Click here to

Session: Machine Intelligence for Treatment Planning and Segmentation [Return to Session]

Performing Fully Automated Treatment Planning Using Meta-Optimization

C Huang*, Y Nomura, Y Yang, L Xing, Stanford University School of Medicine, Stanford, CA


WE-G-BRC-5 (Wednesday, 7/13/2022) 2:45 PM - 3:45 PM [Eastern Time (GMT-4)]

Ballroom C

Purpose: Radiation therapy treatment planning is a manual, iterative process that can be very time-consuming. Typically, planners perform repeated adjustments of hyperparameters until a clinically acceptable plan has been generated. Automating the treatment planning process has the potential to drastically reduce active planning times and streamline clinical workflow. For this purpose, we propose a meta-optimization framework called MetaPlanner (MP).

Methods: The proposed MP algorithm is an open-source framework that performs optimization of treatment planning hyperparameters using a two-loop optimization approach. In the outer loop, the algorithm uses a derivative-free method (i.e. parallel Nelder-Mead simplex search) to search for objective weight configurations that minimize a meta-scoring function. In the inner loop, traditional inverse planning optimization (i.e. fluence map optimization) is performed for a given configuration of objective weights. Meta-scoring of of treatment plans utilizes a tier list to rank planner preferences and mimic the clinical decision-making process. The source code for the MP algorithm is publicly available through github (

Results: The proposed MP method is evaluated on two clinical datasets (21 prostate cases and 6 head and neck cases). MP is applied to both IMRT and VMAT planning and compared to a baseline of manual VMAT plans. MP in both IMRT and VMAT scenarios has comparable or better performance than manual VMAT planning for all evaluated metrics (e.g. target coverage, dose conformity, dose homogeneity, OAR sparing, etc.).

Conclusion: The MP method is an open-source framework for automated treatment planning that requires no active planning and maintains or improves plan quality.

Funding Support, Disclosures, and Conflict of Interest: This research was supported by the National Institutes of Health (NIH) under Grants 1R01 CA176553, R01CA227713, and T32EB009653, as well as a faculty research award from Google Inc.


Optimization, Treatment Planning


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

Contact Email