Medical Image Database Semantic Search via Ontology Annotations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for accessing medical image databases lack the ability to efficiently retrieve relevant medical images using meaningful medically relevant terms for diagnosis and treatment, limiting their effectiveness in clinical decision-making.
Innovation Solution
A system and method that enable semantic matches between medical annotations and medically relevant terms, allowing for the automatic generation of anatomical and diagnostic annotations for image data sets, enabling searches based on anatomically and diagnostically relevant terms through the use of medical ontologies and model-based segmentation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional image database retrieval systems are used, then basic image storage and access is achieved, but the ability to retrieve relevant medical images using medically relevant terms for diagnosis and treatment is lacking
Solution Approach 1:
The patent introduces medical ontologies as an intermediary layer between the query system and the image database. These ontologies contain structured medical knowledge including anatomical hierarchies, diagnostic terms, and relationships between medical concepts. The system uses these ontologies to map user queries to standardized medical annotations, enabling semantically accurate retrieval of medical images based on diagnostically relevant terms rather than simple keyword matching
Solution Approach 2:
The system transforms the search parameter from simple keywords to structured medical annotations derived from ontologies. By changing the parameter representation from unstructured text to standardized medical concepts with hierarchical relationships, the system enables more precise and meaningful search capabilities while maintaining ease of use through natural language query interfaces
2Quantity of substance
If anatomical shape information is used for retrieval, then reduced data amount is achieved, but the system lacks ability to search using high-level medical information meaningful for diagnosis and treatment
Solution Approach 1:
The patent segments the medical image data into anatomically defined subsets using model-based segmentation. By dividing the complete image data into distinct anatomical regions (e.g., organs, tissues, structures) with specific annotations, the system enables retrieval of only the relevant anatomical subsets needed for diagnosis, reducing data transfer and processing while preserving all necessary medical information
Solution Approach 2:
The system performs preliminary annotation of anatomical structures and generation of medical metadata before the retrieval process. Medical annotations including anatomical identifiers, diagnostic information, and ontology-based tags are pre-computed and attached to image subsets, enabling efficient query processing without requiring complex analysis during retrieval, thus reducing operational data requirements while maintaining information completeness
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to a system for accessing a database comprising a plurality of image data sets. The system comprises an acquisition unit for acquiring a query for searching the database for an image data set or an image data subset comprised in an image data set, the query comprising at least one medically relevant term defining search criteria; a determining unit for determining the image data set or the image data subset comprised in the image data set, based on the strength of semantic matches between the at least one medically relevant term and (a) corresponding medical annotation(s) describing the image data set; and a retrieving unit for retrieving the determined image data set or image data subset from the database. By enabling semantic matches between medical annotations describing the image data set and the medically relevant term comprised in the query, this invention enables searching for medical images with high-level medical information that is meaningful for medical diagnosis and treatments.