Modeling and measurement of the variations of CT number distributions for mobile targets in cone‐beam computed tomographic imaging

The purpose of this study was to investigate quantitatively by measurement and modeling the variations in CT number distributions of mobile targets in cone-beam CT (CBCT) imaging. CBCT images were acquired for three targets manufactured from homogenous water-equivalent gel that was inserted into a commercial mobile thorax phantom. The phantom moved with a controlled cyclic motion in one-dimension along the superior-inferior direction to simulate patient respiratory motion. Profiles of the CT number distributions of the static and mobile targets were obtained using CBCT images. A mathematical model was developed that predicted the variations in CT number distributions and their dependence on the motion parameters of targets moving in one-dimension using CBCT imaging. The measured CT number distributions for the mobile targets varied considerably, depending on the motion parameters. The extension of the CT number distribution increased linearly with motion amplitude where maximum target elongation reached twice the motion amplitude. The CT number levels of the mobile targets were smeared over a longer distribution; for example, the CT number level for the 20 mm target dropped by nearly 30% at motion amplitude (A) equal to 20 mm in comparison with the CT number distribution of stationary targets. Frequency of motion played an important role in spatial and level variations of the CT number distributions. For example, the level of the CT number profile for the medium target (20 mm) decreased evenly by nearly 50% at A = 20 mm with high motion frequencies. Motion phase did not affect the CT number distributions for prolonged projection acquisition that included several respiratory cycles. The mathematical model of the CT number distributions of mobile targets in CBCT reproduced well the measured CT number distributions and predicted their dependence on the target size and phantom motion parameters such as speed, amplitude, frequency, and phase. The CT number distributions varied considerably with motion in CBCT. A motion model of CT number distribution for mobile targets has been developed in this work that predicted well the variations in the measured CT number profiles and their dependence on motion parameters. The model corrected the CT number distribution retrospective to CT image reconstruction where it used a first-order linear relationship between the number of projections collected in the imaging window of a mobile voxel to obtain the cumulative CT number. This model provides quantitative characterization of motion artifacts on CT number distributions in CBCT that is useful to determine the validity of CT numbers and the accuracy of localization and volume measurement of tumors in diagnostic imaging and interventional applications, such as radiotherapy.