-
摘要: 快速动态磁共振成像可以通过减少采样量来缩短信号的采集时间.因此，从下采样的数据中重建出高质量的图像成为研究的热点.目前，常见的重建方法利用动态图像序列的稀疏表示实现高质量的重建.本文提出了一种联合相邻帧预测（Joint adjacent-frame prediction，JAFP）的重建方法，首先根据动态图像序列相邻帧之间高度的相似性，联合预测当前帧图像，获得稀疏的图像差；其次，利用图像差序列在时间域的拟周期特性，通过傅里叶变换进一步提高图像差序列的稀疏度.在此基础上构建动态成像模型，并在压缩感知（Compressed sensing，CS）框架下进行求解.该方法可将前一次的重建结果作为新的输入，从而形成迭代算法.采用两个磁共振心脏电影成像数据集对提出的方法进行了实验验证，并与k-t FOCUSS ME/MC和MASTeR进行了比较.实验结果表明，该方法联合相邻帧改进了预测图像的效果，提升了重建图像的质量，具有广泛的应用价值.Abstract: Dynamic magnetic resonance imaging can be accelerated to reduce the signal acquisition time by under-sampling k-space data, so the quality of reconstructed images from under-sampled data has become the focus of research. Currently, the common approaches use sparse representation of dynamic image sequences to improve the reconstruction. This paper proposes a new method named joint adjacent-frame prediction (JAFP) based on the similarity between adjacent frames of dynamic image sequences. The JAFP promotes the quality of the predicted image sequence by jointsing adjacent frames prediction. Meanwhile, observing the quasi-periodicity of the difference of sequences, it further improves the sparsity by applying Fourier transform to the difference sequence along the time direction. Then a dynamic imaging model is setup to incorporate the fidelity constraint and joint sparse promotion, and it is solved in the framework of compressive sensing (CS). Two cardiac cine MR datasets are evaluated to verify the proposed method, and the proposed method is compared with k-t FOCUSS with ME/MC and MASTeR to show the better quality of reconstruction. Experimental results show that JAFP can improve the quality of reconstructed image and has important application value.
-
-
图 1 相邻帧图像与非相邻帧图像之间的差异
Fig. 1 The difference of between adjacent frame and non-adjacent frame
图 4 数据集1在4倍下采样时重建质量比较
Fig. 4 The comparison of reconstruction quality with acceleration factor of 4 for dataset 1
图 5 数据集1在8倍下采样时重建质量比较
Fig. 5 The comparison of reconstruction quality with acceleration factor of 8 for dataset 1
图 6 数据集1在8倍下采样时左心脏壁灰度均值的比较
Fig. 6 The comparison of reconstructed gray average values and errors with acceleration factor of 8 for left myocardial wall of dataset 1
图 7 数据集1在8倍下采样时右心脏壁灰度均值的比较
Fig. 7 The comparison of reconstructed gray average values and errors with acceleration factor of 8 for right myocardial wall of dataset 1
图 8 数据集1在4倍和8倍下采样时, 重建的ROI及其误差图
Fig. 8 The reconstructed ROI and errors with acceleration factors of 4 and 8 for dataset 1
图 9 在4倍和8倍下采样时, 重建的时间域截面及其误差图
Fig. 9 The reconstructed temporal slices and errors with acceleration factors of 4 and 8
图 10 数据集1在不同下采样倍数(2, 4, 8, 10, 16)下, 不同方法重建综合质量对比图
Fig. 10 The comparison of reconstructed quality with acceleration factors of 2, 4, 8, 10 and 16 for dataset 1
图 11 数据集2在4倍下采样时重建质量比较
Fig. 11 The comparison of reconstruction quality with acceleration factor of 4 for dataset 2
图 12 数据集2在8倍下采样时重建质量比较
Fig. 12 The comparison of reconstruction quality with acceleration factor of 8 for dataset 2
图 13 数据集2在8倍下采样时左心脏壁灰度均值的比较
Fig. 13 The comparison of reconstructed gray average values and errors with acceleration factor of 8 for left myocardial wall of dataset 2
图 14 数据集2在8倍下采样时右心脏壁灰度均值的比较
Fig. 14 The comparison of reconstructed gray average values and errors with acceleration factor of 8 for right myocardial wall of dataset 2
图 15 数据集2在4倍和8倍下采样时, 重建的ROI及其误差图
Fig. 15 The reconstructed ROI and errors with acceleration factors of 4 and 8 for dataset 2
图 16 数据集2在4倍和8倍下采样时, 重建的时间域截面及其误差图
Fig. 16 The reconstructed temporal slices and errors with acceleration factors of 4 and 8 for dataset 2
图 18 在8倍下采样时, 不同运动估计方法重建的ROI及其误差图(从左至右依次为心脏电影成像序列的第7帧、第12帧和第24帧ROI图像
Fig. 18 The reconstructed ROI and errors with different motion estimation plan (the reconstructed ROI of the 7th, 12th, 24th frames with acceleration factor of 8)
图 19 在8倍下采样时, 各迭代次数(0 ~ 7)下重建的第5帧图像ROI及其误差图
Fig. 19 The reconstructed the 5th frame ROIs and errors by 0 ~ 7 times iteration
图 20 在不同迭代次数(0 ~ 7)下, JAFP方法重建综合质量对比图
Fig. 20 The comparison of reconstructed quality with 0 ~ 7 iterations
表 1 运行时间对比
Table 1 The comparison of runtime
方法 运动估计耗时(s) 求解耗时(s) 迭代次数(次) k-t FOCUSS ME/MC 201.78 284.31 1 MASTeR 182.41 23 228.25 4 JAFP 183.32 16 250.73 4 
下载: 导出CSV
-
[1] Donoho D L. Compressed sensing. IEEE Transactions on Information Theory, 2006, 52(4):1289-1306 doi: 10.1109/TIT.2006.871582 [2] 李树涛, 魏丹.压缩传感综述.自动化学报, 2009, 35(11):1369-1377 http://www.aas.net.cn/CN/abstract/abstract13592.shtmlLi Shu-Tao, Wei Dan. A survey on compressive sensing. Acta Automatica Sinica, 2009, 35(11):1369-1377 http://www.aas.net.cn/CN/abstract/abstract13592.shtml [3] 荆楠, 毕卫红, 胡正平, 王林.动态压缩感知综述.自动化学报, 2015, 41(1):22-37 http://www.aas.net.cn/CN/abstract/abstract18580.shtmlJing Nan, Bi Wei-Hong, Hu Zheng-Ping, Wang Lin. A survey on dynamic compressed sensing. Acta Automatica Sinica, 2015, 41(1):22-37 http://www.aas.net.cn/CN/abstract/abstract18580.shtml [4] 彭义刚, 索津莉, 戴琼海, 徐文立.从压缩传感到低秩矩阵恢复:理论与应用.自动化学报, 2013, 39(7):981-994 http://www.oalib.com/paper/4417582Peng Yi-Gang, Suo Jin-Li, Dai Qiong-Hai, Xu Wen-Li. From compressed sensing to low-rank matrix recovery:theory and applications. Acta Automatica Sinica, 2013, 39(7):981-994 http://www.oalib.com/paper/4417582 [5] 齐聪慧, 赵志钦, 徐晶, 张海.复杂场景下基于压缩感知的目标电磁散射与成像.强激光与粒子束, 2014, 26(7):Article No. 073206 http://www.opticsjournal.net/abstract.htm?aid=OJ0630000460B9EaHdQi Cong-Hui, Zhao Zhi-Qin, Xu Jing, Zhang Hai. Electromagnetic scattering and image processing of targets under complex environment based on compressive sensing method. High Power Laser and Particle Beams, 2014, 26(7):Article No. 073206 http://www.opticsjournal.net/abstract.htm?aid=OJ0630000460B9EaHd [6] 秦翰林, 韩姣姣, 延翔, 周慧鑫, 李佳, 曾庆杰.基于改进的分块压缩感知红外图像重建.强激光与粒子束, 2014, 26(12):Article No. 121011 http://d.wanfangdata.com.cn/Periodical_qjgylzs201412014.aspxQin Han-Lin, Han Jiao-Jiao, Yan Xiang, Zhou Hui-Xin, Li Jia, Zeng Qing-Jie. Infrared image reconstruction based on a modified block compressed sensing. High Power Laser and Particle Beams, 2014, 26(12):Article No. 121011 http://d.wanfangdata.com.cn/Periodical_qjgylzs201412014.aspx [7] Otazo R, Candés E, Sodickson D K. Low-rank plus sparse matrix decomposition for accelerated dynamic MRI with separation of background and dynamic components. Magnetic Resonance in Medicine, 2015, 73(3):1125-1136 doi: 10.1002/mrm.25240 [8] 任越美, 张艳宁, 李映.压缩感知及其图像处理应用研究进展与展望.自动化学报, 2014, 40(8):1563-1575 http://www.aas.net.cn/CN/abstract/abstract18426.shtmlRen Yue-Mei, Zhang Yan-Ning, Li Ying. Advances and perspective on compressed sensing and application on image processing. Acta Automatica Sinica, 2014, 40(8):1563-1575 http://www.aas.net.cn/CN/abstract/abstract18426.shtml [9] Lai Z Y, Qu X B, Liu Y S, Guo D, Ye J, Zhan Z F, Chen Z. Image reconstruction of compressed sensing MRI using graph-based redundant wavelet transform. Medical Image Analysis, 2016, 27:93-104 doi: 10.1016/j.media.2015.05.012 [10] Song L X, Zhang J G, Wang Q. MRI reconstruction based on three regularizations:total variation and two wavelets. Biomedical Signal Processing and Control, 2016, 30:64-69 doi: 10.1016/j.bspc.2016.06.003 [11] Ye J, Qu X B, Guo H, Liu Y S, Guo D, Chen Z. Patch-based directional redundant wavelets in compressed sensing parallel magnetic resonance imaging with radial sampling trajectory. Journal of Medical Imaging and Health Informatics, 2016, 6(2):387-398 doi: 10.1166/jmihi.2016.1715 [12] Akçakaya M, Basha T A, Pflugi S, Foppa M, Kissinger K V, Hauser T H, Nezafat R. Localized spatio-temporal constraints for accelerated CMR perfusion. Magnetic Resonance in Medicine, 2014, 72(3):629-639 doi: 10.1002/mrm.v72.3 [13] Li Q Y, Qu X B, Liu Y S, Guo D, Lai Z Y, Ye J, Chen Z. Accelerating patch-based directional wavelets with multicore parallel computing in compressed sensing MRI. Magnetic Resonance Imaging, 2015, 33(5):649-658 doi: 10.1016/j.mri.2015.01.014 [14] Majumdar A, Ward R. Learning space-time dictionaries for blind compressed sensing dynamic MRI reconstruction. In: Proceedings of the 2015 IEEE International Conference on Image Processing. Quebec, Canada: IEEE, 2015. 4550-4554 [15] Li J S, Sun J Q, Song Y, Zhao J. Accelerating MRI reconstruction via three-dimensional dual-dictionary learning using CUDA. The Journal of Supercomputing, 2015, 71(7):2381-2396 doi: 10.1007/s11227-015-1386-z [16] Tsao J, Kozerke S. MRI temporal acceleration techniques. Journal of Magnetic Resonance Imaging, 2012, 36(3):543-560 doi: 10.1002/jmri.v36.3 [17] Kim D, Dyvorne H A, Otazo R, Feng L, Sodickson D K, Lee V S. Accelerated phase-contrast cine MRI using k-t SPARSE-SENSE. Magnetic Resonance in Medicine, 2012, 67(4):1054-1064 doi: 10.1002/mrm.23088 [18] Jung H, Ye J C, Kim E Y. Improved k-t BLAST and k-t SENSE using FOCUSS. Physics in Medicine and Biology, 2007, 52(11):3201-3226 doi: 10.1088/0031-9155/52/11/018 [19] Jung H, Sung K, Nayak K S, Kim E Y, Ye J C. k-t FOCUSS:a general compressed sensing framework for high resolution dynamic MRI. Magnetic Resonance in Medicine, 2009, 61(1):103-116 doi: 10.1002/mrm.v61:1 [20] Asif M S, Hamilton L, Brummer M, Romberg J. Motion-adaptive spatio-temporal regularization for accelerated dynamic MRI. Magnetic Resonance in Medicine, 2013, 70(3):800-812 doi: 10.1002/mrm.24524 [21] Dong X L, Liu H W, Xu Y L, Yang X M. Some nonlinear conjugate gradient methods with sufficient descent condition and global convergence. Optimization Letters, 2015, 9(7):1421-1432 doi: 10.1007/s11590-014-0836-5 [22] Gorodnitsky I F, Rao B D. Sparse signal reconstruction from limited data using FOCUSS:a re-weighted minimum norm algorithm. IEEE Transactions on Signal Processing, 1997, 45(3):600-616 doi: 10.1109/78.558475 [23] 阮钦杰. 基于压缩感知技术的快速动态磁共振成像技术研究[硕士学位论文], 杭州电子科技大学, 中国, 2014. http://www.wanfangdata.com.cn/details/detail.do?_type=degree&id=D485080Ruan Qin-Jie. The Study of Fast Dynamic Magnetic Resonance Imaging Based on Compressed Sensing Technique[Master thesis], Hangzhou Dianzi University, China, 2014. http://www.wanfangdata.com.cn/details/detail.do?_type=degree&id=D485080 [24] Orchard M T, Sullivan G J. Overlapped block motion compensation:an estimation-theoretic approach. IEEE Transactions on Image Processing, 1994, 3(5):693-699 doi: 10.1109/83.334974 [25] 姬莉霞, 李学相.基于相邻帧补偿的高速运动目标图像稳像算法及仿真.计算机科学, 2014, 41(7):310-312, 317 doi: 10.11896/j.issn.1002-137X.2014.07.064Ji Li-Xia, Li Xue-Xiang. Algorithm and simulation of image stabilization for high speed moving target images based on adjacent frames compensation. Computer Science, 2014, 41(7):310-312, 317 doi: 10.11896/j.issn.1002-137X.2014.07.064 [26] Zhao T S, Wang J H, Wang Z, Chen C W. SSIM-based coarse-grain scalable video coding. IEEE Transactions on Broadcasting, 2015, 61(2):210-221 doi: 10.1109/TBC.2015.2424012 [27] Maligranda L. Some remarks on the triangle inequality for norms. Banach Journal of Mathematical Analysis, 2008, 2(2):31-41 doi: 10.15352/bjma/1240336290 [28] 方晟. 基于正则化的高倍加速并行磁共振成像技术[博士学位论文], 清华大学, 中国, 2010. http://cdmd.cnki.com.cn/Article/CDMD-10003-1011280424.htmFang Sheng. Regularization-based Parallel Magnetic Resonance Imaging Technique for Highly Accelerated Data Acquisition[Ph. D. dissertation], Tsinghua University, China, 2010. http://cdmd.cnki.com.cn/Article/CDMD-10003-1011280424.htm [29] Sun S J, Gill M, Li Y F, Huang M, Byrd R A. Efficient and generalized processing of multidimensional NUS NMR data:the NESTA algorithm and comparison of regularization terms. Journal of Biomolecular NMR, 2015, 62(1):105-117 doi: 10.1007/s10858-015-9923-x [30] Magarey J, Kingsbury N. Motion estimation using a complex-valued wavelet transform. IEEE Transactions on Signal Processing, 1998, 46(4):1069-1084 doi: 10.1109/78.668557 [31] Brox T, Bruhn A, Papenberg N, Weickert J. High accuracy optical flow estimation based on a theory for warping. In: Proceedings of the 8th European Conference on Computer Vision. Berlin Heidelberg, Germany: Springer, 2004. 25-36 [32] 王兵, 赵荣椿, 蒋晓悦, 俞鸿波.一种基于小波变换的运动估计方法.西北工业大学学报, 2004, 22(4):417-421 http://d.wanfangdata.com.cn/Periodical_xbgydxxb200404004.aspxWang Bing, Zhao Rong-Chun, Jiang Xiao-Yue, Yu Hong-Bo. Motion estimation based on wavelet transform. Journal of Northwestern Polytechnical University, 2004, 22(4):417-421 http://d.wanfangdata.com.cn/Periodical_xbgydxxb200404004.aspx [33] 李志清, 施智平, 李志欣, 史忠植.基于结构相似度的稀疏编码模型.软件学报, 2010, 21(10):2410-2419 https://www.researchgate.net/profile/Zhiping_Shi2/publication/220733094_Sparse_coding_model_based_on_structural_similarity/links/54e1afbb0cf2953c22bb0e90/Sparse-coding-model-based-on-structural-similarity.pdfLi Zhi-Qing, Shi Zhi-Ping, Li Zhi-Xin, Shi Zhong-Zhi. Sparse coding model based on structural similarity. Journal of Software, 2010, 21(10):2410-2419 https://www.researchgate.net/profile/Zhiping_Shi2/publication/220733094_Sparse_coding_model_based_on_structural_similarity/links/54e1afbb0cf2953c22bb0e90/Sparse-coding-model-based-on-structural-similarity.pdf [34] Ravishankar S, Bresler Y. MR image reconstruction from highly undersampled k-space data by dictionary learning. IEEE Transactions on Medical Imaging, 2011, 30(5):1028-1041 doi: 10.1109/TMI.2010.2090538 期刊类型引用(3)
1. 左国玉，徐兆坤，卢佳豪，龚道雄. 基于结构优化的DDAG-SVM上肢康复训练动作识别方法. 自动化学报. 2020(03): 549-561 . 
本站查看2. 徐勇. 电子胃镜运行状态异常智能告警方法研究. 自动化与仪器仪表. 2020(11): 186-189 . 百度学术3. 钱银中，沈一帆. 姿态特征与深度特征在图像动作识别中的混合应用. 自动化学报. 2019(03): 626-636 . 本站查看其他类型引用(13)
-
计量
- 文章访问数: 1993
- HTML全文浏览量: 255
- PDF下载量: 551
- 被引次数: 16
