Motion detection in cone-beam computed tomography incorporating a geometric calibration approach

This work proposed a motion detection method for cone-beam computed tomography (CBCT) that utilizes a calibration phantom of known geometry as the motion detector and an established geometric calibration protocol to provide the motion information. An initial numerical study regarding the consequences of motion and its correction was conducted with a Shepp-Logan and an XCAT phantom. Motion artifacts were induced by acquiring the projections in a simple saddle trajectory scan. Since the scanning trajectory is set, the magnitude of motion for each projection view is already known, the correction of motion can then be efficiently implemented. Motion correction was done prior to the backprojection process of the filtered backprojection (FBP) image reconstruction algorithm. Results showed that motion correction improved the image quality of the reconstructed images. For a known or unknown scanning trajectory, the geometric calibration method can define the geometric information of a scanning system. In the current work, projections of a calibration phantom of known geometry were acquired from a saddle trajectory scan, and geometric parameters for selected projection views were successfully computed from the projection matrix provided by the geometric calibration method. Further studies will involve an experimental investigation wherein a calibration phantom is attached to a randomly moving object and scanned in a circular trajectory. Utilizing the parameters extracted from the geometric calibration, an accurate description of the object motion can be used and adapted for motion correction.