CS-MRI 中评价随机欠采样矩阵的新方法

肖洒1; 2; 吕植成1; 孙献平1; 叶朝辉1; 周欣1*

doi:10.11938/cjmr20150404

波谱学杂志 >

2015 , Vol. 32 >Issue 4: 584 - 595

DOI: https://doi.org/10.11938/cjmr20150404

研究论文

CS-MRI 中评价随机欠采样矩阵的新方法

肖洒1 ,
2 ,
吕植成1 ,
孙献平1 ,
叶朝辉1 ,
周欣1*

展开

1. 波谱与原子分子物理国家重点实验室，中国科学院生物磁共振分析重点实验室，武汉磁共振中心（中国科学院武汉物理与数学研究所），湖北武汉 430071；
2. 中国科学院大学，北京 100049

肖洒(1990-)，男，湖北襄阳人，博士研究生，分析化学专业，主要研究方向为磁共振图像重建与压缩感知． *通讯联系人：周欣，电话： 027-87198802， E-mail: xinzhou@wipm.ac.cn.

收稿日期: 2015-02-15

修回日期: 2015-10-22

网络出版日期: 2015-12-05

基金资助

国家自然科学基金资助项目(81227902, 11174327)

收起

A New Method for Evaluation of Random Undersampling Matrix in Compressed Sensing-MRI

XIAO Sa1 ,
2 ,
LV Zhi-cheng1 ,
SUN Xian-ping1 ,
YE Chao-hui1 ,
ZHOU Xin1*

Expand

1. State Key Laboratory of Magnetic Resonance and Atomic and Molecular Physics, Key Laboratory of Magnetic Resonance in Biological Systems, National Center for Magnetic Resonance in Wuhan (Wuhan Institute of Physics and Mathematics, Chinese Academy of Sciences), Wuhan 430071, China; 2. University of Chinese Academy of Sciences, Beijing 100049, China

*Corresponding author： ZHOU Xin, Tel: +86-27-87198802, E-mail: xinzhou@wipm.ac.cn.

Received date: 2015-02-15

Revised date: 2015-10-22

Online published: 2015-12-05

Supported by

国家自然科学基金资助项目(81227902, 11174327)

Fold

摘要

在压缩感知-磁共振成像(CS-MRI)中，随机欠采样矩阵与重建图像质量密切相关．而选取随机欠采样矩阵一般是通过计算点扩散函数(PSF)，以可能产生的伪影的最大值为评价参数，评估欠采样对图像重建的影响，然而最大值只反应了伪影的最坏情况．该文引入了两种新的统计学评价参数平均值(MV)和标准差(SD)，其中平均值评估了伪影的平均大小，标准差可以反映伪影的波动情况．该文分别使用这3种参数对小鼠和人体脑部MRI数据以不同的采样比率进行CS图像重建，实验结果表明，当采样比率不低于4倍稀疏度时，使用平均值获得了质量更优的重建图像．因此，通过稀疏度先验知识指导合理选取采样比率，并以平均值为评价参数选取随机欠采样矩阵，能够获得更优的CS-MRI重建图像．

关键词： 磁共振成像(MRI); 压缩感知; 随机欠采样矩阵; 点扩散函数

本文引用格式

肖洒1 , 2 , 吕植成1 , 孙献平1 , 叶朝辉1 , 周欣1* . CS-MRI 中评价随机欠采样矩阵的新方法
[J]. 波谱学杂志, 2015 , 32(4) : 584 -595 . DOI: 10.11938/cjmr20150404

Abstract

In compressed sensing magnetic resonance imaging (CS-MRI), the quality of reconstructed image is largely determined by the random undersampling matrix. It is a common practice to select the random undersampling matrix though computation of the point spread function (PSF) and the maximal artifacts possible. In this paper, we proposed to use two novel statistical parameters, mean value (MV) and standard deviation (SD), to guide the selection of random undersampling matrix. The two parameters evaluate the average amplitude and fluctuation of the possible artifacts, respectively. Experiments on mice brain and human brain were used to compare image quality of CS reconstructions of MRI data acquired with random undersampling matrices determined by different criteria. It was shown that reconstruction with MV had better performance when the sampling ratio is above four times of sparsity. It is concluded that better CS-MRI reconstruction quality can be achieved with reasonable selection of sampling ratio guided by prior knowledge of sparsity and MV as random undersampling matrix evaluation parameter.

Key words： MRI; compressed sensing; random undersampling matrix; point spread function

参考文献

[1] Candes E J, Romberg J K, Tao T. Robust uncertainty principles: Exact signal reconstruction from highly incomplete frequency information[J]. IEEE T Inform Theory, 2006, 52(2): 489－509.

[2] Donoho D L. Compressed sensing[J]. IEEE T Inform Theory, 2006, 52(4): 1 289－1 306.

[3] Candes E J, Tao T. Near-optimal signal recovery from random projections: Universal encoding strategies[J]. IEEE T Inform Theory, 2006, 52(12): 5 406－5 425.

[4] Wech T, Lemke A, Medway D, et al. Accelerating cine-MR imaging in mouse hearts using compressed sensing[J]. J Magn Reson Imaging, 2011, 34(5): 1 072－1 079.

