DOC

WG-11 MINUTES

By Douglas Turner,2014-01-20 04:58
15 views 0
WG-11 MINUTES

    NEMA, Suite 1847 th1300 North 17 Street

    Rosslyn, VA 22209

    Ph: (703) 841-3285

    Fx: (703) 841-3385

     WG-11 MINUTES

    DICOM Working Group Eleven

    (Display Function Standard)

    Meeting: December 2 - 3, 2002

     Chicago, IL

    Members Present:

    American College of Radiology Alan Rowberg (M, C, T)

    American College of Radiology Brad Hemminger (M, C, T)

    Cedara Software Minglin Li (C)

    Eastman Kodak Ellie Avraham (C, T)

    Fujifilm Medical Systems USA Makoto Ogoda (C)

    Philips Medical Systems Cor Loef (T)

    R2 Technology Janet Keyes (M, C, T)

    Simon Fraser Univ. Adrian Moise (M, C, T)

    M = Monday morning; C = Monday afternoon Color Session; T = Tuesday

    Members Absent:

    *Food and Drug Administration Kish Chakrabarti

    Siemens Medical Systems Michael Jonas

    * = Does not count toward quorum.

    Others Present

    Agfa Health Care Rob Horn (C)

    BarcoView Geert Carrein

    DR Systems, Inc. Dan Konigsbach (C)

    ETIAM Gilles Mevel (C)

    General Electric Thomas Layden (C)

    General Electric Harry Solomon (C, T)

    Integrated Color Solutions Jack Ladson (C)

    Merge eFilm John Fehrenbach (C)

    Philips Medical Systems Nederland Kees Verduin (C)

    190 WG-11 (Display Function Standard)

    of the DICOM Standards Committee

    December 2 3, 2002

RadPharm David Clunie (M, C, T)

    Roentgen Gesselschaften Peter Mildenberger (C)

    Siemens Medical Solutions USA Chris Amies (C)

    Siemens AG Heinz Blendinger (C)

    Siemens Medical Solutions USA Dennis Falkenstein (C)

    Siemens Medical Solutions Lawrence Tarbox (C)

    Societe Francaise de Radiologie Joel Chabriais (C)

    Tech Source Joseph Lamm (C)

    Toshiba Hidenori Shinoda (C)

    University of Michigan Mike Flynn (M)

    U.S. Food and Drug Administration Aldo Bandano (???) (T)

    NEMA Staff Stephen Vastagh

    Presiding Officer: Brad Hemminger, Chair

    1. Preliminary Events The participants identified themselves and their employers.

    The agenda was accepted.

    Members reviewed minutes of the September 19 20, 2002 meeting. They were approved subject

    to the following two modifications:

    ? In section 6, change last paragraph to read:

    Preliminary versions of the examples based on version 6 of supplement 60 were

    reviewed except for the one from Dr. Rowberg. However, not all were complete.

    This led to several revisions and simplifications of the supplement, in particular

    definitions of positions and size of display set boxes and image boxes have been

    moved just to the image box level. Also, synchronization is defined only of display

    sets (and outside of display sets).

    ? In section 10 change the last paragraph to read:

    Subsequent meetings will be Jan 9-10th 2003 in Rosslyn, and February 21st in

    conjunction with the SPIE meeting in San Diego. The January meeting times are

    9-5pm on Jan 9th and 8:30-3pm on Jan 10th. The Feb meeting time will be from

    9 AM to 5 PM.

    2. Review Previous Homework Assignments

    Brad Hemminger reminded members of the following assignments that were made at the conclusion

    of the September-2002 meeting:

    Everyone:

    ? Should complete their examples and upload them to the NEMA server.

    191 WG-11 (Display Function Standard)

    of the DICOM Standards Committee

    December 2 3, 2002

    ? Review the Japan IC document and related documents.

    ? Prepare for Color discussion at December meeting.

    ? Review highlighted (yellow-noted) items in version 7 of supplement 60 and prepare a

    response for each.

    Janet will:

    ? Complete HP Supplement (version 07) and put it on the NEMA server.

    ? Present an early draft of the supplement to WG-06 at its October meeting. (Schedule for

    January if you cannot get on the October agenda.)

