S1938
Physics - Dose prediction/calculation, optimisation and applications for photon and electron planning
ESTRO 2026
reduction in planning time (from full-day to approximately half-day) while maintaining or improving plan quality. ECHO algorithm was also adapted for VMAT delivery. A retrospective evaluation of 20 randomly selected lung VMAT SBRT cases demonstrated that ECHO produced plans with superior or comparable PTV/CTV coverage and comparable OAR sparing compared to manually generated plans (Figures 1 and 2).
Digital Poster Highlight 3474 Automated and flexible radiotherapy treatment planning with ECHO: development, clinical deployment, and extension to VMAT Masoud Zarepisheh, Gourav Jhanwar, Qijie Huang, Ying Zhou, Jeho Jeong, Jie Yang, Hai Pham, Pengpeng Zhang, Laura Cervino Arriba, Jean Moran, Joseph Deasy, Hong Linda Department of Medical Physics, Memorial Sloan Kettering Cancer Center, New York, USA Purpose/Objective: Automated treatment planning has been studied for more than a decade, yet manual, trial-and-error planning remains common in clinical practice. Knowledge-based planning and recent AI-based dose prediction systems have shown promise, but their adoption has been limited by challenges in adapting trained models to evolving clinical protocols and in handling complex or re-irradiation cases that fall outside the training distribution. We present the development and clinical translation of ECHO (Expedited Constrained Hierarchical Optimization), an in-house automated planning system designed to address these limitations. The system has been deployed clinically for IMRT and, more recently, VMAT planning. Material/Methods: ECHO adopts a constrained optimization framework in which essential clinical requirements are enforced as "hard constraints”, ensuring that mandatory dose limits are satisfied. Within the feasible space defined by these constraints, a hierarchical two-step optimization is performed to achieve desired planning goals as much as possible. The system accepts contours and clinical criteria as inputs and generates optimized fluence maps (IMRT) and leaf trajectories (VMAT). While ECHO completely bypasses the TPS optimization engine, it leverages the TPS dose engine to ensure accuracy and for a final dose validation. For clinical deployment, ECHO was integrated into Eclipse through scripting, enabling planners to launch ECHO directly following contouring and beam/arc setup. Prior to clinical rollout for each treatment site and prescription, retrospective validations were performed by comparing ECHO-generated plans to clinically delivered manual plans for randomly selected historical cases. Sample sizes were chosen based on site complexity (e.g., n=75 paraspinal, n=20 lung SBRT). Results: ECHO-IMRT has been used clinically in over 10,000 treatments across multiple disease sites (paraspinal, oligomets, prostate, lung), including more than 2,000 complex paraspinal re-irradiation cases. Clinical implementation resulted in approximately a 50%
Figure-1: Manual v.s. ECHO-generated VMAT plans for 20 SBRT lung patients
Figure 2: An example ECHO v.s. manual plan Conclusion: ECHO enables consistent, high-quality radiotherapy treatment planning with substantial reductions in planning time for both IMRT and VMAT. The constraint-based formulation of ECHO allows robust handling of complex and re-irradiation cases and supports efficient adaptation to patient-specific needs and evolving clinical protocols. References: Dursun et al. (PMB, 2023) DOI 10.1088/1361- 6560/ace09eZarepisheh et al. (Medical Physics, 2019) DOI: 10.1002/mp.13572Hong et al. (Advances in Radiation Oncology, 2019) DOI: 10.1016/j.adro.2019.11.0052 Keywords: automated treatment planning, IMRT, VMAT
Digital Poster 3527
Optimizing Skin Cancer Radiotherapy: Dosimetric Comparison of VMAT, Electron Therapy, and HDR- Brachytherapy Américo G. Alvarado-Valadez 1 , Ana C. Ahumada- Pámanes 1 , Esperanza S. Ruiz-Sandoval 2 , Karen González- Pérez 1 , Oscar Vidal-Gutierrez 3
Made with FlippingBook - Share PDF online