Related Experiment Videos
Automatic semantic interpretation of anatomic spatial relationships in clinical text
C A Bean1, T C Rindflesch, C A Sneiderman
1National Library of Medicine, Bethesda, MD 20894, USA.
Proceedings. AMIA Symposium
|February 3, 1999
Summary
This study developed semantic rules for natural language processing to understand spatial relationships in medical reports. The system shows promise for analyzing clinical text, particularly coronary angiography reports.
Area of Science:
- Natural Language Processing
- Medical Informatics
- Clinical Text Analysis
Background:
- Understanding spatial relationships in clinical text is crucial for accurate medical data interpretation.
- Current methods for analyzing complex anatomical relationships in reports are limited.
Purpose of the Study:
- To develop and implement semantic interpretation rules for linking syntax and semantics of locative relationships among anatomic entities.
- To assess the system's ability to identify and characterize physico-spatial relationships in coronary angiography reports.
Main Methods:
- Developed a natural language processing system with semantic interpretation rules.
- Conducted two experiments to evaluate the system's performance on coronary angiography reports.
Main Results:
- Branching relationships were the most frequent (75%), followed by PATH (20%) and PART/WHOLE relationships.
- Achieved overall recall of 0.78 and precision of 0.67.
Conclusions:
- The developed approach is viable for semantic processing of clinical text.
- The system effectively identifies and characterizes physico-spatial relationships in medical reports.