Back to Projects List
Tool to anonymize a dataset of medical images.
- Hina Shah (UNC Chapel Hill)
- Juan Carolos Prieto (UNC Chapel Hill)
- Fryderyk Kögl (BWH, TUM)
The very first step to make any medical data available to research community is it’s anonymization. While there are ways to anonymize a single DICOM/non-dicom image in 3D Slicer, there’s no module to do this for a full dataset.
The proposed tool will:
- Anonymize a dataset of images by deleting any identifiable metadata information
- Have options to rename the files using either UUID or custom name.
- Create a crosswalk to get the correspondence between anonymized and original files
- Work as a standalone app or a slicer extension
- Objective A. Write tests
- Objective B. Push the extension to Slicer Extension Index
- Objective C. Find out what other features/enhancements can be added to this extension
Approach and Plan
- Identify existing anonymization pipelines for DICOM
- Modify code to make the extension be available as an extension (not a standalone app), and push it to Slicer Extension Index
- Within the community try to find out what other features would be useful to add to the extension.
Progress and Next Steps
- Worked on a couple of issues for the extension
- The extension has been pushed to the Slicer Extension Index.
- Had a productive discussion with a few folks in the community to understand what are the existing tools/conformances for DICOM anonymization - this needs more introspection and research on our part, and deciding how we want to proceed - especially for the dental research data sharing purposes.
- Will add a few suggested features, for example: letting users chose which metadata to anonymize.
Background and References
- Source code in Github repository