A Dual-energy Spectral CT Regularization Reconstruction Method Based On Segmentation

Dual-energy spectral CT,as a new type of CT imaging technology,can make full use of the x-ray energy spectrum characteristics and further display the detailed information of the sample during the imaging process.Compared with traditional CT,dual-energy spectral CT can achieve energy-and material-selective images.In terms of determination of material composition and concentration,hardening artifact removal,and material feature imaging,it has huge technical advantages that traditional CT cannot match.However,due to the nonlinearity of the dual-energy spectral CT reconstruction problem and the ill-conditioned of the decomposition model,the dual-energy spectral CT is very sensitive to data noise,which seriously affects the quality of the reconstructed images.Therefore,it is great significance to design a dual-energy spectral CT reconstruction method that can effectively suppress data noise.Based on the assumption of the piecewise constant and the dual-spectral CT material decomposition model,this thesis proposes a segmentation-based dual-energy spectral CT regularization reconstruction method.This method makes full use of the prior information shared by the boundaries between the material-based images reconstructed by dual-energy spectral CT,and integrates the image segmentation method into the entire reconstruction process.At the same time,the total variation(TV)minimization is used as the regularization term in the segmentation process to suppress the noise of the sampled data and the adverse effects caused by the model simplification.In terms of solving algorithm,this thesis designs an iterative framework to solve the reconstruction model.The algorithm needs to alternately iterate basis material images reconstruction and basis material images segmentation in the update.The basis material images reconstruction are updated by the E-ART algorithm,and the basis material images segmentation are solved by the primal-dual algorithm.Finally,the effectiveness of the proposed method in suppressing noise and removing hardening artifact are strongly verified by simulation experiments.