2139 - Model-Training Knowledge-Based Planning for Hippocampal-Sparing Whole Brain Radiotherapy: Efficiency and Clinical Outcomes
Presenter(s)
C. F. Zeng, H. L. Hsieh, P. H. Lin, H. Y. Lin, S. K. Hung, and M. F. Lee; Departments of Radiation Oncology, Dalin Tzu Chi Hospital, Buddhist Tzu Chi Medical Foundation, Chiayi, Taiwan
Purpose/Objective(s): Whole brain radiotherapy (WBRT) remains as a possible component of standard treatments for patients with brain metastases. Recent clinical evidence has underscored the cognitive benefits of hippocampal-sparing during WBRT, as outlined in RTOG 0933 guidelines. However, hippocampal avoidance presents substantial challenges in planning efficiency and technical precision. The present study introduces a Knowledge-Based Planning (KBP) model aimed at significantly reducing planning time while maintaining high-quality dosimetry outcomes.
Materials/Methods: A Hippocampal-Sparing Whole Brain radiotherapy (HSWBRT) model was developed using a recursive training process involving 20 clinical cases and validated with an additional four cases. The AI automation tool was employed to delineate critical organs-at-risk (OARs) efficiently. Dosimetry parameters adhered to the RTOG 0933 guideline, with a prescription dose of 30 Gy. Key planning metrics included PTV D2% = 37.5 Gy, V100 = 90%, and hippocampal constraints of D100% = 9 Gy and maximum dose = 17 Gy. Manual planning was used as a comparator, and patient data were collected from May 2018 to May 2024.
Results: All validation cases achieved compliance with the RTOG 0933 criteria. The HSWBRT model demonstrated superior efficiency, with an average planning time of 86.5 minutes compared with 448.0 minutes for manual planning, representing an 80.7% reduction in time. Dosimetry quality metrics for target coverage, dose conformity, and homogeneity were equivalent or superior in the KBP model compared with manually generated plans.
Conclusion: The Knowledge-Based Planning model for HSWBRT significantly enhanced planning efficiency without compromising plan quality. The model offers a streamlined solution for achieving high-quality hippocampal avoidance in clinical practice, aligning with the RTOG 0933 guideline. This advancement underscores the potential of knowledge-driven techniques to address technical challenges in modern radiotherapy planning.