Research On Calibration Method In Cone-Beam XCT

Computed Tomography (CT) is a technique with the developments of physics and computer science. Medical CT and industrial CT are two branches in applications. There is no difference between the basic theories of these two branches. The development of CT imaging consists of parallel imaging, fan-beam imaging, and cone-beam imaging. So far cone-beam CT has become a hot topic not only in the medical CT research, but in the industrial nondestructive testing because of its advantages in fast scanning speed, high spatial resolution and better utilization of X-ray photons. The FDK algorithm is a classic cone-beam approximate reconstruction algorithm, which has been widely used in practice. Ideal imaging geometry is a basic condition in the application of the FDK algorithm. However, it is difficult to satisfy this condition in a practical cone-beam CT system. Reconstructed images will suffer from artifacts caused by the misaligned geometry of the CT system. Therefore, calibration of the CT system in advance is an important and necessary task. Misaligned parameters of the CT system will be taken into the reconstruction algorithm after calibration to improve image qualities in the case of the misaligned CT system.This dissertation, with a purpose on how to reconstruct cross-section slices with better qualities in a geometrical misaligned cone-beam CT system, consists of three tasks.The first task is to design a cone-beam CT simulation system. All kinds of geometrical misalignments of a cone-beam CT are simulated under the system. The effects of the misalignments of the detector, of the X-ray source, and of the turn-table on the reconstructed images are analyzed in detail as well. The high-resolution simulation system can satisfy the practical demands in the theoretical research.The second task is to propose a geometrical calibration method in such a misaligned cone-beam CT system. The projection of a calibration phantom consisting of four point objects is required at only one projection view. Then six parameters of the misaligned CT system will be calculated by solving a set of analytic equations. Previous assumption on any misaligned parameter will be avoided in the calibration method. On account of analytic solutions in the calibration method, the problem of local optimal solution in solving multi-variate functions will be avoidable as well. In addition, the calibration method will not introduce more mechanical error caused by collection of the phantom projections at multiple projection views since the phantom projection at only one projection view is sufficient and necessary for the calibration method. The final task of the dissertation is to propose a filtered back-projection reconstruction algorithm for misalignments correction based on the FDK algorithm after calibration. The improved reconstruction algorithm will use the six calibration parameters calculated in the calibration method to reduce the artifacts in the reconstructed images caused by geometrical misalignments of the CT system. The data of projection coordinates used in the improved algorithm can be computed in advance to improve the efficiency of reconstrucion in practical applications.