3D Organ Motion Estimation for Radiotherapy through Material Point Method-Driven Extrapolation of Multiple Orthogonal Slices.

This study proposes a novel method for estimating three-dimensional organ motion during radiotherapy using real-time MRI data. The approach combines observations from axial, coronal, and sagittal cross-sectional images with vertical extrapolation and a material point method simulation to generate comprehensive three-dimensional displacement fields. Unlike existing methods that rely primarily on geometric deformation within observed slices, our method assists estimation of motion in all directions, including regions far from the observed planes. The effectiveness of the proposed approach was validated through both phantom experiments and human participant trials. The results suggest improvements in accuracy compared to conventional methods, achieving Dice scores of up to 0.748 in human trials and a minimum estimated error of 5.18 mm, a 22.6% reduction, approaching clinically acceptable thresholds for radiotherapy. The method is particularly promising for organs with complex non-periodic motion patterns, such as the pancreas, where accurate motion tracking is critical for effective treatment delivery.