Establishing a knowledge-based planning model for left-sided breast cancer patients receiving hypofractionated...

We aimed to validate and evaluate a RapidPlan model for left-sided breast cancer patients receiving hypofractionated PM-RNI. This study included 89 patients with left-sided breast cancer who were enrolled in a clinical trial and received PM-RNI using IMRT. A RapidPlan KBP model was developed using 40 clinical IMRT plans treated at Institution A. The performance of the KBP model was assessed through internal validation and external validations at Institution A (n = 40) and Institution B (n = 9) by comparing the dosimetric indices and treatment efficiency between the initial plans and the re-optimized KBP plans. The KBP model's benefit for planners with different experience levels was evaluated by comparing the quality of non-KBP and KBP plans created by beginner, junior, and senior planners for 10 cases. With comparable target coverage, the KBP plans showed superior normal tissue sparing than the initial plans. On average, the mean heart dose was significantly reduced in the internal validation (p < 0.001), external validation at Institution A (p < 0.001), and increased in the validation at Institution B (but not significantly). In terms of dose conformality, KBP plans achieved a notable improvement in CI_50% during the external validation at Institution A (p = 0.006). Additionally, treatment efficiency was improved by the KBP model for all validation datasets, with a reduction of 2.8% to 8.8% in treatment time. The beginner planners benefit most from the KBP model, which improved plan quality by reducing the mean heart dose by 13.67 cGy (p = 0.020). The RapidPlan model for left-sided breast cancer receiving hypofractionated PM-RNI effectively spares organs and increases IMRT treatment efficiency. We also demonstrated that RapidPlan helps transfer planning skills from senior planners to beginner planners, thus improving the consistency of complex IMRT plans.