De-Identification of Radiomics Data Retaining Longitudinal Temporal Information

Area of Science:

• Medical Physics
• Radiotherapy Oncology
• Data Privacy

Background:

• Radiation oncology generates sensitive clinical and imaging data, including RTSTRUCT, RTPLAN, and RTDOSE.
• Longitudinal data, such as cone beam CT scans, require secure archival alongside initial records.
• Existing de-identification methods may not adequately address the complexity of radiation oncology data.

Purpose of the Study:

• To develop and present a standalone de-identification system for radiation oncology patient data.
• To ensure the privacy of clinical and annotated imaging data while maintaining data integrity.
• To facilitate potential re-identification of data when necessary through encrypted references.

Main Methods:

• The system operates in a standalone mode for efficient de-identification.
• It processes clinical data (diagnosis, staging, outcome, treatment) and imaging data (diagnostic, planning, verification).
• Original data identities are stored in an encrypted format for secure referencing.

Main Results:

• The system successfully de-identifies radiation oncology data, including RTSTRUCT, RTPLAN, and RTDOSE files.
• Archival of longitudinal verification images and associated clinical data is preserved.
• Encrypted references allow for secure and controlled re-identification.

Conclusions:

• The proposed system offers a robust solution for de-identifying radiation oncology data.
• It balances patient privacy with the potential need for data re-identification.
• This approach supports secure data management in radiation oncology research and practice.