Development and clinical implementation of a comprehensive multifractionation scheme HyperArc-based RapidPlan model...

Area of Science:

• Medical Physics
• Radiation Oncology
• Machine Learning

Background:

• Brain stereotactic radiosurgery/radiotherapy (SRS/SRT) planning involves a significant delay (1-2 weeks) from simulation to treatment.
• Automating and standardizing the treatment planning process is crucial for improving efficiency and patient care.

Purpose of the Study:

• To develop and validate a machine learning model for rapid, automated treatment planning in single-isocenter brain SRS/SRT.
• To mitigate the lengthy treatment planning time for single- and multi-lesion brain SRS/SRT.

Main Methods:

• A HyperArc-based RapidPlan (HARP) model was developed using 89 previously treated HyperArc SRS/SRT plans on a TrueBeam LINAC.
• The model was trained on 69 plans and tested on 20 plans, with subsequent clinical application for patient treatment.
• Plan quality was assessed by target coverage (PTV D95%, GTV D100%), organ-at-risk (OAR) sparing, and dosimetry checks.

Main Results:

• The HARP model generated clinically acceptable plans in under 20 minutes across three fractionation schemes.
• HARP plans demonstrated comparable PTV D95% and superior GTV D100% compared to original manual plans.
• Maximum OAR doses were within SRS/SRT criteria, and all plans achieved acceptable gamma pass rates and Monte Carlo checks.

Conclusions:

• The HARP model significantly reduces brain SRS/SRT treatment planning time, enabling rapid generation of high-quality, clinically acceptable plans.
• This automation standardizes patient care, optimizes clinical workflow, and enhances accessibility to advanced radiation therapy treatments.
• The HARP model effectively handles multiple fractionation schemes while ensuring OAR sparing and therapeutic dose delivery.