Related Experiment Video
Updated: May 12, 2026
05:18
Radiation Planning Assistant - A Web-based Tool to Support High-quality Radiotherapy in Clinics with Limited Resources
Published on: October 6, 2023
Quality assurance incorporating artificial intelligence-generated reference contours in a phase II radiotherapy trial
Wonhyeong Lee1,2, Yeon-Joo Kim1, Jin Hee Kim3
1Department of Radiation Oncology, National Cancer Center, Goyang, the Republic of Korea.
Physics and Imaging in Radiation Oncology
|May 11, 2026
Summary
This study standardized radiotherapy planning for locally advanced breast cancer by using artificial intelligence, significantly reducing variations between institutions for improved treatment consistency.
Area of Science:
- Oncology
- Radiotherapy
- Medical Imaging
Background:
- Prospective phase II trial for locally advanced breast cancer (RTaNAC) involves tailored radiotherapy (RT) post-neoadjuvant chemotherapy (NAC) and surgery.
- Inter-institutional variations in RT dose distributions can negatively impact clinical outcomes.
- A dummy run quality assurance was performed to establish a standardized RT plan protocol.
Purpose of the Study:
- To develop and validate a standardized radiotherapy planning protocol for a multi-institutional trial.
- To minimize inter-institutional variations in dose distribution for tailored RT in locally advanced breast cancer.
- To assess the impact of artificial intelligence (AI) and defined target volumes on RT plan consistency.
Main Methods:
- Computed tomography (CT) images from three clinical scenarios were used, including different lymph node (LN) boost levels.
- Seven institutions developed RT plans in two steps: initial institutional policies and then using AI auto-contoured structures and LN boost information.
- Dose-volume histograms for breast and regional nodal areas were analyzed to compare variations before and after protocol implementation.
Main Results:
- Inter-institutional variations in RT planning significantly improved after implementing the standardized protocol with AI assistance.
- Dose coverage for axillary level I LN improved from 57.5% to 97.5% (p=0.075) and supraclavicular LN from 75.6% to 88.8% (p=0.046) in Scenario 1.
- AI auto-contoured structures and LN boost target volume information effectively mitigated dose/volume metric variations among institutions.
Conclusions:
- Standardized RT planning, incorporating AI, successfully reduced inter-institutional variations in locally advanced breast cancer treatment.
- The dummy run established a consensus RT plan protocol, paving the way for multi-institutional expansion of the RTaNAC trial.
- This approach enhances treatment consistency and supports the reliable execution of complex radiotherapy protocols across multiple centers.
