or a presentation
    state that may be present in the Study depends entirely on what the
    user's expectation and intent is, and is not something that can be
    prescribed a priori to be one way or the other.

    If one is a mammography radiologist performing primary interpretation,
    then almost certainly the hanging protocol should take preference,
    and any presentation state is likely present only to add additional
    information conveyed by the tech (e.g., marks for location of scars,
    skin lesions, etc.), and the user would likely expect to be notified
    that it was available but to have to take action to make it visible
    (to apply); same goes for CAD marks in presentation states in lieu
    of SR objects. A presentation state should NOT be used to flip an
    incorrectly oriented image, since the orientation attributes should
    be corrected within the image; anything else is unsafe.

    On the other hand, a referring physician might well want to see the
    presentation state(s) saved by the radiologist to highlight an image,
    and have that by default override or replace any default hanging
    protocol.

    In other words, this may be a matter for user preferences and/or
    specific work list attributes to convey, and little to do with the
    hanging protocol itself.

    And, of course, there is always the matter of how to make available
    multiple presentation states that apply to the same image(s), and
    which, if any, to apply to default.

    Much of the complication arises because so many different things are
    overloaded into the same presentation state (pan/zoom/flip/window/
    annotate).

    David

    gunter zeilinger wrote:
    > My colleagues responsible for the implementation of DICOM Hanging
    > Protocols in our Image Display stumbled over an issue concerning GSPS
    > in combination with HP. It popped up at the European IHE Connectathon
    > this year.
    >
    > The problem occurs if the HP and the GSPS contain contradictory
    > specifications. The Mammography HP requires some of the images to be
    > flipped. However, some studies contained presentation states with
    > DisplayedAreaTopLeftHandCorner and DisplayedAreaBottomRightHandCorner
    > specified. According to our interpretation the presentation state is
    > 'stronger' and the images were not flipped and thus the hanging did
    > not meet the user expectations.
    >
    > Even if the presentation state would luckily define the spatial
    > transformation as required for the Mammography case, we would run into
    > issues if the HP should support ventral and dorsal mode. As this would
    > require each images to be displayed plain and flipped once.
    >
    > From 'Supplement 60: Hanging Protocols' page 59, C.X.3.1.4 we conclude
    > that the presentation intent with respect to the spatial
    > transformation should be taken from the presentation state, if
    > present. So the Display Set Patient Orientation (0072,0700) would be
    > ignored.
    >
    > In chapter W.4.2.1.4.1.3.1 it is proposed to choose the 'best
    > matching' GSPS for the chosen HP. However, the problem is that there
    > is 'just' one GSPS.
    >
    > So if we privilege the PS, we might result in wrong hangings. The
    > other way around we are violating the DICOM standard and the PS will
    > not be shown initially.
    >
    > One solution would be to define, that the value of Display Set Patient
    > Orientation (0072,0700) in the Hanging Protocol shall override any
    > such intent in associated Presentation States, as defined for HP
    > Presentation Intent Attribute Show Grayscale Inverted (0072,0706).
    >
    > A more comprehensive solution would be to define an additional flag
    > attribute (e.g. "Override Display Set Patient Orientation in
    > Presentation State") which controls, if the Patient Orientation
    > defined in the Hanging Protocol or that specified in an associated
    > Presentation State shall be applied.
    >
    > Such flags may also make sense for other HP Presentation Intent
    > Attributes, in particular for VOI Type (0072,0702), Show Image True
    > Size Flag (0072,0710) and Show Grayscale Inverted (0072,0706).
    >
    > Regards,
    > Gunter Zeilinger
    > Agfa Healthcare
    > Vienna/Europe


  3. Re: Mammo Hanging Protocol problem in combination with PresentationStates

    Hi David,

    I agree, it's not the question if we should meet user expectations OR
    keep our HP implementation strict standard compliant, but how to fix
    the standard to be able to express user expectations / preferences
    without having to "customize" its interpretation.

    So we would like to be able to define HPs for the primary
    interpretation by radiologists, and different HPs for referring
    physicians, reflecting the different needs / expectations of such
    "User Groups".

    I agree that the problem comes from the fact, that one Presentation
    State defines so many different things, whereat some of them may even
    be expected to be applied initially for the primary interpretation.
    For instance, we know one Mammo modality, which use a separate
    Presentation State for just providing a Display Shutter.

    And if I understood you right, you also would expect, that an
    application does not flip the image, when the mammography radiologist
    activates the CAD marks provided by a Presentation State which
    defines a different Spatial Orientation than the Mammo HP Image Box.

