基于全广义变分约束加权最小二乘的低剂量计算机断层重建方法  

Total Generalized Variation Constrained Weighted Least-Squares for Low-Dose Computed Tomography Reconstruction

作　　者：牛善洲[1] 张梦真 邱洋 李硕 梁礼境 刘宏 刘国良[2] Niu Shanzhou;Zhang Mengzhen;Qiu Yang;Li Shuo;Liang Lijing;Liu Hong;Liu Guoliang(Ganzhou Key Laboratory of Computational Imaging,School of Mathematics and Computer Science,Gannan Normal University,Ganzhou 341000,Jiangxi,China;School of Medical Information Engineering,Gannan Medical University,Ganzhou 341000,Jiangxi,China)

机构地区：[1]赣南师范大学数学与计算机科学学院赣州市计算成像重点实验室,江西赣州341000 [2]赣南医学院医学信息工程学院,江西赣州341000

出　　处：《激光与光电子学进展》2023年第4期279-285,共7页Laser & Optoelectronics Progress

基　　金：国家自然科学基金(11701097);江西省科技创新杰出青年人才培养计划(20192BCB23019);江西省重点研发计划一般项目(20202BBE53024);江西省“双千计划”科技创新高端人才青年项目(jxsq2019201061);赣州市科技创新人才计划。

摘　　要：为了减少X射线的辐射剂量,提出了一种基于全广义变分约束加权最小二乘的低剂量计算机断层(CT)重建方法。首先对投影数据进行统计建模,然后将全广义变分正则化作为先验信息引入到投影数据恢复过程中,从而达到抑制噪声的目的,最后使用传统的滤波反投影算法进行CT图像重建。在Shepp-Logan体模实验中,提出方法的重建结果与Gibbs先验约束的惩罚加权最小二乘(Gibbs-WLS)、字典学习先验约束的惩罚加权最小二乘(DL-WLS)和全变分先验约束的惩罚加权最小二乘(TV-WLS)方法的重建结果相比,均方根误差分别降低了25.06%、1.50%和15.21%,信噪比分别提高了10.29%、0.53%和5.68%。在Clock体模实验中,提出方法的重建结果与Gibbs-WLS、DL-WLS和TV-WLS方法的重建结果相比,均方根误差分别降低了42.72%、23.45%和34.63%,信噪比分别提高了27.04%、11.42%和15.49%。实验结果表明,该方法在有效抑制低剂量CT图像的伪影和噪声的同时可以很好地保持图像的边缘信息和结构细节特征。In order to reduce the radiation dose of Xrays,we present a total generalized variation constrained weighted leastsquares approach for lowdose computed tomography(CT)reconstruction.Incorporating the total generalized variation regularization,a total generalized variation constrained weighted leastsquares(TGVWLS)approach is presented to reduce the noise in the projection(sinogram)domain,and the image is then reconstructed using the conventional filtered backprojection(FBP)algorithm.The root mean square errors(RMSEs)of the SheppLogan image reconstructed by the TGVWLS method are reduced by 25.06%,1.497%,and 15.21%,and the signaltonoise ratio(SNR)values increased by 10.29%,0.53%,and 5.68%,respectively,as compared with those of the Gibbs constrained weighted leastsquares(GibbsWLS),dictionary learning constrained weighted leastsquares(DLWLS),and total variation constrained weighted leastsquares(TVWLS)methods.In addition,for the Clock images reconstructed by the TGVWLS method,the RMSEs are reduced by 42.72%,23.45%,and 34.63%,and SNR values increased by 27.04%,11.42%,and 15.49%,respectively,as compared with those of the Gibbs-,DL-,and TVWLS methods.The experimental results show that the TGVWLS method can achieve noticeable gains in terms of noiseinduced artifact suppression and edge information and structural details preservation.

关 键 词：成像系统 低剂量计算机断层重建 全广义变分 加权最小二乘 图像重建 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]