    Brad will:

    ? Respond to David Clunie.

    ? Respond to Doug Sluis. Also, Alan et al. will send an e-mail to Doug. Craig Cornelius will:

    ? Prepare a document describing the appropriateness of the ICC standard for standardized

    medical image display; or at least what the difficulties or shortcomings might be based on Ed

    Giorgianni’s book and Craig’s discussions with him. Upload this document to the NEMA

    server.

    3. Hanging Protocols

    A detailed list of issues raised by David Clunie and other participants during the review of Sup60

    (and the associated examples) is given below. Follow-up discussion and changes are noted with

    each issue; the remainder were considered open discussion issues that would be dealt with later, after

    further review.

    A. Incorporation of GSPS in HP. David Clunie felt that it was an issue expecting a reference to an

    existing GSPS i.e. this would create a great burden on creating GSPS for possible conditions. He

    gave the use case of where GSPSs may not be in use, but HPs were and the users (radiologists)

    wanted to indicate a preferred IW setting. After some discussion, members felt the best path would

    be to have some GSPS type information included in HP object. This would provide a default, which

    would be overridden if a GSPS for that image study were present. That is, the GSPS still takes

    precedence and provides the final say on processing (so that there is a single point of definition of

    what processing is applied). A second purpose of having the HP define some presentation items is

    that, if more than one GSPS is present, then, the HP definitions could be used to help select the most

    appropriate GSPS. Overall, this would allow HP to work well with and without the presence of

    GSPS for the image study being reviewed. Follow-up work includes a need to identify the parts that

    should be internalized in the HP. Members expect to base this on what they previously identified

    (see page 55 in supplement 60 version 7). Members will review these and incorporate the results at

    the Jan meeting. [To be included in Sup60 v9 in Jan.]

    B. Extend choicesuse an extensible coded list. Be sure to pass by the modalities to get feedback

    to make sure we incorporating what their needs are. Maybe have to allow flexibility on

    conformance to support this item.

    192 WG-11 (Display Function Standard)

    of the DICOM Standards Committee

    December 2 3, 2002

C. Scope and interoperability should clearly define expectations. In general, we should review and

    more carefully state what the expected outcomes are. As an example of this, David Clunie

    suggested we choose to say:

    ? If image set Y is missing (not retrieved) we expect the behavior to be (?)

    ? Display areas the display set depends on Image Set Y should just display as empty, i.e.

    maintaining the same layout as originally intended (as if Image Set Y was present).

    ? Vendor chooses how to adapt to handling missing data

    That is, HP should spell out expected conformance. In this case we added a new attribute Partial

    Data Display Handling in Display Sets to allow the user to specify their preference between the two

    above options. [Modified in sup60 v8.]

    D. Regarding our nested relationships to handle our OR and AND sequences. This is new thing

    that has not been handled previously in DICOM. Members want to make sure this is done in a

    reasonable fashion. It should also be done in a way that is general enough to be reused later. After

    some discussion it seemed that what we had was acceptable.

    E. Choose a single existing coding scheme and values i.e. one representation of anatomy/Chest instead of a allowing a specification via ORs of CPT, SNOMED and DICOM group/element.

    Choose one schema, and give that schema and value; and let vendors match to other schemas. So,

    for instance, on the Chest CR example, don’t do Item 1 (anatomy /chest) and Item 2 (SNM3 /T-

    D3000/meaning Chest). Do just one instead. Note, if the procedure intent is condensed into one

    code, then it may be necessary to handle local codes differently. For instance, a CT/MR surgery

    example condense everything into one but keep the local procedure code as well. Note, we still

    want OR sequence to be able to specify multiple conditions like ―MR brain or MR head‖.

    Members discussed this further on Tuesday and proposed the following modification: first, use

    Modality and Anatomy context groups as baselines; i.e., have single controlled vocabulary for these

    coming from DICOM, and allowing for extension from the baseline. Second, break out Procedure

    intent into Procedure and Reason. Second, make an Open Issue: Advocate DICOM creation of

    context group for ―procedure‖ and ―reason‖. Until then, keep the existing methods for procedure and for new ―reason.‖

    Modality: CP 310, context group 29, letter ballot completed this week.

