DICOM Ontology
(DO) UpdateBuilding a cathedral, one
grain of sand at a time
March, 2009
Daniel Rubin M.D., M.S.Assistant Professor of Radiology
Stanford University
David S. Channin, MDAssociate Professor of Radiology
Northwestern University
Curtis P. Langlotz, MD, PhDAssociate Professor of Radiology
University of Pennsylvania
Charles E. Kahn, Jr., MD, MSProfessor of Radiology
Medical College of Wisconsin
Presenters
DICOM
• DICOM is the global standard for medical imaging• Comprises 20+ years of combined knowledge
from engineers and radiologists from academia and industry
• 16 Parts; Over 125 Supplements; 3583 PAGES!• Subsets of DICOM used extensively in many
caBIG Imaging Workspace projects• Current DICOM standard not computationally-
accessible to reuse; limits interoperability
Needed by many,understood by few
• DICOM lacks a computable reference information model• Hinders interoperability in applications• Inconsistent use of controlled terminology or relations
• Specified in PDF documents; not computable• Duplication of effort in workspace projects• Potential conflicts among different models that represent
same imaging information
• DICOM is HUGE• time consuming to review and understand
DOverview: Death by Bureaucracy
• DO Phase I FY08• Planning contract
• DO Phase II FY09• Pilot Ontology development of 1 IOD (CT)
• DO Phase III FY10 ??• IODs 2,3,4,5,6
• DO Phase IV FY11 ??• IODs 7,8,9,10,11
• DO Phase V FY12 ??• IODs 11,12,13,14,15
• • • DO Phase XIV FY20 ??
• IODs 50,51,52,53,54• DO Phase XV-XX FY21-FY26 ?? (DSC,DLR,CPL,CK retire at age 65)
• New IODs since FY10
Overview of Phase II (FY09)
• Purpose: Develop a unified knowledge model based on the scope of radiology domain in DICOM
• Scope: DICOM is BIG; project is scoped initially to reviewing/modeling one IOD (CT)
• Goals: Develop prototype ontology and begin developing synergies with existing projects
Phase II Goals
• Produce initial skeletal structure of the DICOM ontology based on reviewing a single Information Object Definition (IOD)
• Create DICOM ontology containing the entities and relations needed to describe the IOD content.
• Define approach to generalize our approach to extend DICOM ontology to rest of the DICOM standard
Project Tasks
• Identify and review pertinent portions of DICOM: Focus on a single IOD
• Review E-R diagrams in DICOM: Identify elements incorporate into the oncology
• Reconcile DO for modeling inconsistencies in DICOM• Compile ontology components: Enumerate the key
terms comprising the ontology, and identify the content that should be modeled as classes and relations
• Build initial DICOM Ontology: Translate relevant portions of DICOM standard into a structured format that can be imported into an ontology authoring tool such as Protégé
DO Progress
• Reviewed CT IOD and DICOM standard• CT IOD• Relevant parts of DICOM extracted from part 6,
part 3 and some of part 16• Produced initial DICOM ontology
• Patient (Study, Series, and Image remain) • DICOM Modules and Macros• DICOM Data Dictionary
• Data Elements and Data Element Types
Raw CT IOD in DICOM
DICOM Ontology in Protege
Patient Module in DICOM
Patient Name in DICOM Ontology
Questions?