3731 - Maximizing Research Potential and Re-Use of Archival Radiotherapy Datasets
Presenter(s)
R. F. Thompson1, R. Kapoor2, W. C. Sleeman3, N. Johnson4, C. Madison4, J. A. Lynch5, N. G. Nickols6,7, E. Katsoulakis8, J. R. Palta9, and M. D. Kelly10; 1Department of Radiation Medicine, Oregon Health & Science University, Portland, OR, 2National Radiation Oncology Program, Veterans Healthcare Administration, Richmond, VA, 3Virginia Commonwealth University, Richmond, VA, 4VA Portland Healthcare System, Portland, OR, 5Department of Internal Medicine, University of Utah School of Medicine, Salt Lake City, UT, 6University of California Los Angeles, Department of Radiation Oncology, Los Angeles, CA, 7VA Greater Los Angeles Health System, Los Angeles, CA, 8Veterans Affairs, Tampa, FL, 9Department of Radiation Oncology, Virginia Commonwealth University, Richmond, VA, 10National Radiation Oncology Program, Veterans’ Healthcare Administration, Washington, DC
Purpose/Objective(s):
Radiation oncology treatment planning and delivery systems are predominantly designed as silos, centered around the care of individual patients and generally disconnected from the broader health record. This poses significant challenges for cohort or population scale research, particularly when trying to analyze the nuances and details of treatments. We sought to design, develop, and implement a platform-agnostic tool to extract clinically meaningful treatment details from DICOM-RT data at two distinct treatment facilities and merge the resulting dataset with the broader electronic health record to empower arbitrary queries for the first time.Materials/Methods:
We developed a python script which takes as input a DICOM-RT dataset and outputs a dictionary of extracted data elements including but not limited to structure-specific dose volume histogram data, individual beam-level treatment details, and verified delivered fraction data. We applied this to historical data from two treatment facilities and linked the resulting data with the broader electronic health record on an individual patient level.Results:
We demonstrate successful export of clinically meaningful treatment details from two real-world cohorts of patients treated between 2012-2024, representing well over 5,000 radiation courses. We confirmed the ability to arbitrarily query these cohorts based on both the intrinsic data export as well as its linkages to the broader electronic health record.Conclusion:
This is a proof-of-principle study demonstrating the ability to extract and integrate detailed radiotherapy data with the broader health record, as well as enable unprecedented arbitrary queries at population scale.