Transcript MIRC at the National Cancer Institute

Medical Image Database
Resource for Software
Tools Evaluation and
MIRC
Laurence P Clarke, PhD
Barbara Croft PhD
Carl Jaffe MD
Gary Becker MD
Dan Sullivan MD
Cancer Imaging Program
Resource for Benchmarking Application
Specific Software Performance.

Need to develop a consensus or standards for assessment of
application specific software for cancer imaging.

Software methods and related informatics tools lag behind the
front end imaging technology in terms of their performance
(MRI,CT, Molecular Imaging..).

Need to engage the national and international community in
the development and assessment of advanced software tools.

Need to explore if this resource could be employed to reduce
delay or uncertainty in FDA approval.

Need to develop IT resource to establish this database and
provide web access to this resource to encourage standards
for software assessment.
Example Projects: Evaluation of
Software Tools as required Lung
Cancer using X-ray CT.


Lung Image Database Consortium (LIDC):
Database to permit the benchmarking of
CAD methods for cancer screening.
Lung Database for Drug Response.
Database to permit the benchmarking of
software metrics such as volumetric
measures of tumor response to drug or
radiation therapy.
U01: Lung Image Database Consortium
(LIDC) (U01: 2001-2006):

NCI: ACRIN/NLST (2002-2008: $200M/9 years) is
supporting a large lung cancer screening trial using
helical CT, where subsets of images can be collected.

LIDC ($7M/5 years: 5 Academic Sites). Goals:
Develop an image database for the comparison of
CAD methods, and encourage standards for
assessment of application-specific software.

LIDC Steering Committee: Includes academic
members from each site, FDA CDRH scientists and
consultants that are engaged in the developing broad
understanding of all the issues involved in the
development of a public database, as a transparent
process.
Foundation for NIH: Facilitate
Public Private Partnerships

Authorized by Congress in 1991--and established in
1996-- to support the NIH in its mission through
collaboration with academia, industry, individuals, and
other non-profits.

An independent non-profit, not a government
organization.

Governed by independent Board of Directors, including
the Director of NIH and the FDA Commissioner as exofficio members.
Partnership Model
Based on the Osteoarthritis Initiative
$$ Flow
Private
Partners
FNIH
Information Flow
Steering
Committee
FDA
NIH
Grants or
Contracts
FNIH-NCI Plans (2005-06):
Expand the proposed databases as an
industry resource

Form an overarching steering committee with
FNIH, NCI, Academic and Industry
representation

Identify specific milestones for early delivery of
databases acquired from NCI clinical trails.
• Lung Cancer Screening: Expand the number of sites
for X-ray CT include digital chest X ray (Jan 2005)
• Lung Cancer Drug Response: X-ray CT. Expand Pilot
Project (Sept 2004): PPP timetable: Summer Fall 2005
The Database Resource
Research Community
MIRC sites:
•
•
•
•
•
•
•
•
•
NCI
FDA CDRH
Cornell University
MD Anderson Cancer Center
Memorial Sloan Kettering Cancer Center
University of California, Los Angeles
University of Chicago
University of Iowa
University of Michigan…
• Industry Partners
Meta Data: Results of Unblinded Reads: Nodule Location
4/4 Markings
2/4 Markings
2/4 Markings
We will capture one aspect of reader variability in this way
Metadata: Description of Nodule
• A spectrum of abnormalities?
(
Nodules
)
Focal Abnormalities
Scar
Spiculated
Nodule
Calcified
Nodule
Meta Data: Nodule Characteristics
Library to Standardize Radiologist
Interpretation.
Meta Data: Nodule Boundaries
Example Spiculated Nodule: Comparison of Contours
Meta Data: Probabilistic Description
of Boundary
RSNA: 2003:LIDC: Fundamental Issues for the Creation of a International
Research Resource for the Image Informatics Community.
NCI Current and Future Clinical Trials.
Potential Sources of Data and Timeline
Year 1
Year 2
Year 3
Meta Data: NCI Informatics Infrastructure
Planned Reference Database Resource to be Fully
Integrated into caBIG Informatics GRID.
GRID
Principles
•Open source
•Open access
access
portals
Clinical
Trials
•Open
development
Mouse
Models
Group
Nodes:
Cancer
Centers
SPORES
CGAP
MMHCC
Others
Molecular
Signatures
NCI
Core
•Federated
participating
group nodes
TGI
Cancer
Genomics
GAI
CCAP
Database
Resource
Users
NCI and FDA
Cornell
MIRC
Site
MDA
MSK
UCLA
Chicago
Iowa
Mich.
MIRC sites
Internet
PACS
n
Workstations
PACS
DICOM
Database Resource Several
Demonstration Projects 2005-06
Modality
CTMS
(images)
DICOM
(csv data)
MIRC
Clinical
Trial
Service
8443
HTTP(S)
8080
MIRC
Clinical
Trial
Service
HTTP
MIRC
Storage
Field Center Sites
Required extensions:
1. image export in groups
2. CSV converter to metadata for NID
MIRC
Storage
Principal Investigator Site
Desirable extensions:
1. name spaces
2. cde viewing / query
3. extend author service