Exploration Of Grating-based Imaging System With A Large Field Of View And A Conventional X-ray Source

Compared with traditional X-ray absorption imaging,grating-based imaging(GI)can obtain absorption,phase-contrast and dark-field simultaneously,which is a research hotspot in the field of X-ray imaging in recent years.The GI system overcomes the resolution limitation in the absorption image when imaging the material composed of light elements,and is applicable to the conventional incoherent X-ray source.As a result,GI has greater application prospects in clinical diagnosis,industrial production and safety inspection.This thesis focuses on the exploration of grating-based imaging system with a large field of view and a conventional X-ray source,including the design of the large-field-of-view projected-fringe-based GI system(LP-GI),the investigation of the traditional and novel information retrieval algorithms and the imaging methods and applications of the LP-GI system.Firstly,the design of the LP-GI system is studied.Since the traditional GI system is of small field of view(FOV)and high radiation dose,the LP-GI system has larger FOV,faster scanning and lower dose.In the new system,gratings of large areas are applied to enlarge the FOV,which is up to 400 × 500 mm2.The between-G1-and-G2 design can reduce the dose to half theoretically than the between-G0-and-G1 design,with the similar level of image quality and even better image contrast and sharpness.Trigger-based imaging mode is performed in the system,with a high tube current and a short exposure time,resulting in the increasing overall scanning efficiency.This imaging mode can also avoid the generation of imaging-independent X-rays,further reducing the radiation dose.Secondly,the information retrieval algorithms in GI are investigated.In the conven-tional CMA algorithm,the phase-contrast image may have a phase wrapping problem,while the conventional SAXS algorithm takes a long time to finish the iterative decon-volution,where the iterative method and times affect the retrieved images.Since the phase-stepping curve of background accords well with the cosine model,only the zeroth and first orders of the Fourier series of the scattering distribution contribute to the convolu-tion relationship.As a consequence,novel fast information retrieval algorithms based on the analytic scattering distribution,the principal component analysis and the convolution theorem are proposed,respectively.The simulation and real experiments prove that these novel algorithms greatly improve the computational efficiency,with the similar level of image quality and even more rich structural details.Finally,the imaging methods and applications of the LP-GI system are performed.To overcome the limitation of the resolution direction of the one-dimensional grating,a multi-directional imaging method is proposed to obtain the complete expression of the two-dimensional structure,using an image fusion method based on the NSCT transform.The dark-field imaging of lung is carried out with quantitative indexes,where parts of lung overlapped by organs such as heart and liver are clearly observed,demonstrating that GI is important to the diagnosis of lung diseases.Besides,food quality inspection is performed.It is found that phase-contrast images,especially dark-field images are able to distinguish the fresh and rotten fruits or meats.