    So perhaps adding a Flag "Apply Available Presentation State" [YES/NO]
    and changing the definition of Display Set Patient Orientation
    (0072,0700) to override the Spatial Orientation defined by Displayed
    Area TLHC/BRHC of applied Presentation State would cover most of the
    use cases. - Even we would prefer a more fine-grained control about
    which part of associated Presentation States shall be applied, as
    already provided for Graphic Annotations by attribute Show Graphic
    Annotation Flag (0072,0712).

    gunter

    On Sep 3, 3:05*pm, David Clunie wrote:
    > Hi Gunter
    >
    > I think that you are looking at this from the wrong perspective.
    >
    > Whether or not to use the default hanging protocol or a presentation
    > state that may be present in the Study depends entirely on what the
    > user's expectation and intent is, and is not something that can be
    > prescribed a priori to be one way or the other.
    >
    > If one is a mammography radiologist performing primary interpretation,
    > then almost certainly the hanging protocol should take preference,
    > and any presentation state is likely present only to add additional
    > information conveyed by the tech (e.g., marks for location of scars,
    > skin lesions, etc.), and the user would likely expect to be notified
    > that it was available but to have to take action to make it visible
    > (to apply); same goes for CAD marks in presentation states in lieu
    > of SR objects. A presentation state should NOT be used to flip an
    > incorrectly oriented image, since the orientation attributes should
    > be corrected within the image; anything else is unsafe.
    >
    > On the other hand, a referring physician might well want to see the
    > presentation state(s) saved by the radiologist to highlight an image,
    > and have that by default override or replace any default hanging
    > protocol.
    >
    > In other words, this may be a matter for user preferences and/or
    > specific work list attributes to convey, and little to do with the
    > hanging protocol itself.
    >
    > And, of course, there is always the matter of how to make available
    > multiple presentation states that apply to the same image(s), and
    > which, if any, to apply to default.
    >
    > Much of the complication arises because so many different things are
    > overloaded into the same presentation state (pan/zoom/flip/window/
    > annotate).
    >
    > David
    >
    > gunter zeilinger wrote:
    > > My colleagues responsible for the implementation of DICOM Hanging
    > > Protocols in our Image Display stumbled over an issue concerning GSPS
    > > in combination with HP. It popped up at the European IHE Connectathon
    > > this year.

    >
    > > The problem occurs if the HP and the GSPS contain contradictory
    > > specifications. The Mammography HP requires some of the images to be
    > > flipped. However, some studies contained presentation states with
    > > DisplayedAreaTopLeftHandCorner and DisplayedAreaBottomRightHandCorner
    > > specified. According to our interpretation the presentation state is
    > > 'stronger' and the images were not flipped and thus the hanging did
    > > not meet the user expectations.

    >
    > > Even if the presentation state would luckily define the spatial
    > > transformation as required for the Mammography case, we would run into
    > > issues if the HP should support ventral and dorsal mode. As this would
    > > require each images to be displayed plain and flipped once.

    >
    > > From 'Supplement 60: Hanging Protocols' page 59, C.X.3.1.4 we conclude
    > > that the presentation intent with respect to the spatial
    > > transformation should be taken from the presentation state, if
    > > present. So the Display Set Patient Orientation (0072,0700) would be
    > > ignored.

    >
    > > In chapter W.4.2.1.4.1.3.1 it is proposed to choose the 'best
    > > matching' GSPS for the chosen HP. However, the problem is that there
    > > is 'just' one GSPS.

    >
    > > So if we privilege the PS, we might result in wrong hangings. The
    > > other way around we are violating the DICOM standard and the PS will
    > > not be shown initially.

    >
    > > One solution would be to define, that the value of Display Set Patient
    > > Orientation (0072,0700) in the Hanging Protocol shall override any
    > > such intent in associated Presentation States, as defined for HP
    > > Presentation Intent Attribute Show Grayscale Inverted (0072,0706).

    >
    > > A more comprehensive solution would be to define an additional flag
    > > attribute (e.g. "Override Display Set Patient Orientation in
    > > Presentation State") which controls, if the Patient Orientation
    > > defined in the Hanging Protocol or that specified in an associated
    > > Presentation State shall be applied.

    >
    > > Such flags may also make sense for other HP Presentation Intent
    > > Attributes, in particular for VOI Type (0072,0702), Show Image True
    > > Size Flag (0072,0710) and Show Grayscale Inverted (0072,0706).

    >
    > > Regards,
    > > Gunter Zeilinger
    > > Agfa Healthcare
    > > Vienna/Europe



+ Reply to Thread