Presentation Description

No description available.


Presentation Transcript

Linking Ontologies with Three-Dimensional Models of Anatomy to Predict Physiological Effects of Penetrating Injuries: 

Linking Ontologies with Three-Dimensional Models of Anatomy to Predict Physiological Effects of Penetrating Injuries Daniel L. Rubin Yasser Bashir David Grossman Parvati Dev Mark A. Musen Stanford Medical Informatics Stanford University

Projectile Injury: 

Projectile Injury Penetrating trauma responsible for large proportion of civilian deaths; major cause of battlefield fatalities Survivability after projectile injury depends on rapid acquisition/interpretation of knowledge Need to know anatomic structures injured and extent of organ damage Need to know physiological consequences Need to make triage decision (immediate surgery, medi-vac, observe, etc.)


Objectives Predict consequences of projectile injuries Help triage patients at time of injury Provide on-scene decision support Computable model of knowledge of human anatomy (“holomer”) Predict anatomic structures injured by penetrating trauma Predict extent of organ damage (e.g., volume of injury) Use to enable decision support for triage Provide graphical display of anatomy, bullet trajectory, and tissue damage


Challenges Making knowledge explicit and computable Geometric knowledge: implicitly represented in 3-d models Anatomic/physiologic knowledge: usually in head of observer Difficult to automate reasoning Richness and complexity of this domain suggest using Ontologies for representation and computation

Ontologic approach to geometric models of anatomy: 

Ontologic approach to geometric models of anatomy Input: segmented CT data (pre-injury) Build an integrated 3-d model of anatomy: Represent 3-d geometry of anatomic structures Integrate tissue physical properties and biomechanics Simulate geometric effects of penetrating injury Geometric model links to knowledge source of anatomy (FMA ontology) (Use physiologic models to predict consequence of the injury) Predict and display organ injury

What Is An Ontology?: 

What Is An Ontology? Enumerates concepts, attributes of concepts, and relationships among concepts, thus defining a structure (“model”) for the application area Can be comprehended by people and processed by machines Provides a “domain of discourse” for characterizing some application area; a common vocabulary (shared understanding)


Protégé-2000 Ontology editor Modeling concepts, attributes, and relationships Tools Visualize ontologies and knowledge bases Storage Archive ontologies and knowledge bases in a variety of formats Java API Linking knowledge bases to other applications A world-wide community of active users


The Ontology in Protégé-2000 Ontology Slots

Knowledge Sources: 

Knowledge Sources Geometric Knowledge Geometry ontology constructed from computer graphics principles Specifies data structures used to represent geometric models Anatomic Knowledge Foundational Model of Anatomy (FMA) ontology Specifies anatomical entities and relationships (e.g., partonomies, continuities, adjacencies) Logical model as opposed to spatial

Ontology of Geometric Modeling: 

Ontology of Geometric Modeling

The Foundational Model of Anatomy (FMA) Ontology: 

The Foundational Model of Anatomy (FMA) Ontology


Hollow Viscus

Important Knowledge in FMA: 

Important Knowledge in FMA Catalog of organs; controlled vocabulary of names Organ parts and compositionality Adjacencies for organs and organ parts Connectivity Containment Arterial supply/venous drainage

FMA Browser: 

FMA Browser Part-of hierarchy Properties (“slots”) containing knowledge

Building Geometric Models: 

Building Geometric Models

Visible Human Raw Data: 

Visible Human Raw Data

Visible Human Segmented Data: 

Visible Human Segmented Data

Geometric Model Building: 

Geometric Model Building CT data is segmented into organ parts Segmented organ parts are stacked in 3D Construct mesh models (spatial objects) from segmented organ parts using VTK and ITK Biomechanical and other information added to create “abstract geometric objects” Tissue physical properties; density Abstract geometrical objects are organized into a hierarchy based on FMA: Heart  pericardium; LA; LV; RA; RV LA  LA wall; inter-atrial septum; chamber This model is extensible; can include other info

Object Hierarchy Derived from Anatomy Ontology: 

Object Hierarchy Derived from Anatomy Ontology Each node above is an abstract geometric object

Abstract geometric objects: 

Geometry Tissue Density Elasticity Surface area … Abstract geometric objects Rendering of Abstract Geometric Object Abstract Geometric Object

Predicting Organ Injury: 

Predicting Organ Injury Direct organ injury 1. Organs visible on CT: injury predicted by intersection with “cone of damage” 2. Organs not visible on CT: inferred from adjacencies in FMA Secondary organ injury 3. Use knowledge of arterial anatomy to infer downstream consequences to arterial damage

Injury caused by Cone of Damage: 

Injury caused by Cone of Damage We infer injured tissues using FMA: • names of injured tissues • knowledge of organ adjacencies


Bullet entry site Area of injured tissue surface Conical volume of injury


Conical region of tissue injury

Software Architecture Diagram: 

Software Architecture Diagram


Conclusions Benefits of integrating geometric models with ontologies Makes anatomic knowledge and relationships explicit and computer-accessible Useful for reasoning (e.g., organ adjacencies, impact of vascular injury, small structures) Benefits of integrating additional information in abstract geometric data objects (spatial objects) Biomechanical and other data needed for simulation Extensibility to accommodate future data


Thank you. Contact info:


FMA Segmented Data Additional Invisible Object Landmarks Biomechanics Injury FMA Invisible Objects Biomech Injury Geometric Models Pre-processing of link Between {geometry (image+mesh), biomechanics, etc} and FMA. How? Spatial objects (abstract geometry objects)

authorStream Live Help