磁共振图像处理算法的研究
|
摘要
磁共振成像技术由于具有成像参数多、分辨率高、无电离辐射损伤、可任意层面断层成像等特点而在医学上得到广泛的应用。但是,磁共振图像也具有成像时间长、软组织对比度不高、成像有噪声、有伪影等缺陷,这些缺陷严重阻碍了磁共振成像技术的推广和发展。为了克服磁共振图像的这些缺陷,本文从磁共振图像的增强和伪影校正这两个方面来展开对磁共振图像处理算法的研究。
在磁共振图像的增强方面,本文对基于尺度相关性的阈值处理方法进行改进,实现对磁共振图像的增强。考虑到在同一幅图像中,图像的去噪、增强这两种操作会相互影响,造成图像的质量下降,基于磁共振图像的特点,本文对图像像素点按相似度区别对待,对当前基于小波多尺度积的图像增强算法进行改进,采用一种基于尺度相关性的阈值处理方法来对磁共振图像进行增强操作。实验证明,该方法能够在有效增强图像的同时,抑制甚至消除图像中的噪声。
在磁共振图像伪影校正算法的研究上,本文将运动伪影分为平移运动伪影和旋转运动伪影两种,并分别进行研究。在平移运动伪影的校正方面:采用一种基于能量约束的方法,实现磁共振图像平移运动伪影的校正。在深入分析磁共振图像特点的基础上,提出基于能量约束的平移运动伪影校正算法,选取方向信息测度和磁共振图像非目标区0值像素点个数的组合作为能量约束函数。通过与目前研究较为成熟的自动逆向迭代校正算法进行对比实验,可知,本文算法在平移运动伪影的处理上,具有一定的优势。在旋转运动伪影校正方面:采用基于稀疏采样成像的方法,实现磁共振图像快速序列旋转运动伪影校正。考虑到磁共振图像旋转运动伪影的K空间数据被校正后,会出现数据分布不均匀的情况,若使用插值方法对数据空间进行填充,因填充方法的不同,会导致图像的重建效果不同,因此,本文选用稀疏采样成像方法。K空间数据具有变换稀疏性,可使用小波变换对K空间数据进行变换,得到具有稀疏性的数据空间,再进行稀疏采样成像。实验证明,本方法能够有效地校正磁共振图像的旋转运动伪影,且图像重建效果较好。
Magnetic resonance imaging technique is applied widely because this technology has lots of features such as multi-imaging parameters, high resolution, non-invasive and arbitrary slice imaging, etc. However, this technology also has some defects such as required a long time during imaging, the images of soft tissue are low contrast, and lots of images have noise or artifacts. The spread and development of the magnetic resonance imaging technique are obstructed by these defects. In order to conquer those defects, this paper provides related processing algorithms on image enhancement and artifact correction.
Firstly, in enhancement aspect, this paper improved a threshold processing algorithm based on scale dependencies to enhance the image. Consider in the same image, this paper considering that these two operations of image noise inhibition and enhance will affect each other, and lead to the image quality decline. Based on the characters of magnetic resonance image, according to the similarity between pixels this paper adopts a method that deal with the pixels differently, improve the enhancement algorithm on the basis of wavelet multi-scale product, using a threshold processing method based on the scale correlation to enhance the magnetic resonance image. The experiments show that this method can enhance the image effectively, and can inhibit and even eliminate the noise of the image at the same time.
And then, two motion artifacts including translational motion artifact and rotational motion artifact are studied respectively. In the aspect of translational motion artifact, this paper adopts a method based on energy constraint to correct the artifact. On the basis of character analysis of the magnetic resonance image, this paper raised a algorithm based on energy constraint. This algorithm chooses direction information measure and the number of 0 pixels in the target zone of the magnetic resonance image as the energy function. Compare to the automatic reverse iterative algorithm, experiments show that this algorithm has certain advantages on correction of translational motion artifact. In the aspect of rotational motion artifact correction, this paper using a method based on sparse sampling to correct the rotating motion artifact of fast sequential magnetic resonance. Considering that after the correction of the K-space data, the data distribution will be uneven. Interpolation methods are often used to fill the data space, but the effect of the reconstruct image will change with different methods. Therefore, this paper selects sparse sampling imaging methods. K-space data has sparseness after transformation, in order to achieve sparse data space, we can use wavelet transform to transform the K-space data, then use the sparse sampling imaging. Experiments have proved that this method can correct the rotating motion artifact of magnetic resonance imaging effectively, and the effect of image reconstruction is good.
