东北大学学报:自然科学版 ›› 2018, Vol. 39 ›› Issue (7): 964-969.DOI: 10.12068/j.issn.1005-3026.2018.07.011

• 信息与控制 • 上一篇    下一篇

基于增广拉格朗日的全变分正则化CT迭代重建算法

孝大宇, 郭洋, 李建华, 康雁   

  1. (东北大学 中荷生物医学与信息工程学院, 辽宁 沈阳110169)
  • 收稿日期:2017-03-21 修回日期:2017-03-21 出版日期:2018-07-15 发布日期:2018-07-11
  • 通讯作者: 孝大宇
  • 作者简介:孝大宇(1980-),男,辽宁阜新人,东北大学博士研究生; 康雁(1964-),男,辽宁沈阳人,东北大学教授,博士生导师.冯明杰(1971-), 男, 河南禹州人, 东北大学副教授; 王恩刚(1962-), 男, 辽宁沈阳人, 东北大学教授,博士生导师.
  • 基金资助:
    国家自然科学基金资助项目(61372014).国家自然科学基金资助项目(51171041).

Total Variation Regularization CT Iterative Reconstruction Algorithm Based on Augmented Lagrangian Method

XIAO Da-yu, GUO Yang, LI Jian-hua, KANG Yan   

  1. School of Sino-Dutch Biomedical & Information Engineering, Northeastern University, Shenyang 110169, China.
  • Received:2017-03-21 Revised:2017-03-21 Online:2018-07-15 Published:2018-07-11
  • Contact: KANG Yan
  • About author:-
  • Supported by:
    -

摘要: 采用一种基于增广拉格朗日方法(augmented Lagrangian method)求解全变分正则化(total variation regularization)算法(ALMTVR)来进行CT图像重建.将ALMTVR算法与经典的代数重建算法(algebraic reconstruction technique,ART)进行比较,并采用仿真数据与实际数据进行实验.在实验中,使用ALMTVR算法与ART算法分别进行图像重建,并对重建图像进行对比分析.实验结果表明:所提算法与ART算法相比,显著提高了图像重建的质量与速度,显示了其对图像重建的有效性及在CT成像系统中潜在的应用价值.

关键词: CT迭代重建, 增广拉格朗日方法, 全变分正则化, 仿真数据, 实际投影数据

Abstract: A novel algorithm based on augmented Lagrangian method was presented to solve total variation regularization problem (ALMTVR) of the CT iterative reconstruction. The classical algebraic reconstruction technique (ART) was compared with the ALMTVR algorithm, the simulation data and actual data are used in the experiment. The ALMTVR algorithm and the ART algorithm were used to reconstruct the images respectively, and the reconstruction images were compared and analyzed. Results showed that, compared with ART algorithm, the proposed algorithm can significantly improve image quality and reconstruction speed, which indicates the proposed algorithm is effective and has potential applications in the CT imaging system.

Key words: CT iterative reconstruction, augmented Lagrangian method, total variation regularization, simulation data, real projection data

中图分类号: