Noise Analysis Of X-ray Differential Phase Contrast Imaging

Since 1895,when the German physicist Johann Roentgen discovered X-ray and obtained the first X-ray image in human history,X-ray imaging has been a hot research topic,and has been widely used in medical imaging,industrial nondestructive testing and other fields.However,due to the weakly absorption of X-ray for light elements,the conventional absorption-based imaging technology is not effective for the imaging of elements with objects composed of low atomic numbers.In the 1990 s,the X-ray phase contrast imaging method was proposed to effectively solve this problem.Because the phase shift cross section of the light element is usually 1000 times of the absorption cross section in the hard X-ray,which makes the phase contrast image more sensitive,and higher quality images can be obtained by using the phase information.After many years of development,X-ray phase contrast imaging has developed a variety of different imaging techniques,including crystal interferometer imaging,phase propagation imaging,diffraction enhanced imaging,grating phase contrast imaging and edge illumination.In each imaging technology,different information extraction methods have been developed to decompose the mixed information collected by the detector into three different kinds of information: absorption,refraction and scattering.In this paper,the information extraction methods of diffraction enhanced imaging and grating phase contrast imaging are systematically introduced,and the noise properties of the extracted images are studied.The work mainly includes the following two aspects:(1)Noise analysis of diffraction enhanced imagingThe three-image information extraction method based on Gaussian function fitting rocking curve in diffraction enhanced imaging can effectively extract three complementary signals and obtain information of different aspects of objects.Although three images can obtain three kinds of information,the quality of images collected from different angles is obviously different.In this paper,based on extracted results of three images method,using the error propagation formula,the noises model of absorption,refraction and scattering signals is established,and based on MATLAB platform,the numerical calculation is implemented to study the optimal angular position of analysis crystal of minimized noise variance.In other words,one can reduce image noise and improve image quality by angular optimization.(2)Revised noise model of grating-based dark-field imagingThe experimental results show that the existing noises models of dark-field will significantly overestimate the dark field image's noise under high visibility,which misleads image interpretation and image denoising.Based on the theoretical extraction formula of dark-field signal of phase-stepping method,we propose a revised noise model of dark-field image by using error propagation formula.And compared with the existing noise models,the revised noise model has a great agreement with the experimental results under both low and high visibilities.