[5] Gao Ming-sheng(高明生), Xie Hai-bin(谢海滨), Yan Xu(严序), et al. Selective dual-direction sequential compressed sensing for dynamic MR imaging(选择性双向顺序压缩感知重建动态磁共振成像)[J]. Chinese J Magn Reson(波谱学杂志), 2013, 30(2): 194－203.

[6] Li Hai-dong(李海东), Zhang Zhi-ying(张智颖), Han Ye-qing(韩叶清), et al. Lung MRI using hyperpolarized gases(超极化气体肺部磁共振成像)[J]. Chinese J Magn Reson(波谱学杂志), 2014, 31(3): 307－320.

[7] Lustig M, Donoho D L, Pauly J M. Sparse MRI: The application of compressed sensing for rapid MR imaging[J]. Magn Reson Med, 2007, 58(6): 1 182－1 195.

[8] Lustig M, Donoho D L, Santos J M, et al. Compressed sensing MRI[J]. IEEE Signal Proc Mag, 2008, 25(2): 72－82.

[9] Ajraoui S, Lee K J, Deppe M H, et al. Compressed sensing in hyperpolarized He-3 lung MRI[J]. Magn Reson Med, 2010, 63(4): 1 059－1 069.

[10] Qu X B, Zhang W R, Guo D, et al. Iterative thresholding compressed sensing MRI based on contourlet transform[J]. Inverse Probl Sci En, 2010, 18(6):737－758.

[11] Ravishankar S, Bresler Y. MR image reconstruction from highly undersampled k-space data by dictionary learning[J]. IEEE T Med Imaging, 2011, 30(5): 1 028－1 041.

[12] Qu X B, Guo D, Ning B D, et al. Undersampled MRI reconstruction with patch-based directional wavelets[J]. Magn Reson Imaging, 2012, 30(7): 964－977.

[13] Song Y, Zhu Z, Lu Y, et al. Reconstruction of magnetic resonance imaging by three-dimensional dual-dictionary learning[J]. Magn Reson Med, 2014, 71(3): 1 285－1 298.

[14] Qu X B, Hou Y K, Lam F, et al. Magnetic resonance image reconstruction from undersampled measurements using a patch-based nonlocal operator[J]. Med Image Anal, 2014, 18(6): 843－856.

[15] Chen S S B, Donoho D L, Saunders M A. Atomic decomposition by basis pursuit[J]. Siam J Sci Comput, 1998, 20(1): 33－61.

[16] Ye J C, Tak S, Han Y, et al. Projection reconstruction MR imaging using FOCUSS[J]. Magn Reson Med, 2007, 57(4): 764－775.

[17] Donoho D L, Elad M, Temlyakov V N. Stable recovery of sparse overcomplete representations in the presence of noise[J]. IEEE T Inform Theory, 2006, 52(1): 6－18.

[18] Daubechies I, Defrise M, De Mol C. An iterative thresholding algorithm for linear inverse problems with a sparsity constraint[J]. Commun Pur Appl Math, 2004, 57(11): 1 413－1 457.

[19] Starck J L, Elad M, Donoho D L. Image decomposition via the combination of sparse representations and a variational approach[J]. IEEE T Image Process, 2005, 14(10): 1 570－1 582.

[20] Elad M, Matalon B, Zibulevsky M. Coordinate and subspace optimization methods for linear least squares with non-quadratic regularization[J]. Appl Comput Harmon Anal, 2007, 23(3): 346－367.

[21] Kim S J, Koh K, Lustig M, et al. An interior-point method for large-scale l(1)-regularized least squares[J]. IEEE J STSP, 2007, 1(4): 606－617.

[22] Ajraoui S, Parra-Robles J, Wild J M. Incorporation of prior knowledge in compressed sensing for faster acquisition of hyperpolarized gas images[J]. Magn Reson Med, 2013, 69(2): 360－369.

[23] Imai H, Kimura A, Hori Y, et al. Hyperpolarized ¹²⁹Xe lung MRI in spontaneously breathing mice with respiratory gated fast imaging and its application to pulmonary functional imaging[J]. NMR Biomed, 2011, 24(10): 1 343－1 352.

[24] Pang Y, Zhang X. Interpolated compressed sensing for 2D multiple slice fast MR imaging[J]. PLOS ONE, 2013, 8(2): e56098.

[25] Candes E J, Wakin M B. An introduction to compressive sampling[J]. IEEE Signal Proc Mag, 2008, 25(2): 21－30.

[26] Sun Yu-bao(孙玉宝), Wei Zhi-hui(韦志辉), Wu Min(吴敏), et al. Image poisson denoising using sparse representations (稀疏性正则化的图像泊松去噪算法)[J]. Acta Electronica Sinica(电子学报), 2011, 39(2): 285－290.

[27] Fang L, Li S, Nie Q, et al. Sparsity based denoising of spectral domain optical coherence tomography images[J]. Biomedical Opt Express, 2012, 3(5): 927－942.

Options

文章导航

模态框（Modal）标题

摘要

本文引用格式

Abstract

参考文献