    Anatomy: Context 4 (already defined).

    Procedure: no baseline context groups, attributes that exist are free text

    Reason: no baseline context groups

    [Draft revision in Sup60 v8.]

    F. CPT codes are licensed. It may not be possible to use them for the standard. It is important to

    determine whether or not we can use CPT codes without licensing. Given the new proposal in (E),

    this may become moot if we do not use CPT codes.

    G. Is ref_patient really needed? Keep to minimum important stuff. Members decided to move this

    back out to an open issue to get feedback and to initially not include it. Brad Hemminger (who

    originally proposed it) felt the uses were mainly academic vs. business use cases (i.e. for teaching

    files or to compare against composite ―average‖ or ―model‖ patient). [Revise in Sup60 v8.]

    193 WG-11 (Display Function Standard)

    of the DICOM Standards Committee

    December 2 3, 2002

H. David Clunie suggested improving the clarity and information content of the figures (e.g.,

    include more of the time sequence in figure 1). He advocated using UML to convey interaction

    sequence for instances like figure 1.

    I. There seemed to be some confusion in the display environment module regarding ―Hanging

    intent‖ vs. ―hanging description.‖ That is, some of the terms would likely be filled out with the

    display environment values of the workstation that was used to create it. The description should

    more clearly indicate that they should be used as the recommended display environment to match.

    A second, related issue was that, maybe in some cases, the specification should be ―don’t care‖

    instead of a value i.e. don’t care if color display or not – versus specifying 8 bits of color.

    Members would like to have a way to represent this. Members decided to make these type 2, so that

    they must be specified but could be empty. If empty, then, this will mean ―don’t care,‖ i.e., no

    specification was intended. [Revise in Sup60 v8.]

    During this discussion Adrian Moise suggested the concept of ranges of values to help in better

    specifying intent. This was deemed a good idea, and members decided to allow a range specification

    as a double value, as well as the original single value specification. [Revise in Sup60 v8.].

    J. David Clunie suggested talking with print folks to learn from their experience regarding the

    display module. Members had already reviewed the hardcopy film work when coming up with their

    display sets and image boxes. They plan to encourage comment from print people during the Public

    Comment period.

    K. David Clunie noted that members suggested that they have:

    ? General format model (for everything except image_type_modality)

    ? Specific information (like what they have for image type modality, except not limited to

    type. For instance, in Sup49, MR is extending to add new attributes, not just type values). Note, also, that there are now image level and frame level (values mixed).

    After review, members realized this was the same as OPEN ISSUE #10:

    The attributes of the Filter Operations Sequence specify only a Filter-by

    Category, Operator, and Value. Is there a need for a Filter-by Attribute, to

    allow specifying a DICOM tag and value by which to filter the image set?

    Based on the apparent need and the increasing definitions of new attributes (as in Sup49), members

    decided to add ―filter-by,‖ as discussed in #10, but keep the open issue to get discussion on this topic

    at review time. [Revise in Sup60 v8.]

    L. Image set max compression: David Clunie suggested that this be left out. After some

    discussion, members agreed that it should not be included in the standard yet. They are not sure

    how soon it will be, but, according to Alan Rowberg, it would be after HP. They reasoned that it

    would be better to add afterwards, if there is a real need, and be able at that time to base it upon

    what is actually put in standard. Members decided to remove this from Sup60 but include it as an

    open issue to keep attention on this during the review process. [Revise in Sup60 v8.]

M. Sup49 has added some new terms, i.e. MIP, MIN, subtraction, diffusion, etc. Members believe

    they should incorporate these terms (as opposed to making up their own). Examples cited by David

    194 WG-11 (Display Function Standard)

    of the DICOM Standards Committee

    December 2 3, 2002

Clunie included Value 4 of Image Type and Volumetric Properties, Volume based calculation

    technique. [Revise by adding in Sup60 v8.]

    N. Review/Reconsider choice of composite vs. normalized. David Clunie suggested that

    descriptions would be easier if composite and that we cannot exchange on CD (media) unless

    composite. Note that this was previously proposed to WG-06, which recommended that WG-11 to

    pursue normalized, which we have done successfully. Initially, this seemed to be a major change,

    but, after some discussion with David Clunie and Janet Keyes, it may not be unreasonable.

