Evaluation of Rapidplan Knowledge-Based Planning Models for VMAT Chestwall and Regional Lymph Node Irradiation
Abstract
Purpose
VMAT post-mastectomy and regional lymph node irradiation is a highly complex technique that requires multiple planning iterations to achieve adequate coverage while minimizing OAR doses. RapidPlan, a Varian knowledge-based modeling program, has improved planning efficiency for many sites. Current models are focused on whole or partial breast irradiation with few published models evaluating post-mastectomy and nodal irradiation and little evaluation on the impact of fractionation. The goal is to develop a comprehensive RapidPlan model for VMAT chestwall and nodal irradiation that encompasses varying unilaterality and fractionation schemes aiming to improve planning efficiency.
Methods
We developed an individual RapidPlan model for the 5000cGy/25Fx scheme trained on clinically acceptable VMAT Chestwall and nodal plans to assess feasibility of this planning technique. RapidPlan-generated plans, using only 1 iteration, were compared with manual plans assessing target coverage, OAR, overall plan metrics and in-house clinical goals using the Wilcoxon Ranked Sum metric.
Results
The 5000cGy/25Fx RapidPlan model improved planning efficiency for 10 VMAT Chestwall cases, with an average optimization time of 280.5 seconds. Target coverage was statistically indistinguishable between RapidPlan and manual plans, with mean PTV sum V95% of 90.64% versus 89.55% (p=0.705), respectively. Achieved OAR doses were similar including mean heart dose (4.86Gy vs. 4.02Gy, p=0.174), ipsilateral lung V20Gy (30.72% vs. 28.12%, p=0.406), and maximum spinal cord dose (23.36Gy vs. 25.07Gy, p=0.406). These results indicate clinically acceptable plan quality can be obtained with significant planning time savings. Future work will expand training and testing datasets to further validate model robustness.
Conclusion
The initial RapidPlan model for 5000cGy/25Fx produced plans with target coverage and effective OAR sparing comparable to manual plans. This technique is expected to significantly improve planning efficiency. Expanding the model to multiple fractionation schemes and unilaterality to support a wider range of VMAT Chestwall cases is currently under investigation.