Spectral CT Noise Reduction Reconstruction Algorithm Based On Compressed Sensing

Compared with the traditional CT,Spectral CT based on photon counting detector has attracted great attention because of its ability to identify and distinguish different materials.Photon counting detectors can divide the energy channels into several channels,increasing the resolution of energies.However,due to the division of energy channels,the number of photons in the single energy channel is much smaller than the total number of detected photons,so the noise in each energy channel increases dramatically.Therefore,how to reconstruct a high quality CT Image from the noise projection has become one of the hot issues in the field of current Spectral CT research.Based on this problem,the main research work of this paper is as follows:(1)This paper proposes an iterative reconstruction algorithm based on dictionary learning for spectral CT reconstruction,which relies on the theory of compressed sensing,Apply the alternating minimization method to optimize the related objective function and solve it by Split-Bregman algorithm.At the same time,the ordered subset method is used to accelerate the iterative convergence process and improve the operation rate.In order to validate and evaluate the proposed method,a simulation experiment was performed using a simple model and an actual clinical mouse model.The experimental results show that the proposed algorithm has a better denoising and detail preservation ability.(2)Although the traditional algorithm has a certain denoising effect,but when the noise is too large and the model structure is complex,the performance will be greatly reduced.Therefore,a new type of spectral CT reconstruction algorithm is to be proposed.Since the intraimage sparsity and interimage similarity are important prior knowledge for image reconstruction,Inspired by this observation,using a priori image constraints and dictionary learning to improve quality.Using the correlation of reconstructed images under different spectral channels,using high-quality omni-spectral reconstructed image as a priori image to guide narrow-spectral CT reconstruction,combined with dictionary learning theory,a constrained compression sensing framework based on prior image is proposed.the performance of the algorithm is verified by simulation experiments.In order to validate and evaluate theproposed method,a simulation experiment was performed using a simple model and an actual clinical mouse model.The experimental results show that the proposed algorithm has a better denoising and detail preservation ability.