    Members agree to keep this as an open issue until they can get an initial review from WG-06.

    O. Mike Flynn joined the meeting and brought up the issue about GSDF and the effect of

    luminance range on perceptual standardization. There does seem to be some effect. Members

    agreed that it would potentially be good to include this as an attribute (max luminance /min

    luminance). This would help vendors more appropriately choose processing. (Thus, as

    demonstrated by Mike Flynn in InfoRad 2001, modifying the IW settings can compensate for

    different luminance ranges to better maintain similarity of image presentation). This is related to

    similar thoughts regarding MTF, as proposed by Hartwig Blume, to include some specification-

    expected display. Members will consider working these two through together as change proposals.

    The CPs would add attributes to indicate the expected display’s max/min luminance and some

    measure of MTF (not defined yet).

    P. Structured display use cases Harry Solomon cited a use case where physicians want to play

    back set of images in a structured format for conferencing like a softcopy version of print box.

    Harry Solomon indicated that he would be willing to shepherd this and is going to prepare a white

    paper proposal on this.

    Q. Harry suggested using Cine attributes from Part 3 Cine module (C.7.6.5) if possible to avoid

    creating new terms/attributes.

    4. Adding Color to GSPS Approximately 25 individuals attended the Open Meeting on Color that was held on the afternoon of

    December 2. As the session proceeded, WG-11 Chair Brad Hemminger recorded the following

    observations:

    Brainstorming Thoughts

    Marketing vs. scientific (It’s important to have some consistency from the marketing perspective.

    The question is how much is scientifically needed?)

    Think about the semantics; where should this information reside. E.g. in marking up lesion or

    indicating blood flow in US an association is established between label and semantic meaning. Should semantics be captured in the presentation/display objects? WG11 has generally tried to keep

    them separate (except for the proposed future ―Structured Display‖).

    Assumptions

    ? Want to use (and support) off the shelf technology digital cameras, monitors, and existing

    calibration.

    ? Utilize a public existing standard if possible. Some standards include VESA, ICC, CIE

    standards, and prepress proprietary standards.

    195 WG-11 (Display Function Standard)

    of the DICOM Standards Committee

    December 2 3, 2002

General Needs

    ? Color consistency (both coarse accuracy and more exacting accuracy):

    ? Color presentation in conjunction with accurate greyscale presention (i.e. have color in

    conjunction with GSDF&GSPS).

    Accuracy/Fidelity

    There seem to be two main uses:

    ? Labeling (coarse, for instance Doppler on US) or

    ? Continuous (visible light, 3D volume rendering in continuous scales).

A number of use cases were also identified (partial listing, many more, but mostly overlap the others

    recorded during the brainstorming session).

Need for Color Consistency

    Endoscopy: (visible light)

    3D volume rendering

    secondary capture

    196 WG-11 (Display Function Standard)

    of the DICOM Standards Committee

    December 2 3, 2002

Need for CSPS

    US: color Doppler, grayscale structure. Want to apply GSPS to zoom, rotate, etc., but cannot

    because GSPS doesn’t support color. Would like integrated color and gray presentation. Want

    luminance to abide by GSPS and want to add pseudo color along with this.

    Unclear needs

    Functional allocations: fusion, color coding veins vs arteries, organs different colors.

    Example: Functional MR: z score in color, anatomy in greyscale.

    3D volume renderingneeds pipeline that may be superset of 2D CSPS. 3D visualizations: volume rendering, polygonal rendering

    Multi-parameter (dimension) fusion:

    Combining multiple modalities (acquisitions)

    Combing functional and anatomical

    Should we include modeling light sources? I.e. assumed currently; because more of an issue with

    3D rendering (ability to change/modify the lightsource(s)). Endoscopy takes a reference calibration

    image to calibrate the light source/scene.

Issues

    If they have to work hard to calibrate, users aren’t interested. Expectation is that most will use

    default ICC profile.

    Use of something like ICC for color consistency may place some restrictions on older slower

    graphics cards that may not support such transformations in ―real-time‖ for multiple windows on the

    scene.

