In-Situ Validation of a Surrogate-based Lung Motion Model for the Long-term Capture of Cycle-To-Cycle Variations with 4DCT

Abstract

We propose a novel volumetric surrogate-based motion model (SMM) to address limitations of single cycle respiratory-correlated 4DCT in capturing breathing variations. SMMs are constructed based on the a priori correlation between an external surrogate and the internal motion observed during the planning CT acquisition. Our machine-learning based volumetric SMM exploits the internal-external correlation observed at the time of treatment delivery, minimizing the loss of accuracy resulting from commonly occurring changes in this correlation. We evaluated improvements in target position estimation from our SMM compared to 4DCT by analyzing 2,369 fluoroscopic (FL) images.