基于正则项的能谱CT投影分解方法研究

Research on Projection Decomposition Method of Spectral CT Based on Regularization

能谱CT通过探索不同X射线能量下衰减特性的差异,具有定量分析材料的能力。然而,利用光子计数探测器获得的多个投影数据在进行材料分解的过程中,会受到噪声放大的影响。本文在特定材料正则项的基础上,加入张量分解的正则项,用于降低噪声,提高分解效果。实验结果表明,本文提出的算法可以清楚地分解出不同材料的投影图,利用重建算法对材料投影图进行层析重建后,材料图像的边缘结构比较清晰,并且抑制了分解过程中的噪声,提高了图像质量。

Spectral CT has the ability to quantitative analysis materials by exploring the differences of attenuation characteristics under different X-ray energy.However,during the process of multiple projection data obtained by photon-counting detectors for material decomposition,it will be affected by noise amplification.In this paper,on the basis of the regular term of specific materials,the regular term of tensor decomposition is added to reduce noise and improve the decomposition effect.The experiment results show that the algorithm proposed can clearly decompose the projection diagram of different materials.After using the reconstruction algorithm to perform tomographic reconstruction on the material projection images,the edge structure of the reconstructed material image is relatively clear,and the noise in the decomposition process is suppressed,and the image quality is improved.