Transcript Presentation - Association for Pathology Informatics

Evaluation of
DICOM Supplement 122
at CWRU
Ashok Patel
Rajnish Gupta John Gilbertson
Case Western Reserve University
Cleveland, Ohio
Massachusetts General Hospital
APIII 2007
Harvard Medical School
Radiology had a problem 1980s
RIS
CT Scanner
CT Scanner
Archive
Archive
Dedicated Workstation
Dedicated Workstation
Massachusetts General Hospital
APIII 2007
Harvard Medical School
Massachusetts General Hospital
APIII 2007
Harvard Medical School
Massachusetts General Hospital
APIII 2007
Harvard Medical School
Two Important Inventions
RIS
PACS
Image Exchange Standard
CT Scanner
CT Scanner
Archive
Archive
Dedicated Workstation
Dedicated Workstation
Massachusetts General Hospital
APIII 2007
Harvard Medical School
DICOM – an Image Exchange Standard:
Vendor 2
Vendor 1
CT Scanner
•
1985: ACR (American College of
Radiology) and NEMA (National
Electrical Manufactures
Association) published the first
ACR-NEMA standard for radiology
•
1993: DICOM Digital Image
COmmunications in Medicine
RIS
DICOM
PACS
Workstation
Vendor 3
Massachusetts General Hospital
Vendor 4
APIII 2007
Harvard Medical School
DICOM
•
A remarkably successful standard  Is the basis for virtually all PACS and
multi-specialty Clinical Image Archives…
•
•
•
Very large client community
Very strong vendor community
Over time, it has been responsive to technical and practice changes
•
It managed by NEMA through open, collaborative, international working
groups
•
Working Groups maintain and extend different parts of the standard
Massachusetts General Hospital
APIII 2007
Harvard Medical School
Device
Device
(CT Scan)
(Archive)
Internal Information
Model and Protocols
Internal Information
Model and Protocols
“CT Working Group”
Convert
Convert
DICOM Information
Model and Protocols
DICOM Information
Model and Protocols
Multiple Modules
Transaction
Made up of
DICOM Protocol / Service  “Store”
Image
CT Scan
Information Object
Definition (IOD)
Defined
Type: CT
Patient ID
Study ID
Machine ID
Date
Size
Radiologist
Etc.
Image
Object
Criminally Simplified DICOM
Image & standardized clinical data and metadata
Different for each image object class
Working groups from different specialties can define IODs as needed…
Massachusetts General Hospital
APIII 2007
Harvard Medical School
Pathology, DICOM and WG 26
•
It was designed to be used by other (non-radiology) specialties and many
have done so
•
Initial work between Pathology and DICOM in the middle nineties
•
October 2005: DICOM Strategic Planning Working Group (WG 11) invited a
number of pathologist to Budapest to discuss the possibility a Pathology
Working Group in DICOM
•
December 2005: DICOM WG 6 established WG 26, with scope over all of
pathology imaging
Massachusetts General Hospital
APIII 2007
Harvard Medical School
Pathology has a problem…
Bar coded
Slides
LIS / Histology
WSI Robot
WSI Robot
Image
Archive / Server
Image
Archive / Server
“Virtual Microscope”
On Pathologist’s PC
“Virtual Microscope”
On Pathologist’s PC
Massachusetts General Hospital
APIII 2007
Harvard Medical School
WG 26
•
It is made up of pathologists, the WSI industry and senior members of
DICOM who act as mentors
•
Anyone can join, show up (and work)
•
Seven formal meetings - Phoenix January 06, Madrid, Vancouver, Chicago,
DC., Cologne, Pittsburgh (plus conference calls)
Massachusetts General Hospital
APIII 2007
Harvard Medical School
WG 26
•
Major “initial findings””
– Most of DICOM could be used directly in Pathology, but
– for DICOM to work in Pathology, three main issues had to be solved…
Massachusetts General Hospital
APIII 2007
Harvard Medical School
•
Image
Server
– “Sub image access”
– “Image size”
He’s panning to
the left!
Entire Image
Two of the three issues:
Are outside the scope of this study
Here are the
appropriate tiles
Image
Client
Current View
•
Members of WG 26 are working
with other parts of the DICOM
community to solve these problems
“Sub-image level access”
Massachusetts General Hospital
APIII 2007
Harvard Medical School
Specimens
Massachusetts General Hospital
APIII 2007
Harvard Medical School
Basic DICOM Hierarchy
Specimens in DICOM
Patient
Study
Series
?
•
DICOM expects a Patient to be the
subject of every image
•
In pathology a Specimen is the
subject of an image
•
WG 26 has offered for public
comment DICOM Supplement 122
which defines the place of a
specimen in the DICOM
information model as well as the
specimen attributes that should be
collected when a specimen is the
subject of a DICOM image
Specimen
Image
Very Simplified DICOM
Massachusetts General Hospital
APIII 2007
Harvard Medical School
WG 26: Specimens in the DICOM Model
LIS
Patient
Pathology Modality
Information Objects
Definitions? (IOD)
Modality
Relevant Specimen
Identification and
Processing data
(expected to come
from the LIS)
Study
Series
Image
Specimen
Specimen
Information Object
Definition (IOD)
Criminally Simplified DICOM
Massachusetts General Hospital
“Specimen” can be
associated with any
type of image
APIII 2007
Second draft
this summer
Harvard Medical School
Supplement 122: Specimen Attributes
SS.3.1
Scope
• The Specimen Module (see PS3.3) defines formal DICOM attributes for the
identification and description of laboratory specimens when said specimens are the
subject of a DICOM image. The Module is focused on the specimen and laboratory
attributes necessary to understand and interpret the image. These include:
– Attributes that identify (specify) the specimen (within a given institution and
across institutions).
– Attributes that identify and describe the container in which the specimen resides.
Containers are intimately associated with specimens in laboratory processes,
often “carry” a specimen’s identity, and sometimes are intimately part of the
imaging process, as when a glass slide and cover slip are in the optical path in
microscope imaging.
– Attributes that describe specimen collection, sampling and processing. Knowing
how a specimen was collected, sampled, processed and stained is vital in
interpreting an image of a specimen. One can make a strong case that those
laboratory steps are part of the imaging process.
– Attributes that describe the specimen or its ancestors (see Section SS.1, above)
when these descriptions help with the interpretation of the image.
• Attributes that convey diagnostic opinions or interpretations are not within the scope
of the Specimen Module. The DICOM Specimen Module does not seek to replace or
mirror the pathologist’s report.
Massachusetts General Hospital
APIII 2007
Harvard Medical School
SPECIMEN ATTRIBUTES IN DICOM: SUPPLEMENT 122
Device
Device
(WSI)
(Archive)
Internal Information
Model and Protocols
Internal Information
Model and Protocols
Working Group 26
Specimen
Module
Convert
Convert
DICOM Information
Model and Protocols
DICOM Information
Model and Protocols
Multiple Modules
Transaction
Made up of
DICOM Protocol / Service  “Store”
Image
Information Object
Definition (IOD)
Defined
Type: CT
Patient ID
Study ID
Machine ID
Date
Size
Radiologist
Etc.
Image
Object
Very Simplified DICOM
Image & standardized clinical data and metadata
Different for each image object class
Working groups from different specialties can define IODs as needed…
Massachusetts General Hospital
APIII 2007
Harvard Medical School
Table A.32.X-1
SPECIMEN VL MICROSCOPIC IMAGE IOD MODULES
Module
Reference
Usage
Patient
C.7.1.1
M
Clinical Trial Subject
C.7.1.3
U
General Study
C.7.2.1
M
Patient Study
C.7.2.2
U
Clinical Trial Study
C.7.2.3
U
General Series
C.7.3.1
M
Clinical Trial Series
C.7.3.2
U
Equipment
General Equipment
C.7.5.1
M
Image
General Image
C.7.6.1
M
Image Pixel
C.7.6.3
M
Acquisition Context
C.7.6.14
M
VL Image
C.8.12.1
M
Overlay Plane
C.9.2
U
Specimen
C.7.6.20
M
C.12.1
M
IE
Patient
Study
Series
SOP Common
Massachusetts General Hospital
APIII 2007
Specimen Module are
Defined in a document,
Written by a working group,
Called a Supplement
Harvard Medical School
Specimens in DICOM
•
Supplement 122 asks for the following data elements:
1.
2.
3.
4.
5.
6.
Specimen Container Identifier
Container Identifier
Specimen Identifier
Short Description of the Specimen (Text)
Detailed Description of the Specimen (Text)
Coded Description of the Specimen
1. (code | code system | code meaning)
7. Processing History especially fixation, embedding and staining
1. Specimen ID of Specimen Processed
2. Date Time of Processing
3. Type of Processing
4. Description of Processing
Massachusetts General Hospital
APIII 2007
Harvard Medical School
The study
•
After the DICOM Supplement 122 became available for public comment,
investigators at Case Western Reserve University attempted to implement
the supplement in their LIS to determine if the LIS could provide the
specimen attributes requested in Sup. 122 and if there was important LIS
data that the Supplement should request but does not
•
Essentially we tried to write an LIS (Cerner Copath) query/report that could
retrieve the proposed DICOM specimen data on Gross Specimens and
Slides
•
Looked at 5000 cases (~ 30000 gross specimens and slides)
Massachusetts General Hospital
APIII 2007
Harvard Medical School
Specimen ID
Massachusetts General Hospital
•
Part and Block identifiers were
not problem
•
A unique slide identifier could be
constructed, but it required
joining two tables. The result
was a x.x slide number
•
•
•
S05-100 A 1 2.3
A05-100 A 1 2.5
A05-100 A 1 3.1
APIII 2007
Harvard Medical School
Specimen Description
Massachusetts General Hospital
•
Pathologists descriptions of parts
and blocks were “hidden” in large,
multi-specimen narrative field that
required extensive parsing
•
This was a serious problem – the
“native” LIS could not reliably
provide specimen descriptions
•
Coded specimen descriptions (part
types, etc) were available but were
often non-specific and nondescriptive (‘Big – Other”)
APIII 2007
Harvard Medical School
Specimen Processing
•
How was it fixed? FIXATION and STAINING are from the same dictionary
•
How was it Embedded? NO DATA
•
How was it Stained? FIXATION and STAINING are from the same
dictionary
• An slide stained in H&E and fixed in B5 was reported as “stain =
B5”
• There were multiple ‘types” of H&E processes depending on when
the staining was done: HE Initial, HE, etc.
Massachusetts General Hospital
APIII 2007
Harvard Medical School
Conclusions
•
Our LIS could not implement SUP 122 without major changes
•
This is not about DICOM, this is about our LIS systems
•
The study showed serious limitations in the information model and
implementation of our LIS when it comes to specimen level information
•
Narrative field that contain data on multiple specimens (parts and blocks)
are serious problems
•
Histology data is not stored in sufficient detail
•
Dictionaries are poorly implemented
Massachusetts General Hospital
APIII 2007
Harvard Medical School