Predictive Tracking for Respiratory Induced Motion Compensation in Adaptive Radiotherapy

Abstract

This paper considers a compensation strategy for respiratory-induced tumour motion for adaptive radiotherapy using a controlled patient support system (PSS). A model predictive control (MPC) scheme for the PSS to track the tumour motion is proposed together with two methods for predicting tumour motion, including a Kalman filter and neural networks. Simulation results using a clinical data set consisting of 27 traces of respiratory motion show the potential of the proposed control strategy.