Fully Automated, Feedback-Driven VMAT Planning In Elekta Monaco for Localized Prostate Cancer
Abstract
Purpose
VMAT planning for prostate cancer in Elekta Monaco can be slow and variable due to differences in planner experience. Existing automation tools often depend on external prediction models or provide limited optimization support. We developed a fully automated workflow that operates entirely within Monaco and dynamically adjusts optimization objectives in response to real-time DVH feedback, aiming to deliver consistent, clinically acceptable plans without manual input.
Methods
Fifty consecutive patients treated with 60 Gy in 20 fractions were retrospectively analysed. For each case, an automated VMAT plan was generated using a scripted pipeline that launched planning from a standardized template and iteratively modified isoconstraints and optimization parameters during Phase 1 and Phase 2. Automated plans were compared with clinically delivered manual plans using Wilcoxon signed rank tests. Metrics included target coverage, conformity, OAR sparing, MU, and modulation complexity. Ten randomly selected plans (20%) underwent ArcCHECK QA using 3%/2 mm gamma criteria.
Results
All automated plans met institutional clinical constraints. Target coverage matched manual plans with no significant difference. Automated plans demonstrated slightly improved conformity and consistent reductions in low and mid-dose rectum metrics while maintaining similar high-dose values. For example, rectum V24Gy decreased by 9.76% (p < 0.001) and V32Gy by 5.85% (p < 0.001). Bladder results were mixed but clinically acceptable. MU and complexity scores increased modestly, but all QA tests passed with ≥99.8% gamma agreement, indicating no loss of deliverability. Average automated planning time was 38.7 minutes, substantially faster than typical manual workflows.
Conclusion
A fully autonomous, feedback-driven Monaco workflow can consistently generate prostate VMAT plans that match or improve manual plan quality, reduce planning time, and minimize interplanner variability. This demonstrates a practical path toward routine, TPS-integrated automated planning.