Using ontologies linked with geometric models to reason about penetrating injuries

Area of Science:

• Medical imaging analysis
• Computational anatomy
• Artificial intelligence in medicine

Background:

• Assessing penetrating injuries requires integrating physical exams and CT scans with extensive medical knowledge.
• Current methods are knowledge-intensive and time-consuming, impacting triage and treatment decisions.
• Automating this process can significantly improve the speed and accuracy of injury assessment.

Purpose of the Study:

• To develop a methodology for automating the reasoning process in penetrating injury assessment.
• To combine subject-specific image data with canonical medical knowledge for injury analysis.
• To create intelligent computer applications for aiding medical practitioners in evaluating penetrating trauma.

Main Methods:

• Constructing a 3D geometric model from segmented medical images of the subject.
• Linking image regions to ontologies of anatomy and regional perfusion.
• Developing computer reasoning services ('problem solvers') to analyze the geometric model and deduce injury consequences.

Main Results:

• Demonstrated the ability to identify injured organs based on projectile trajectories.
• Successfully determined if vital structures, like coronary arteries, were compromised.
• Predicted the spread of injury following damage to critical anatomical components.

Conclusions:

• Validated the use of ontologies with medical images for computer-assisted injury reasoning.
• Showcased a novel approach for intelligent computer applications that reason with medical image data.
• Highlighted the potential value of this methodology in clinical practice for penetrating injury assessment.