Research On Several Key Issues In Cone-beam X-ray Imaging Technique

For great advantages such as high longitudinal resolution, rapid scanning speed,high x-ray utilization, cone-beam x-ray imaging has been widely utilized in vivoanimal imaging and the woman breast diagnosis. It has become the most activeadvanced research in medical imaging field. However, how to improve the cone-beamx-ray imaging quality has significant academic value and application value.Corresponding solutions to several key factors that affect the cone-beam x-rayimaging quality are investigated in this study. The paper is mainly composed by threeparts.Firstly, according to the ray tracing theory and the x-ray attenuation law, asimulation platform of cone-beam x-ray imaging based on Shepp-Logan model wasdeveloped to verify and evaluate the cone-beam reconstruction algorithm. On thisbasis, aiming at the significance of the small animals in biomedical anatomy imagingresearch, an enclosed cone-beam micro-CT system was built up with digital specimenradiography system, and then experiment was conducted with the self-made animalphantom to evaluate the imaging system. Finaly, to investigate the small animalimaging, the vivo mouse experiment was implemented using open cone-beam CTsystem.Secondly, in response to the axial intensity fall-off caused by the incompleteprojection data collection of the FDK-based reconstruction, an empirical weightedfunction was adopted to modify the original FDK algorithm. Under the condition ofnon-additional scanning trajectory, the reciprocal Gaussian function was employed asthe pre-weighted factor for FDK algorithm to effectively compensate the axialintensity of the reconstructed images. Then the noise suppression effect of themodified algorithm was quantitatively evaluated through the simulation investigationand the phantom experiment verified its compensation effect.Thirdly, to solve the reconstruction artifacts and image distortion resulted fromthe imaging system geometric error, a phantom-based calibration method wasintroduced for the elimation of the geometric misalignments in the study. Using ourcalibration phantom, the projection matrices of the cone-beam x-ray imaging systemwere previously acquired under all projection views. Then each extracted projection matrix can be employed for the corresponding projection view to calibrate themis-mappings between the object coordinates and the detector projections. Theverification experiment for the calibration effectiveness was performed under therotation axis horizontal-shift from the central x-ray. The experimental resultsdemonstrated that the phantom-based calibration method could effectively minimizethe reconstruction artifacts for the imaging system and hold potential to benefit other3D medical imaging applications.Since corresponding solutions were proposed to overcome several key effects inthe cone-beam x-ray imaging. This study promotes the investigation of cone-beamx-ray imaging with crucial practical significance.