Report of the Medical Image De-Identification (MIDI) Task Group -- Best Practices and Recommendations

Area of Science:

• Medical Imaging Informatics
• Data Privacy
• Health Informatics

Background:

• Medical images and associated data contain sensitive patient information.
• Public sharing of medical data is crucial for research and AI development.
• Existing de-identification methods require careful technical consideration for broad data sharing.

Purpose of the Study:

• To address the technical aspects of de-identifying medical images and related data.
• To ensure reduced re-identification risk for unrestricted public sharing.
• To focus on Digital Imaging and Communications in Medicine (DICOM) data and associated objects.

Main Methods:

• Review of technical de-identification strategies for various medical image formats.
• Emphasis on embedded data elements within DICOM files.
• Inclusion of image-like objects (Segmentations, Parametric Maps) and related non-image objects (RT Structure Sets, Plans).

Main Results:

• Technical guidelines for de-identifying medical images and associated data are presented.
• Methods ensure sufficient reduction of re-identification risk for public release.
• The report covers a wide range of medical imaging data, including DICOM and related objects.

Conclusions:

• Effective de-identification techniques are essential for enabling unrestricted public sharing of medical imaging data.
• This report provides a technical framework for de-identifying diverse medical imaging data types.
• Adherence to these technical aspects facilitates ethical and legal data sharing for research purposes.