Related Experiment Video
Updated: Sep 15, 2026
Radiation Planning Assistant - A Web-based Tool to Support High-quality Radiotherapy in Clinics with Limited Resources
Published on: October 6, 2023
A Generalized Software Framework for Consolidation of Radiation Therapy Planning and Delivery Data From Diverse Data
Yasin Abdulkadir1, Justin Hink1, Peter Boyle1
1Department of Radiation Oncology, University of California, Los Angeles, California.
Purpose:
Large-scale radiation therapy data sets drive predictive modeling, automated segmentation and planning, and personalized treatment, yet remain fragmented because treatment planning (TPS) and record-and-verify (R&V) systems differ, Digital Imaging and Communications in Medicine (DICOM) implementations are inconsistent, and automated linkage tools are lacking. We developed a generalizable framework that automatically reconstructs complete planning and delivery data sets across diverse clinical environments with minimal manual effort.
Methods And Materials:
We designed and implemented a software framework capable of automating the collection and integration of radiation therapy data from multiple institutions and TPS/R&V combinations. The system begins with Radiation Therapy Treatment Records and recursively traces unidirectional DICOM references to retrieve linked radiation therapy treatment plans, doses, structure sets, planning images, image registrations, and associated diagnostic images. Core components of the framework include automated DICOM queries, secure data transfer, integrity verification, linkage mapping, and detailed logging. To support diverse environments, we developed custom modules for non-DICOM-compliant systems, file format conversions, and robust error handling.
Results:
The framework was deployed across 4 institutions using 6 different combinations of TPS and R&V systems. In a focused 2-clinic implementation spanning 11 years of retrospective data, the system successfully processed and integrated data from 6164 patients and 13,871 radiation therapy plans. The pipeline achieved a 99.76% success rate in identifying and linking complete treatment data sets, with an average processing time of 18 minutes per patient, demonstrating its efficiency and scalability in real-world conditions.
Conclusions:
This automated framework provides a scalable and reliable solution for large-scale aggregation of radiation therapy data. It is compatible with heterogeneous clinical systems, including those lacking DICOM Query/Retrieve support, and overcomes key technical barriers to data integration. By enabling the creation of comprehensive, high-quality data sets, the framework supports advanced research and contributes to the improvement of clinical care in radiation oncology.
