DICOM Supplement 49 Extended MR DICOM Objects Korean PACS Conference

1
DICOM Supplement 49Extended MR DICOM
ObjectsKorean PACS Conference
GE Medical Systems
Charles Parisot May 5, 2002
2
New MR Image Objects Why ?

• To address with full interoperability, add
  acquisition techniques such as
• Diffusion Imaging,
• Perfusion Imaging,
• Angio Imaging,
• fMRI Imaging,
• Cardiac Imaging and
• Spectroscopy for MR

3
3 New MR Image ObjectsDICOM Supplement 49

• Enhanced MR Image Object

• MR Spectroscopy Object

• Raw Image Data Object
• standardized enough to allow network, archive
• otherwise vendor dependent

4
A new Generation of Image Objects Why ?

• Performance and ease to managethe Exploding
  number of imagesin an MR acquisition

5
A new Generation of Image Objects Why ?

• Performance and ease to managethe Exploding
  number of imagesin an MR acquisition

• More complex inter-relation between these images

• Real-Time Imaging increasingly used

Solution Concatenation of Multiframe Image
Objects
6
MR Multiframe High Level Requirements

• Provide a way of efficiently organizing large
  groups of images (1,000 to 10,000 Images)
• Provide a way of organizing any group of images
  (cine loop, Peripheral Vascular stations/phases)
• Allow for organizing of image sets associated
  with a single complex multiphase application
  (Peripheral Vascular localizers, stations/phases)

Dynamic Imageswith up to 100 dimensions !
7
Multiframe OrganizationFlexible separation of
Static vs Dynamic Attributes

• Always Static Attributes - They never change per
  frame in a multiframe
• Will never need to change within an Image Object
  gt changes in these require the start of a new
  Image Object
• Goal is to reduce complexity in receiving
  application
• Facilitates use of current toolkit technology.
  Examples
• Pixel ( bits, matrix size, etc.)
• Pulse Sequences

8
Multiframe OrganizationDynamic Attribute
Groupings

• Dynamic Attributes may change per frame in a
  multiframe. They are separated in Functional
  Groups that often change together
• To reduce the number of changeable entities
  within a multiframe object
• to allow for modality independent reuse
  modality independent vs dependent
• to convey semantics to the receiving application
  e.g., MIP program may not accept a MF object that
  has an orientation group that changes
• Special care has been taken to "balance" the size
  of groupings.

9
MR Multiframe Organization 24 Dynamic
Functional Groups
10
MR Multiframe Organization 24 Dynamic
Functional Groups
MR Image Instance

• For a Specific MR Image Instance
• some Functional Groups are shared across all
  frames,
• some vary per frame

11
A new Generation of Image Objects Why ?

• Performance and ease to managethe Exploding
  number of imagesin an MR acquisition

• More complex inter-relation between these images

• Real-Time Imaging increasingly used

Solution Concatenation of Multiframe Image
Objects
12
Reasons to break upMultiFrame Objects

• File systems file, partitions, or storage media
  size limits
• To provide pseudo real-time streaming (fMRI from
  scanner to workstation for real-time monitoring
  and processing)
• To provide for retransmission in chunks in the
  case of network transmission failures
• Standard Forced Breakup due to dynamic
  attribute being defined as a static attribute

13
Concatenations
An object may be split up into two or more SOP
Instances
e.g. After frame-numbers 2000, 4000 and 4200
3
4
2
1
Image attributes
Pixel data for frames of set n
n
Concatenation Frame Offset Number (e.g.1, 2001,
4001 and 4201)
Shared Dimension Moduleattributes
14
Concatenation of MF Objects

• Examples total body scan stations, fMRI broken
  into time segments
• Concatenation UID is used to group image objects
  belonging to the same concatenation
• All concatenated objects must have the same
• Instance Number
• Frame of reference
• Series number UID
• Dimension modules

15
A new Generation of Image Objects Why ?

• Most attributes are made mandatory for greater
  interoperability
• Many old attributes not used removed
• Anatomy specification required to facilitate PACS
  handling
• Image Relationship Referencing generalized
  and added coded reasons for reference
• New image pipeline (LUT and Color Palette)
• Image Types
• One sacrificeCreate a New Enhanced MR Image
  Object, different and incompatible with the
  existing MR image Object

For A Higher Level of Compatibility
16
MR Object Relationships
17
LUTs
18
Real World Value LUT

• Sometimes the integer pixel values and what the
  user wants to see for pixel values based on real
  world units (such as blood flow velocity at a
  pixel location) are different.
• The Real World Value LUT maps the pixel data to
  the units the user wants to see (cm/sec or
  mm/sec)
• Multiple overlapping regions of pixel values can
  be mapped to multiple LUTs including regions that
  are not mapped at all (e.g. functional data
  versus physiological data).

