应用先验插值校正CT金属伪影  

Metal artifact reduction in computed tomography based on prior interpolation

作　　者：李铭[1,2,3] 卢彦飞[1,2,3] 袁刚[1] 吴中毅[1] 张涛[2] 

机构地区：[1]中国科学院苏州生物医学工程技术研究所,江苏苏州215163 [2]中国科学院长春光学精密机械与物理研究所,吉林长春130033 [3]中国科学院大学,北京100049

出　　处：《液晶与显示》2015年第6期1032-1039,共8页Chinese Journal of Liquid Crystals and Displays

基　　金：国家自然科学基金(No.61301042;No.61201117);国家重大科学仪器设备开发专项(No.2011YQ040082);中国科学院青年创新促进会(No.2014281);苏州市科技计划项目(No.ZXY2013001);江苏省自然基金项目(No.BK2012189;No.BK20151232)~~

摘　　要：针对CT系统在实际应用中出现的金属伪影问题,提出一种基于先验插值的金属伪影校正算法。文中通过预滤波、骨骼分割和软组织恢复步骤计算先验图像,并利用先验图像的正向投影对原始投影中的金属投影区进行插值校正。应用该算法对数值仿真图像和临床CT图像分别进行了校正重建实验。数值仿真实验表明,用提出算法校正的结果比线性插值金属伪影校正算法、归一化金属伪影校正算法校正的结果更接近理想体模。临床数据实验表明:该算法的重建结果有效抑制了金属伪影,清晰重建出金属边缘细节,极大地提高了重建图像的质量。Presence of metal artifacts in computed tomography （CT）is a major reason of degradation of image quality.In this work,a novel metal artifact reduction （MAR）algorithm using prior interpola-tion was developed to suppress metal artifacts in CT images.The prior image was generated by pre-filtering, bone extraction, and soft tissue restoring procedures. This prior image was then reprojected,and the estimation of surrogate data in the metal traces could be performed via the repro-jections.Next,the corrected experiments were conducted on digital phantom and clinical data respec-tively.The simulation experiment shows that results stem from prior interpolation MAR （PI-MAR） have higher similarity of ideal phantom than those from linear-interpolation MAR （LI-MAR）and Nor-malized MAR （NMAR）algorithms.Clinical experimental results demonstrate that the new algorithm can successfully suppress metal artifacts and clearly restore anatomical structure close to metallic implants.More-over,the quality of reconstructed images can be greatly improved by the proposed algorithm.

关 键 词：金属伪影 先验图像 预滤波 软组织恢复 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]