Room: AAPM ePoster Library
Purpose: To develop and test a closed-loop control framework for motion management in MRI-guided radiotherapy in which the cumulative deposited dose is continuously monitored and, if warranted, the radiotherapy plan is adjusted in real time to correct for dose discrepancies.
Methods: Starting with a base radiotherapy plan and a 4D-CT obtained prior to treatment, we calculate the dose volumes per phase and aperture of the base plan, which are used along with the cine MRI acquisition to dynamically estimate the cumulative dose. Additionally, the cine MRI is used to predict the intra-fraction motion and, in particular, the amount of time spent at each of a discrete set of anatomical states. The estimated cumulative dose and the motion prediction serve as feedback and feedforward signals, respectively, to dynamically re-optimize the remainder of the radiotherapy plan using a model-predictive control (MPC) strategy.
Results: We investigate the performance of the MPC framework by retrospectively applying it to a liver case treated on an MRI-guided radiotherapy system integrating three Co-60 heads and a 0.35T MRI scanner using a step-and-shoot delivery mode. The base plan consists of seven gantry angles with simultaneous delivery from the three heads at each angle. Prior to the delivery of each gantry angle, the MPC method re-optimizes aperture intensities for the remaining angles. We compare the clinical quality and delivery efficiency of the dynamically re-optimized plan against the base plan delivered under three scenarios, namely, static anatomy, free breathing, and respiration gated. Results validate that the dynamically re-optimized plan achieves a dose quality similar to the respiration-gated plan while improving the delivery efficiency by as much as 15%.
Conclusion: A closed-loop control framework was proposed for intra-fraction motion adaptation in MRI-guided radiotherapy. Compared to respiration-gated delivery, dynamic (re-)optimization of aperture intensities improves the delivery efficiency while maintaining the dose quality.
Funding Support, Disclosures, and Conflict of Interest: National Science Foundation, Award# 1662819