DICOM 3D Ultrasound Supplement 43 Overview

1
DICOM 3D UltrasoundSupplement 43 Overview

• Working Group 12
• Ultrasound (3D)
• Joe Luszcz, Chair Pro Tem

2
Goal of Supplement 43

• Storage of
• 3D Ultrasound Volume Datasets,
• 2D views derived from them,
• and related information

3
Current Status of Sup 43

• Just exited Public Comment phase
• ftp//medical.nema.org/medical/Dicom/Supps/sup43_
  pc.pdf
• WG-12 (3D) meeting this week to review comments
  and prepare Letter Ballot draft
• Present Letter Ballot draft to WG-6 in March, 2008

4
Characteristics of US Modality

• Operator-dependent interactive acquisition
• Multiple data types possible in each acquisition
  plane
• Non-image data types acquired with image
• Lack of a fixed frame of reference
• Different probe geometries and oblique
  acquisition planes
• Real-time imaging at relatively high rates

5
3D Ultrasound Use Cases 1

• Data-related cases
• 3D Volume Data
• Static and Dynamic Volume Datasets 1,2,22
• Suitable for rendering and slicing 3,4
• Volume calibration 5
• 2D views derived from Volume Data 6,7,8
• Static and Dynamic varieties 9
• Calibrated frame data 10
• Separation of Data Types 11
• Integrate physio waveforms with image 12

6
3D Ultrasound Use Cases 2

• Workflow-related cases
• Permit Two-Step Review 13
• Frame of Reference
• Frame and Transducer-relative FoR
• Patient-relative FoR 14,15
• Anatomical FoR (Fiducials) 16
• Key Image identification 20
• 3D Presentation State 17,18,19
• 3D spatial coordinates in SR 21

7
Supplement 43 Concepts

• Uses same Enhanced Image structure used by MR,
  CT, XA/XRF in which a new Enhanced US Image IOD
  represents a 3D dataset as a cube of voxels
  (i.e., a spatial sequence of XY planes)
• Data types in separate planes (2D, Flow, etc.)
  with no embedded graphics or text
• Transducer Frame of Reference (primary) with
  optional Gantry and Patient Frames of Reference
• New Enhanced Blending and Display Pipeline for
  combining data types for display
• Uses companion Waveform IODs for related physio
  waveforms
• Derived 2D images represented by US Image IODs

8
Ultrasound 3D Image Types
9
Why a role for 2D US IODs?

• Ease the transition for 3D ultrasound images on
  legacy PACS
• Provide separate SOP Classes for two-stage
  query/retrieval
• Maintain images with integrated data types for
  direct presentation on viewers
• Defer the work of generalizing Enhanced US for 2D
  images

10
Not in Supplement 43

• Line or frustum data representations (if desired,
  use linked Raw Data IOD)
• Use of Enhanced US IOD as 2D replacement (future
  activity)
• Oblique volume datasets (oblique views relegated
  to US-MF IOD)
• Attempts to standardize rendering and slicing
  algorithms (may differ in appearance)

11
What is covered by Sup43?

• New Enhanced US Image IOD for 3D volume datasets
• Single (3D) or Multiple temporally-related
  volumes (aka 4D, 3D loop, 3D animation)
• Slices consisting of one or more frames of
  distinct data types at same space and time
• 3 new Waveform IODs
• Frame of reference for hand-held transducer as
  well as gantry and patient Frames of reference

12
What is covered (continued)

• MONOCHROME2 data representation
• New Enhanced Blending and Display Pipeline
  defines grayscale and/or color presentation for
  single or multiple data types
• Support for multi-palette color LUTs for data
  frames in the same image
• New Ultrasound Application Profiles
• Carry-over of ultrasound exam context from
  existing US image IODs, many as Code Sequences

13
Q A
14
Backup Slides
15
Slice Registration
16
Separation of Data Types
17
Use of Multi-Frame Dimension

• Attributes defining data type alignment use
  existing Multi-frame Dimensions
• Typical order of significance
• Stack ID (most significant)
• Temporal Position Index
• In-Stack Position
• Data Type (least significant)

18
Common Values Sequence

• New Common Values SQ provides high-level view
  of dimensional symmetry
• For each dimension, indicates those attributes
  that share the same value in every frame with the
  same value of the dimension attribute
• Augments, but doesnt replace Per-Frame
  Functional Groups

19
Transducer Marker Orientation

• Attribute specifying the orientation of a
  tactile marker relative to the image data.
  Enumerated values
• POS_X the image is oriented such that the
  tactile marker is in the direction of positive X
• NEG_X the image is oriented such that the
  tactile marker is in the direction of negative X

