Q Huang, G Jhanwar*, J Yang, H Pham, L Cervino, J Deasy, M Zarepisheh, L Hong, Memorial Sloan Kettering Cancer Center, New York, NY

• 
  G Jhanwar

MO-C930-IePD-F4-4 (Monday, 7/11/2022) 9:30 AM - 10:00 AM [Eastern Time (GMT-4)]

Exhibit Hall | Forum 4

Purpose: To present our clinical experience with a fully automated home-grown treatment plan optimization system integrated with our treatment planning system (TPS).

Methods: An automatic treatment planning workflow was created using in-house-developed automated optimization system named Expedited Constrained Hierarchical Optimization (ECHO) which is integrated into the TPS using API scripting. ECHO applies advanced optimization tools such as hierarchical constrained optimization, convex approximations, and Lagrangian methods, to produce Pareto optimal plans. Clinical criteria, including maximum and mean doses, and dose volume metrics, were separated into 2 categories: limits and guidelines. Limits were strictly enforced by ECHO, while guidelines were optimized sequentially as much as possible. For implementing ECHO in clinic, we established 2 steps for each disease site: training and validation. Firstly, a template file containing department clinical criteria was created and optimization parameters were fine-tuned using training set of just a few patients (varies from 6 to 29). For validation, a larger number of auto-plans were generated applying the same parameters retrospectively to compare with manual plans. Dose metrics were evaluated for plans in terms of tumor coverage and normal tissue sparing. After clinical implementation, we assessed the performance of ECHO.

Results: Since 2017, ECHO has been used to generate 5697 treatment plans for 5000 patients across 4 disease sites: paraspinal SBRT (3673), oligometastasis SBRT (1626), prostate (371), and lung (27). PTV sizes ranged from 2cc to 1854cc. The latest version of ECHO achieves total planning time of 27 min (8 min-106 min). For all treated ECHO plans, 79% of plans were accepted clinically after single run of the program.

Conclusion: We developed an automated treatment planning system, provided as a plug-in in the TPS, for different disease site. ECHO improved consistency of plan quality for treatment planning and enabled expedited treatment including same day SBRT treatment in our clinic.

Funding Support, Disclosures, and Conflict of Interest: This work was partially supported by the MSK Cancer Center Support Grant/Core Grant (NIH P30 CA008748)

Keywords

Treatment Planning, Optimization

Taxonomy

TH- External Beam- Photons: Treatment planning using machine learning/Knowledge Based Planning/automation

Contact Email