Routine Clinical Use of Clinical-Criteria–Driven Automated IMRT and VMAT Planning
Abstract
Purpose
Despite extensive research on automated treatment planning, manual trial-and-error optimization remains common in clinical practice. Knowledge-based and AI-driven approaches show promise but often lack robustness to evolving clinical protocols due to the need for frequent model re-training. These methods may also be inadequate for re-irradiation cases, where clinical criteria and achievable doses depend on prior radiation exposure. We present the development and clinical translation of ECHO (Expedited-Constrained-Hierarchical-Optimization), an in-house automated planning system designed to address these limitations. ECHO has been clinically deployed for IMRT and recently extended to VMAT planning.
Methods
ECHO employs a hierarchical constrained-optimization framework in which patient-specific clinical criteria explicitly guide the planning process. Essential clinical requirements are enforced as hard-constraints, ensuring mandatory dose limits are satisfied. Within this feasible region, a two-stage optimization is performed to maximize target coverage and normal-tissue sparing. Using the Eclipse API, ECHO accepts contours and clinical-criteria as inputs and generates optimized fluence for IMRT and leaf-trajectories for VMAT. Optimization is performed externally, bypassing the TPS optimization engine, while the TPS dose-calculation engine is used for final dose computation and validation. Prior to site-specific rollout, retrospective validation studies were conducted by comparing ECHO-generated plans with manually delivered clinical plans from randomly selected historical cohorts, with sample sizes tailored to treatment complexity (e.g., 75 paraspinal and 20 lung SBRT cases).
Results
ECHO-IMRT has been used clinically in over 10,000 treatments across multiple disease sites, including more than 2,000 complex paraspinal re-irradiation cases. Clinical deployment reduced planning time by approximately 50% while maintaining or improving plan quality. Extension to VMAT demonstrated comparable or improved target coverage and similar organ-at-risk sparing relative to manual lung VMAT SBRT plans in a 20-case retrospective evaluation.
Conclusion
ECHO enables consistent, high-quality IMRT and VMAT planning with substantial efficiency gains and robust performance for complex and re-irradiation scenarios.