20
US Frames of Reference
21
Patient Frame Of Reference
22
Patient FoR Transformation
Transducer Gantry Position and Orientation
attributes containing a translation vector SXYZ
and 3 directional cosines XXYZ, YXYZ, and ZXYZ
indicating the position and orientation of the
Gantry FoR relative to the Transducer FoR
according to the following transformation
23
Representation of Color Images

• MONOCHROME2 photometric interpretation used for
  all Enhanced US Image data values
• Presentation State used to associate palette
  color LUTs for color representation (colorized
  tissue or flow/TDI data)
• Types combined in the new Enhanced Blending and
  Display Pipeline

24
Blending and Display Pipeline
25
Is Blending for 2D or 3D?

• Written in terms of 2D Data Frames, but it is
  applicable to
• any 2D result of a slicing or rendering operation
  that preserves individual data types
• transformation into RGB volume prior to slicing
  or rendering operation

26
Physio Waveforms

• Conveyed in separate Waveform objects
• Linked with the image object via bi-directional
  SOP Instance UID references
• Synchronized via Acquisition Datetime values in a
  common time reference
• With each waveform type in a separate Waveform
  SOP Class
• With gated acquisition info via R-Wave
  identification through the Waveform Annotation
  Module and excluded beats identified via the
  Volume Timing module, permitting our Live 3D
  stacked ECG display

27
Stacked ECG Display
28
Volume Timing
29
Known Open Issues

• 1) Section C.8.X.1 Coverage of 3D Stress and
  improved 2D Protocols (Staged and Non-Staged) is
  intended. Does this provide enough flexibility
  for so-called custom protocols?
• 2) Section C.8.X.1 Not sure what should be
  included in Protocol Context or its Modifier.
  added v25
• 9) The Enhanced Palette Color Lookup Table
  describes an improved Enhanced Blending and
  Display Pipeline. Is the level of complexity
  correct (flexible enough, yet implementable in
  readers). Please review this pipeline to verify
  whether the flexibility is sufficient yet not
  excessively complex. added v31
• 10) No new Presentation State SOP Class has been
  defined to reference the Enhanced Blending and
  Display Pipeline. added v31
• 11) Add authoritative references for sample
  frequency range for Arterial Pulse Waveform,
  Respiratory Waveform and Phonocardiographic
  Waveform. added v32
• 12) What are the appropriate numbers of channels
  and sample frequencies for the new Pulse,
  Respiration, and Phonocardiologic Waveform IODs?
  Are the Waveform Sample Interpretation attribute
  values correctly defined (linear, logarithmic,
  etc.)? added v33

30
More Known Open Issues

• 13) Should the Enhanced US IOD require the use of
  the VOI LUT module or macro? added v33
• 14) Would a blending pipeline with symmetric
  primary and secondary path be more helpful in
  terms of easier implementation considering that
  it would be implementation of symmetric path or
  would that increase the complexity of the
  pipeline without much value to it? added v33
• 15) What should be the media application profile
  for Ultrasound? Should the General Purpose media
  profile be used, or should the existing
  Ultrasound application profile be extended for
  Enhanced Ultrasound object, as appears in this
  draft? added v35

31
More Known Open Issues

• 16) The Enhanced US Image Module (Section
  C.8.X.3) includes a number of optional attributes
  from the existing US Image Module (ps3.3 Section
  C.8.5.6). Is it appropriate to keep the
  following Type 3 attributes in the Enhanced US
  Image Module, and if so, is the form correct
  (defined terms vs. code sequence values, for
  example)?
• Beat Rejection Flag (0018,1080)
• Transducer Data (0018,5010)
• Focus Depth (0018,5012)
• Preprocessing Function (0018,5020)
• Mechanical Index (0018,5022)
• Bone Thermal Index (0018,5024)
• Cranial Thermal Index (0018,5026)
• Soft Tissue Thermal Index (0018,5027)
• Soft Tissue-focus Thermal Index (0018,5028)
• Soft Tissue-surface Thermal Index (0018,5029)
• Depth of Scan Field (0018,5050)
• 17) The Transducer Type attribute (0018,6031)
  attribute has been replaced with the two code
  sequence attributes Transducer Type Code
  Sequence (0018,xx09) and Transducer Type Modifier
  Code Sequence (0018,xx0A), with corresponding
  Context Groups CID xxx05 and CID xxx06. Is this
  information necessary? Is this scheme acceptable
  and are the choices in the Context Groups correct
  and complete? added v35 and modified v38