Spatial-symbolic Query Engine in Anatomy

Area of Science:

• Medical Informatics
• Anatomy
• Semantic Web Technologies

Background:

• Traditional methods for anatomical queries rely on textbooks and online searches.
• A need exists for more dynamic and interactive anatomical information retrieval systems.

Purpose of the Study:

• To develop a semantic web service and spatial query processor for anatomical questions.
• To enable graphical posing of spatial and symbolic anatomical queries.

Main Methods:

• Anatomy experts defined spatial relations (e.g., anterior).
• A 3-D labeled thoracic image volume was queried for anatomical structures and relations.
• A web interface integrated the Foundational Model of Anatomy ontology (FMA) with spatial query results.

Main Results:

• The system achieved an F-measure of 0.90 (rater 1) and 0.56 (rater 2) when evaluated against expert opinions.
• High observed agreement (99%) and Cohen's kappa (0.51) were noted.
• Discrepancies were primarily attributed to dataset labeling issues, not the query tool itself.

Conclusions:

• The system serves as a functional end-user application and a framework for developing new applications.
• It offers practical utility for accessing and navigating integrated spatial and symbolic anatomical data.
• Potential applications include visualizing anatomical data in 3-D scenes.