Analysis And Improvement Of Phase Contrast CT Image Quality

Conventional X-ray absorption imaging technology measures the absorption capacity of substances to X-rays.When the samples have similar absorption coefficients,it is difficult to obtain satisfactory contrast from the reconstruction images.Since the 1960s,X-ray phase contrast imaging technology has been developed.This method detects the phase modulation capacity of substances to X-rays and distinguishes the internal structure of the sample from another dimension,which can be used as a supplement to the absorption imaging technology.However,most phase contrast imaging technologies rely on synchrotron radiation or micro-focus sources,so their practicality is limited.The grating-based X-ray phase contrast imaging technology uses conventional X-ray tube as the source,and can obtain three kinds of information of absorption,refraction,and scattering in one data acquisition,so it has the potential for clinical medical applications.The quality of reconstruction images is crucial for X-ray imaging technology.Since the development of X-ray imaging technology,how to remove the ring artifacts in computed tomography has been a hot topic in the scientific and industrial circles.In this thesis,the grating-based X-ray phase contrast imaging system in the National Synchrotron Radiation Laboratory,University of Science and Technology of China is used to discuss the ring artifacts.By analyzing the generation mechanism and performance characteristics of artifacts,artifact removal algorithms for absorption signals and refraction signals are proposed respectively.Finally,the effectiveness of the algorithm is proved through experimental results and evaluation indicators.For the absorption signal,because the slope filter kernel of the reconstruction algorithm will amplify the high-frequency noise in the sinogram,the algorithm is divided into two parts:the sinogram domain and the reconstruction domain.In the sinogram domain,according to the spatial continuous distribution of the samples,the algorithms of sorting,filtering and resorting are used to remove artifacts.In the reconstruction domain,the characteristics of the ring artifacts are analyzed,and the image is converted to the polar coordinate system.At this time,the ring artifacts appear as stripe perpendicular to the polar radius direction.Filtering along the polar direction and making a difference to the images before and after filtering can obtain residual images containing artifacts and sample edges.The image segmentation algorithm based on machine learning is used to obtain the pixel distribution of each type of sample.In order to protect the edge information of the image,the internal pixel regions of the sample are obtained through morphological operations.Finally,the distribution characteristics of the residual image are used to locate the artifact pixels,and the mean value of adjacent non-artifact pixels are used instead.For the refracted signal,the anti-symmetry of the conjugate rays in the sinogram is analyzed according to the projection geometry of the fan beam CT scan.Calculating the average value of the sinogram along the angular direction.In theory,at the position of the conjugate detector,the average value is opposite to each other.The artifact regions can be located by this property,and the average value of the artifact regions can be fitted by interpolation of the average value of the surrounding non-artifact regions.Finally,the original sinogram can be corrected according to the fitted sinogram.The main content of this thesis includes the following six parts:1.Overview of the development of X-ray imaging technology;2.Briefly introduce the imaging system used in the experiment;3.Introduce the artifact removal algorithm of the absorption signal in detail;4.Introduce the artifact removal algorithm of the refraction signal in detail;5.Use existing models to evaluate algorithms;6.Summarize and look forward to this thesis.