Skip to content

Instantly share code, notes, and snippets.

@jdmichaud
Last active July 17, 2020 13:01
Show Gist options
  • Save jdmichaud/af46ad5a36739e0f43ec35723a05c8aa to your computer and use it in GitHub Desktop.
Save jdmichaud/af46ad5a36739e0f43ec35723a05c8aa to your computer and use it in GitHub Desktop.

from: http://dicom.nema.org/dicom/Conf-2005/Day-2_Selected_Papers/B304_Harvey&DC_UserInterfaceIssuesForPresentationStates.pdf

User Interface Issues for DICOM Presentation StatesDICOM Presentation States

Overview of Softcopy Presentation State (GSPS)Presentation State (GSPS)

  • Added in 1999 as supplement 33
  • Major departure for DICOM moving from data content and transmission into display
  • Closely related to (but independent of) Supplement 28/Part 14 (Grayscale Standard Display Function)

GSPS Design

  • 3GSPS DesignGSPS Design
  • GSPS objects may be created at any time after imaging by any equipment
  • GSPS are independent of the original image(s) and do NOT require modifications to the original image(s)
  • Handled exactly like other instance objects
    • C-STORE, G-FIND, C-MOVE etc.
    • Easy for existing SCPs to handle
  • No specific GSPS support in the Q/R model

Items controlled by GSPS

  • All LUTs/transformations
    • Modality
    • VOI
    • Presentation
  • Displayed Area & Shuttering
  • Rotation & Flipping
  • Mask Subtraction
  • Independent Annotations
    • Text & graphics
    • Both image & display relative
  • Activation of overlays in the image, or in the GSPS object

Place of GSPS objects in the DICOM Model

  • Strictly part of one study
  • Only refer to image(s) in that study
  • In their own series
  • May be stored on off-line media etc.
  • Generally, they may be treated just like any other composite object

Relationship to Images

  • One "manifest" (Referenced Series Sequence) sequence defines a list of images to which this PS object relates
  • Huge choice of granularity
    • All series in a study
    • One or more specific series
    • One or more specific images
    • One of more specific frames within an image
    • Any combination of the above!

"Top level" Items: No reference to series / images

  • Masking
  • Shuttering (geometric & overlay)
  • Spatial Transformation (Rotate & Flip)
  • Modality LUT/rescale
  • Presentation LUT
  • Overlay Activation

"Image specific" Items: With Reference Image Sequence

  • Displayed Area
  • Annotations - graphic & Text
  • VOI LUT/Windowing

How PS objects may be created

  • Automatically by the modality
    • As default or reflection of radiographer choices
    • As part of PGP
  • Explicitly by users
    • As part of reporting
    • For conferences etc.
    • For teaching
    • etc.

Key Object Note Support

  • KON (and SR) objects can specify the relationship between images and GSPS objects
  • Achieved using the “Referenced SOP Sequence” element (0008,1199) in the Image Reference Macro
    • This is specialised to allow only a single entry, and to specify a GSPS object only.

Clinical Scenarios (in a GSPS capable viewer):

  • User has queried for and retrieved image instances directly
    • The PACS has PS objects for that study
    • Possibly many for the same image with different scope
  • User selects and retrieves GSPS object
  • User selects and retrieves KON object

GSPS Deficiencies, Ambiguities & Problems:

  • No C-FIND method to find GSPS objects for a given instance
  • So need to retrieve every PS object in the study to inspect them internally
  • What should the viewer do if > 1 possible GSPS for an image?
  • There is no predefined coding (like there is in KON) to indicate “intent” of the GSPS object
  • How to handle overlapping GSPS

Better Clinical Scenarios

  • User retrieves a KON (or SR) with GSPS references
    • All is well-defined and unambiguous
  • User retrieves a PS and associated images
    • e.g. IHE presentation of grouped procedures (PGP)
    • semantically suspect but technically OK

Summary of GSPS Issues

  • There is no clearly defined mechanism to indicate which GSPS object should be used as the default for which image
  • There is even a lack of supporting data (such as the document title used in KON – CID 7010) to assist choice algorithms

SOP Class Support

  • Greyscale Softcopy Presentation State
  • Colour Softcopy Presentation State
  • Pseudo-Colour Softcopy Presentation State
  • Record application of false colour LUT to greyscale images
  • Private Attributes
  • Convolution Filters (edge enhancement etc.)
  • ROI stats

Creation Granularity

  • Presentation state created for series
  • Multiple states used to record “conceptual” state of one series
  • “Streamlined” method to create presentation states for study

Application Granularity

  • Automatic application of “best” presentation state
  • Apply to series
  • Coalescence of “similar” states to form single conceptual state
  • Apply to image

Metadata

  • Prompt for metadata on creation
  • Content Label (pick list)
  • Description
  • Creator name and date/time
  • Used to choose “best” initial state

AutoAuto

  • Load presentation state + referenced images
  • Load study/series/images + relevant presentation states
  • Applies to Query/Retrieve and DICOMDIR sources

Loose Ends

  • Font selection
  • Curve algorithms (for INTERPOLATED graphic annotation)
  • Representation of existing graphic annotations
  • Angle measurements
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment