A Method to Estimate Mean Position, Motion Magnitude, Motion Correlation, and Trajectory of a Tumor From Cone-Beam CT Projections for Image-Guided Radiotherapy

Purpose To develop a probability-based method for estimating the mean position, motion magnitude, and trajectory of a tumor using cone-beam CT (CBCT) projections. Method and Materials CBCT acquisition was simulated for more than 80 hours of patient-measured trajectories for thoracic/abdominal tumors and prostate. The trajectories were divided into 60-second segments for which CBCT was simulated by projecting the tumor position onto a rotating imager. Tumor (surrogate) visibility on all projections was assumed. The mean and standard deviation of the tumor position and motion correlation along the three axes were determined with maximum likelihood estimation based on the projection data, assuming a Gaussian spatial distribution. The unknown position component along the imager axis was approximated by its expectation value, determined by the Gaussian distribution. Transformation of the resulting three-dimensional position to patient coordinates provided the estimated trajectory. Two trajectories were experimentally investigated by CBCT acquisition of a phantom. Results The root-mean-square error of the estimated mean position was 0.05 mm. The root-mean-square error of the trajectories was