19
Palette Color Pipeline
20
Palette Color LUT

• Palette Color LUT is used to map Monochrome2
  pixels to color
• Mixed Monochrome2 grayscale pixels and palette
  color pixels can be shown in the same image (e.g.
  to show functional data in color on top of
  physiological data).
• Only a single grayscale and single color range of
  pixel values can be represented.

21
Image Type AttributesMR Image Description Macro

• MR has a large, rich set of image types
• Applications need a way to determine if an image
  set is compatible with its processing
• Supplement 49 proposes a reasonably orthogonal
  set of attributes for image type useful to
  reading applications
• Image Type (0008,0008) values
• 1 Original/Derived redefined
• 2 Primary/Secondary Only Primary valid for MR
• 3 Image Flavor the overall most important
  characteristic of this Image e.g. flow encoded,
  max-IP, Perfussion, Stress, T1, T2, etc.
• 4 Derived Contrast Diffusion aniso,
  Subtraction, Velocity, None generally an
  indication of post processing performed

22
Image Type Attributes 2MR Image Description
Macro

• Other Image Types are separate attributes
• Pixel Presentation (Palette) Color/Monochrome
  (color supported or not)
• Volumetric Properties Volume, Sampled,
  Distorted (used by Grx, 3D to determine image
  compatibility with the application)
• Volume Based Calculation Technique MAX_IP, MPR,
  Curved-MPR (used by Grx, 3D to determine image
  compatibility)
• Complex Image Component - Magnitude, Phase, Real,
  Imaginary (standard MR transformations of the raw
  data)
• Acquisition Contrast - T1, T2, Perfusion,
  Combination (MR acquisition contrast types)

23
New MR Image Objects Why ?

• To address with full interoperability add
  acquisition techniques such as
• Diffusion Imaging,
• Perfusion Imaging,
• Angio Imaging,
• fMRI Imaging,
• Cardiac Imaging and
• Spectroscopy for MR

Each Application has specific viewing
characteristics..
24
Order in Viewing ?

• Many parameters can change from frame to frame.
• For the most important, those that define certain
  relations between slices, specific tags have been
  defined to indicate and order the relation.
• A dimension will consist of number of... tags
  with the highest ordinal number of every
  dimension.
• Sorting images according to those ordinal
  numbers, and repeating that for another
  dimension, will enhance receiving applications
  and interoperability

25
Dimension Attributes Examples

• Stations
• Stacks
• Positions
• In-stack slice Position (slice relative to
  stack)
• Orientations
• Trigger delay times
• Temporal positions
• Diffusion B values
• Metabolite maps
• Echoes

26
Dimensions Use of Indexes

• Examples can be given in many areas, but in
  general the mechanism uncouples the actual value
  of a certain attribute actual position vs.
  position number
• In some cases the increase of attribute values
  will be related to that of the index numbere.g.
  Trigger delay time (in ms) increases with the
  Trigger delay time index 1....6
• In other cases these are completely uncoupled
  Orientation(patient) and Orientation Index
• In some cases the index looks very much like the
  attribute value.

27
Examples of properties that may change,
cardiac phase
b-value
orientation
time
position
volume
time
28
Cardiac Example 1 station, 1 stack, n trigger
delay times
Trigger delay time index
1
2
3
Frame number 1-6
Frame number 13-18
Frame number 7-12
time
29
Diffusion Example 1 station, 1 stack , 3 b-values
B-value Index
3
2
1
Frame number 1-6
Frame number 13-18
Frame number 7-12
time
30
Example 1 Station, 3 stacks
31
Multi Stack example (parallel and non-parallel,
2D and 3D)
Stack 3
Frame number 13-18
Stack 2
Frame number 7-12
Stack 1
Frame number 1-6
32
A new Generation of MR Image Objects
Functional Groups of Dynamic Attributes
Dimensions and Indexes
Multiframe
Concatenation
New Enhanced MR Image
Anatomy specification required
Most attributes mandatory
MR Object Relationship and Referencing
LUT for Real Values and Color Maps
Raw Images
MR Spectro
33
DICOM Web Site

• http// medical . nema . org