波谱学杂志

• 低场磁共振技术与应用专栏 • 上一篇    下一篇

低场核磁共振扩散-横向弛豫二维反演算法

李新军1,聂生东1*,王远军1,杨培强2   

  1. 1.  上海理工大学 医学影像工程研究所,上海 200093; 
    2.  上海纽迈电子科技有限公司,上海 200333
  • 收稿日期:2012-05-10 修回日期:2012-08-20 出版日期:2013-09-05 发布日期:2013-09-05
  • 作者简介:李新军(1985-),男,山东潍坊人,硕士研究生,研究方向为微弱信号处理与分析. * 通讯联系人:聂生东(1962-),男,山东人,博士,教授,博士生导师,主要研究方向为医学成像技术及图像处理,电话:021-55271172,E-mail: nsd4647@163.com.
  • 基金资助:

    国家自然科学基金资助项目(60972122),上海市重点学科建设资助项目(P0502).

An Improved Two- Dimensional Inversion Algorithm for Low-Field NMR Diffusion- Transverse Relaxation Correlation Data

LI Xin-jun1,NIE Sheng-dong1*,WANG Yuan-jun1,YANG Pei-qiang2   

  1. 1. Institute of Medical Imaging Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China;
    2. Shanghai Niumag Corporation, Shanghai 200333, China
  • Received:2012-05-10 Revised:2012-08-20 Online:2013-09-05 Published:2013-09-05
  • About author:Nie Sheng-dong, Tel: 021-55271172,E-mail: nsd4647@163.com.
  • Supported by:

    国家自然科学基金资助项目(60972122),上海市重点学科建设资助项目(P0502).

摘要:

针对低场核磁共振一维反演中无法分辨一维谱中重叠组分和目前报道的扩散-横向弛豫二维反演算法计算量大、计算耗时长的问题,提出了一种计算量小、计算效率高、耗时短的扩散-横向弛豫二维反演算法. 首先对扩散系数D-和横向弛豫时间T2进行布点;其次根据信号采集条件计算出两个核心矩阵,并分别进行奇异值分解;然后,由所采集信号计算出两个核心矩阵的奇异值截断值,分别对两个核心矩阵的奇异值矩阵进行截断并求其逆矩阵;最后计算出初始反演结果,并添加非负约束经过多次迭代得到最终反演结果. 实验结果证明,提出的扩散-横向弛豫二维反演算法在不影响反演结果准确性的基础上,能极大提高计算效率.

关键词: 核磁共振(NMR), 扩散, 横向弛豫,  , 二维反演

Abstract:

NMR spectra acquired at low field often show severe signal overlap. Under such circumstance, two-dimensional (2D) diffusion-transverse relaxation (D-T2) correlation data are useful to differentiate signals from different components in the sample. The currently available inversion algorithms for D-T2 correlation data are time- and memory-consuming. To overcome this problem, we developed an improved 2D D-T2 inversion algorithm, which is more effective and consumes less memory and time than the traditional inversion algorithms. In brief, the logarithm values of discrete diffusion coefficient and transverse relaxation time were first obtained and used to construct two core matrices according to the signal collection conditions. Singular value decomposition was then carried out for the two matrices, followed by application of a cutoff value to remove negligible values. The inverse matrices of the thresholded core matrices were calculated and used to derive the initial inversion result. Finally, the initial inversion results were iterated with non-negative constraints to derived the final inversion results.

Key words: nuclear magnetic resonance, diffusion, transverse relaxation, two-dimensional inversion

中图分类号: