LinkedDicom: Indexing DICOM Metadata Using Semantic Web Technologies

Area of Science:

• Medical Informatics
• Bioinformatics
• Semantic Web Technologies

Background:

• Digital Imaging and Communications in Medicine (DICOM) standard offers rich metadata, but archiving systems store limited subsets.
• Integrating imaging metadata with clinical and experimental data is crucial for research.
• Existing methods for DICOM metadata management can be resource-intensive.

Purpose of the Study:

• To develop a lightweight Python package, LinkedDicom, for converting DICOM metadata into Resource Description Framework (RDF) triples.
• To enable querying of DICOM metadata using Semantic Web technologies.
• To provide a resource-efficient and flexible solution for research data integration.

Main Methods:

• Developed LinkedDicom, a Python package for DICOM to RDF conversion.
• Utilized Semantic Web technologies for metadata querying.
• Evaluated LinkedDicom on a public head and neck cancer dataset.
• Compared query performance using SPARQL endpoints and per-patient file storage.

Main Results:

• LinkedDicom demonstrated efficient and scalable representation of imaging metadata.
• On-disk storage with selective in-memory analysis proved a flexible alternative to resource-intensive databases.
• Practical trade-offs between database-driven and on-demand querying were identified.

Conclusions:

• LinkedDicom enables FAIR-compliant representation of imaging metadata for research.
• The package offers a practical and resource-efficient approach to DICOM metadata management.
• LinkedDicom facilitates the integration of imaging data with other research datasets.