Research On Image Reconstruction Algorithms Of Dual Energy Computed Tomography

Computed Tomography technique uses X-ray transmission attenuation information about object at different angles to obtain the attenuation characteristic distribution information of the object through the image reconstruction algorithm,in order to perspective imaging of the internal structure of the object without destruction or injury.Compared with the traditional single spectrum CT technology,Dual Energy CT(DECT)imaging using transmission attenuation difference of the object in different X-ray energy spectrum to obtain more abundant information of object's internal physical characteristic distribution,can effectively improve the resolution of the materials,suppress or eliminate beam hardening artifacts,and provide more accurate quatitative information,which has become the research hotspot and development trend of CT technology.However,DECT imaging results are more easily affected by noise fluctuations and inconsistency in projection data,which lead to a significant decrease in signal to noise ratio of image reconstruction.Therefore,an important issue need to be researched on promoting image quality of DECT.In this paper,with the focuses on the DECT imaging technology in high precision and fast decomposition of high and low energy projection data,noise suppression of the reconstructed image of decomposition coefficient distribution and DECT image reconstruction in sparse projection angle,DECT image reconstruction algorithms of projection-domain preprocessing class,image-domain postprocessing class and iterative class are studied.The main research achievements obtained are as follows:1.In view of the ill-condition of the matrix of the measurement system,a X-ray energy spectrum algorithm based on weighted difference regularization is proposed.X-ray energy spectrum estimation is established by using weighted difference regularization model with comprehensive utilization of the fidelity of transmission attenuation measurement data,the continuous characteristics of bremsstrahlung spectrum part,the discrete characteristics of the characteristic radiation spectrum part,non-negative and normalization of energy spectrum and the average effective attenuation coefficient and other information.L curve criterion in the regularization theory is used to determine the regularization parameter in order to ensure the accuracy of the problem solving.Experimental results show that compared with EM and ART estimation methods,the proposed method is not affected by the initial energy spectrum,and the accuracy of estimation by the proposed method is impoved significantly.2.High order polynomial fitting and iterative solution after taking the negative logarithm projection is needed in the iterative method,resulting in quite complex computation and low decomposition efficiency,and the look-up table method is limited by storage size and computation speed with a finite precision.A fast projection decomposition algorithm base on iso-transmission curve fitting for DECT is proposed.In view of the above problems,iso-transmission curve fitting model of high and low energy projection is established to transform the high and low energy projection joint nonlinear solving problem into solving high and low energy projection iso-transimission curve intersection problem.Two order Taylor series expansion of the projection partition is applied to get the analytical solution with high accuracy for a given polychromatic projection value.A quadratic polynomial fitting model of iso-transmission curve is separatively built for high and low energy projection value.Through the analytic solution of the intersection point of the iso-transmission curves,the fast solution of the high and low energy projection decomposition is realized.Experimental results show that compared to projection matching method,the projection decomposition precision of the proposed method is increased significantly.At the same time,computation efficiency is increased.3.To deal with the problem of noise fluctuation and directional artifacts in DECT image reconstruction based on projection domain preprocessing,a DECT image reconstruction algorithm based on adaptive bilateral filtering and weighted fidelity term is proposed.The adaptive bilateral filtering is used to suppress the noise fluctuation and maintain the edge of decomposition coefficient projection data.The fidelity term is weighted by calculating the error between the measurement polychromatic projection and the estimated polychromatic projection of two filtered basis function decomposition coefficient projections,to avoid interference of image reconstruction results with the projections with serious deviation from the true value.DECT basis function decomposition coefficient images are reconstructed by alternating direction Total Variation(TV)minimization method.Experiments show that the proposed method can effectively suppress the noise fluctuation in DECT image reconstruction,and eliminate the directional artifacts in the reconstructed image.4.A DECT image reconstruction algorithm based on TV minimum and penalized least squares decomposition is proposed in terms of DECT image reconstruction based on image-domain postprocessing.The reconstruction algorithm is formulated as an optimization problem,which contains total variation regularization term and data fidelity term.The alternating direction method is utilized to design the corresponding algorithm which shows faster convergence speed compared with the existing ones.The image domain decomposition applies the penalized least square estimation(PLS)on decomposing the material mappings.The PLS includes linear combination term and the regularization term which enforcing the smoothness on estimation images.The experiments in full angles and sparse angles shows that the proposed method has advantages on noise suppression and edge reservation,without blurring the fine structures in the sinus area in the phantom.Furthermore,the proposed methods consume much less time than the compared iterative reconstruction algorithms.5.A DECT image reconstruction algorithm based on first order linear approximation with alternating direction method is proposed in terms of the iterative DECT image reconstruction.TV regularization and material decomposition model is established for DECT image reconstruction.The model is transformed into four alternative subproblems with the augmented Lagrangian method under alternating direction framework.In the nonlinear subproblem solving process,the first order Taylor expansion method is used to linear approximation,and the analytic solution based on FFT is obtained by matrix diagonalization,so that all the subproblems have analytical solution with fast calculation.In the absence of noise and additive noise cases as well as different sparse angle sampling cases,experiments show that compared to ESART algorithm and its improved algorithm ESART-TV algorithm,the proposed algorithm has significant advantages in the quality of the reconstructed image and the the performance of the iterative convergence.