| Grating-based X-ray multi-contrast imaging can achieve precise modulation of Xrays by utilizing micro-grating elements,and simultaneously retrieve absorption,refraction and dark-field contrasts of the imaging object.Nowadays grating-based X-ray multicontrast imaging is the world-wide research frontier in the field of X-ray imaging.Due to the complementary advantages of different contrasts,this technology is going from bench to bed,and potential to produce the next generation of high-end X-ray imaging equipment.To promote practical applications of this technology,researches on information retrieval,dose optimization and fast scanning are carried out in this thesis.Information retrieval is the core theoretical cornerstone of grating-based X-ray multicontrast imaging.In this thesis,the author conducts in-depth analysis of the physical model and the system response in grating-based x-ray imaging,and reveals the integral property of the phase-stepping curve for the first time,which is an important guidance for the generalization and optimization of analytic information retrieval algorithms.By applying the integral property to the direct multi-order moment analysis,this thesis innovatively proposes a universal kernel transformation,and finds the noise-optimal form,which can reduce about 40%noise variance compared to the original direct multi-order moment analysis and demonstrate practical values of this property.In addition,this thesis innovatively deduces the Fourier-based form of multi-order moments of the scattering distribution,and establishes theoretical connections between two mainstream information retrieval methods(Fourier component analysis and multi-order moment analysis)for the first time,which can explain many published results of the similarities and differences between the two methods.Based on the theoretical form,an efficient truncated analytic multi-order moment analysis is proposed,and improves the computation efficiency by dozens of times compared with the original deconvolution-based method.Moreover,this truncated analytic method can retrieve a hybrid-field contrast that naturally fuses the refraction and dark-field signals,which provides more practical values.Dose optimization is one of the most critical problems for wide applications of grating-based X-ray multi-contrast imaging.This thesis breaks the paradigm of adopting the constant fluence distribution in the phase-stepping process for the first time,and proposes an fluence adaptation strategy for dose optimization.Then this thesis carries out strict mathematical modeling and theoretical derivations for our proposed universal kernel transformation and the commonly-used Fourier component analysis.Finally we found the contrast-based optimal adaptive fluence distributions,which can reduce about 20%dose compared with the traditional constant fluence under the same image quality.Fast scanning is the future trend of practical applications of grating-based X-ray multi-contrast imaging.In this thesis,a complete experimental scheme of fringe scanning is designed based on our laboratory’s existing imaging platform.By using an iterative fitting information retrieval method combined with the flat-field image matching prior,artifact-free images are obtained,which validates the effectiveness of our proposed method.In addition,this thesis designs a grating-based X-ray multi-contrast imaging system for security inspection,and conducts preliminary experiments on powdered contraband such as drugs,which shows that the dark-field signal can be several times higher than the traditional absorption signal,and therefore improve the detection rate of such contraband greatly.The findings of this thesis improve the information retrieval theoretical framework of grating-based X-ray multi-contrast imaging,and develop a new paradigm of dose optimization and a set of practical solutions for fast scanning,which provides important theoretical guidances and key technical supports for wide applications of grating-based X-ray multi-contrast imaging. |
