The deployment of ProKnow for cloud-based clinical research in radiotherapy

Area of Science:

• Medical Physics
• Radiotherapy Informatics
• Data Management in Healthcare

Background:

• Radiation oncology data management requires robust tools for analysis and comparison.
• Current de-identification processes can hinder longitudinal data analysis.
• The Elekta ProKnow DS system offers a cloud-based solution for Picture Archiving and Communications in Radiotherapy (RT-PACS).

Purpose of the Study:

• To present a comprehensive implementation of the Elekta ProKnow DS system within a clinical department.
• To detail and evaluate two distinct de-identification workflows for DICOM data.
• To establish a foundation for advanced data analysis, including longitudinal outcomes and feasibility studies.

Main Methods:

• Implementation of the Elekta ProKnow DS system, including its user interface and Application Programming Interface (API).
• Development and comparison of two de-identification workflows: ProKnow Dicom Agent (PDA) and a trusted third-party service.
• Utilizing Python scripts for data extraction, metric calculation, and data upload via the API.
• Collaboration with MeDICLMU for an automated de-identification process to enable merging of clinical and de-identified data.
• Retrospective dosimetric feasibility study for glioblastoma radiotherapy and a dummy run for a prospective trial.

Main Results:

• Successful implementation of ProKnow DS, enabling plan analysis and comparison with an integrated Dose-Volume Histogram (DVH) engine.
• Demonstrated two de-identification workflows, highlighting a drawback in matching IDs for subsequent clinical data uploads.
• Established an automated de-identification process via a trusted third party to facilitate longitudinal data merging.
• Conducted a retrospective feasibility study and a multi-institute dummy run, validating protocol compliance and data analysis capabilities.
• RT-structures were automatically downloaded via API, and Dice Score and Hausdorff Distance were calculated and set as metrics in ProKnow.

Conclusions:

• The implemented ProKnow DS system provides a versatile platform for radiation oncology data management and analysis.
• Improved de-identification strategies, particularly through trusted third-party services, are crucial for longitudinal outcomes analysis.
• The system's API facilitates automated data processing and metric calculation, supporting clinical trial preparation and retrospective studies.
• Integration with local databases enhances the potential for comprehensive patient data analysis over time.