Research On Sparse Projection Reconstruction Algorithm For Tube Excited X-Ray Fluorescence CT

As a molecular imaging mode,X-ray fluorescence computed tomography（XFCT）can obtain the distribution and content of high Z elements in organisms with a non-invasive and non-destructive manner,synchronously presenting structural and functional information within organisms,and has broad application prospects in the early diagnosis of diseases such as cancer.Among them,tube excited X-ray fluorescence CT has become a research hotspot due to its low cost,mobility,and other advantages.However,at present,the research on tube excited X-ray fluorescence CT is still in initial stage,and there are still many technical issues to be resolved.So,this paper based on the compressed sensing theory and the tube excited X-ray fluorescence CT imaging system,the reconstruction algorithm of X-ray fluorescence CT with sparse projection was studied.The main research contents of this paper are as follows:（1）A sparse projection reconstruction algorithm for X-ray fluorescence CT based on total variation was studied.Due to the traditional XFCT reconstruction algorithm has poor image quality in sparse projection.Therefore,this paper combined OSEM algorithm with total variation to propose OSEM-TV reconstruction algorithm.At the same time,a fan-beam X-ray fluorescence CT imaging system based on parallel hole collimation was simulated and designed by Geent4 software.And the reconstructed image quality of OSEM algorithm and OSEM-TV algorithm was analyzed and compared through the evaluation parameter CNR.（2）A sparse projection reconstruction algorithm for X-ray fluorescence CT based on multi-directional total variation was studied.Because the traditional total variation only utilizes the gradient information of the image in the horizontal and vertical directions,and does not utilize the gradient information of the image in other directions,there is still some room for improvement.Based on this,this paper proposed Md TV-EM algorithm by combining traditional total variation with diagonal total variation.It utilized the gradient information of the image in the horizontal,vertical,and diagonal directions.After one iteration of the MLEM algorithm,it successively introduces total variation and diagonal total variation to optimize the reconstructed image.At the same time,the evaluation parameter CNR was used to analyze and compare the reconstructed images of the proposed algorithm.（3）A sparse projection reconstruction algorithm based on L_1/2-norm for X-ray fluorescence CT was studied.In image reconstruction algorithms based on compressed sensing theory,the L₁-norm is often used as the regularization norm.For example,the total variation applied in the research contents（1）and（2）is essentially the L₁-norm of gradient image.However,in fact,compared to the L₁-norm,the L_1/2-norm is closer to the L₀-norm and sparser.Based on this,this paper proposed L_1/2-EM algorithm by combining the L_1/2-norm with the Split Bregman algorithm,and simulation experiments have verified the effectiveness of the algorithm.At the same time,XFCT reconstructed images are analyzed and compared by evaluation parameters RMSE and UQI.