    Some use cases need to match human perceived colors: i.e. camera images show similar to what

    human observer would see.

    Scene spectrum: measurable photonic energy in wavelengths (independent of sensor system)

    Display possibilities

    ? Realisticmatching human visual system

    ? Syntheticpsuedocolor/arbitrary (synthetic only with respect to human observers)

    197 WG-11 (Display Function Standard)

    of the DICOM Standards Committee

    December 2 3, 2002

How to accomplish these two proposed work items in DICOM?

    ICC:

    Based on previous analysis done by WG-11 and discussion, members propose using ICC, device

    independent paradigm, and perceptual method. Note, related to this, do we need to constrain to

    perceptual space in ICC; or can this be chosen dynamically? I.e. so as to not rule out using one of

    the other four methods should they be desired by applications.

    ICC can profile most everything. Facilitates device independence between cameras and scanners

    and medical imaging equipment which already have ICC profiles. Small devices (that don’t have

    resources to profile each device) could use ICC default profiles.

    Can consistency be done just via ICC with the rest of the use case needs being separately handled

    through Color SPS?

What in DICOM needs to change?

    ? Where stored?

    ? Options: new optional attribute at image level, has to have some text to explain

    assumptions if not present. (make new object if need to change palette).

    ? Single attribute as a blob containing ICC profile

    ? Behavior/conformance for part 4

    ? Need input from print vendors

What do device manufacturers need to provide?

    ? Standardization: use defaults or potentially need to calibrate

    ? May require image science experts to help characterize device (i.e. steps not spelled out like

    GSDF, requires some expertise).

    ? Measuring lighting/viewing environment

    ? Choice of palettes suitable to support users with anomalies in color vision

CSPS:

    Do separately or extension to GSPS? I.e. CP to GSPS? New CSPS paralleling GSPS? Completely

    new color IOD? Expected plan is to do CSPS paralleling existing GSPS.

    Next steps: work item requests for each to DICOM committee.

    198 WG-11 (Display Function Standard)

    of the DICOM Standards Committee

    December 2 3, 2002

    Questions if we do CSPS

    ? It was suggested that we work to improve handling of multi-frame. For example in cine

    module. Use cases US and MR multiframe color & grey presentations.

    ? Do we want to specify or allow operation between two images. I.e. have chroma in one

    image (functional MR) and anatomical in greyscale in second image.

    ? What kind of operations between two images would we want to support?

    ? Thresholding

    ? Colorization

    ? Are there standards for color transformations? Many operations, mostly all very

    specialized. Not clear if there is a set (or any) that would be useful to standardize. ? If we add color to GSPS would have to add via optional type 3 attributes. This may be

    problematic if we want to include these as mandatory attributes.

    ? Should the new Color/Grey SPS replace usage of the GSPS, or be used in conjunction with

    the GSPS.

    ? New CG/SPS could cause legacy issues with GSPS if it contains new required attributes.

    Maybe better to have second CSPS that works in conjunction with original GSPS. If so,

    perhaps best to wait to have significant important improvements, e.g. merging two images.

    Could be tied to work in WG17 on registration and fusion.

    ? Should a new CSPS include 3D as well as 2D?

    General Issues

    ? What group(s) should address this?

    ? ICC: WG11,(WG13?)

    ? CSPS: WG11, (WG12?, WG13?)

    ? What participants?

    ? Brad Hemminger asked whether anyone was willing to serve as an editor? There were

    no volunteers at this time.

    ? Brad Hemminger asked whether anyone was willing to participate in meetings and work

    on the proposed two work items? (About six attendees, including representatives from

    Kodak, Agfa, Techsource and GE, indicated an interest.) They expressed an interest in

    teleconferences versus in-person meetings.

    ? Next Step?

    ? Follow-up meeting during the second half of the WG-11 meeting at SPIE on Feb 21st.

    This will be to follow up on these issues, to have an informative presentation on ICC,

    and to discuss further the proposed work and work items, as well as trying again to

    199 WG-11 (Display Function Standard)

    of the DICOM Standards Committee

    December 2 3, 2002

Report this document

For any questions or suggestions please email
cust-service@docsford.com