在磁共振图像的增强方面,本文对基于尺度相关性的阈值处理方法进行改进,实现对磁共振图像的增强。考虑到在同一幅图像中,图像的去噪、增强这两种操作会相互影响,造成图像的质量下降,基于磁共振图像的特点,本文对图像像素点按相似度区别对待,对当前基于小波多尺度积的图像增强算法进行改进,采用一种基于尺度相关性的阈值处理方法来对磁共振图像进行增强操作。实验证明,该方法能够在有效增强图像的同时,抑制甚至消除图像中的噪声。
在磁共振图像伪影校正算法的研究上,本文将运动伪影分为平移运动伪影和旋转运动伪影两种,并分别进行研究。在平移运动伪影的校正方面:采用一种基于能量约束的方法,实现磁共振图像平移运动伪影的校正。在深入分析磁共振图像特点的基础上,提出基于能量约束的平移运动伪影校正算法,选取方向信息测度和磁共振图像非目标区0值像素点个数的组合作为能量约束函数。通过与目前研究较为成熟的自动逆向迭代校正算法进行对比实验,可知,本文算法在平移运动伪影的处理上,具有一定的优势。在旋转运动伪影校正方面:采用基于稀疏采样成像的方法,实现磁共振图像快速序列旋转运动伪影校正。考虑到磁共振图像旋转运动伪影的K空间数据被校正后,会出现数据分布不均匀的情况,若使用插值方法对数据空间进行填充,因填充方法的不同,会导致图像的重建效果不同,因此,本文选用稀疏采样成像方法。K空间数据具有变换稀疏性,可使用小波变换对K空间数据进行变换,得到具有稀疏性的数据空间,再进行稀疏采样成像。实验证明,本方法能够有效地校正磁共振图像的旋转运动伪影,且图像重建效果较好。
Magnetic resonance imaging technique is applied widely because this technology has lots of features such as multi-imaging parameters, high resolution, non-invasive and arbitrary slice imaging, etc. However, this technology also has some defects such as required a long time during imaging, the images of soft tissue are low contrast, and lots of images have noise or artifacts. The spread and development of the magnetic resonance imaging technique are obstructed by these defects. In order to conquer those defects, this paper provides related processing algorithms on image enhancement and artifact correction.
Firstly, in enhancement aspect, this paper improved a threshold processing algorithm based on scale dependencies to enhance the image. Consider in the same image, this paper considering that these two operations of image noise inhibition and enhance will affect each other, and lead to the image quality decline. Based on the characters of magnetic resonance image, according to the similarity between pixels this paper adopts a method that deal with the pixels differently, improve the enhancement algorithm on the basis of wavelet multi-scale product, using a threshold processing method based on the scale correlation to enhance the magnetic resonance image. The experiments show that this method can enhance the image effectively, and can inhibit and even eliminate the noise of the image at the same time.
And then, two motion artifacts including translational motion artifact and rotational motion artifact are studied respectively. In the aspect of translational motion artifact, this paper adopts a method based on energy constraint to correct the artifact. On the basis of character analysis of the magnetic resonance image, this paper raised a algorithm based on energy constraint. This algorithm chooses direction information measure and the number of 0 pixels in the target zone of the magnetic resonance image as the energy function. Compare to the automatic reverse iterative algorithm, experiments show that this algorithm has certain advantages on correction of translational motion artifact. In the aspect of rotational motion artifact correction, this paper using a method based on sparse sampling to correct the rotating motion artifact of fast sequential magnetic resonance. Considering that after the correction of the K-space data, the data distribution will be uneven. Interpolation methods are often used to fill the data space, but the effect of the reconstruct image will change with different methods. Therefore, this paper selects sparse sampling imaging methods. K-space data has sparseness after transformation, in order to achieve sparse data space, we can use wavelet transform to transform the K-space data, then use the sparse sampling imaging. Experiments have proved that this method can correct the rotating motion artifact of magnetic resonance imaging effectively, and the effect of image reconstruction is good.
引文
[1]黄其流.临床磁共振成像[M].北京:人民军医出版社.991:274.
[2]赵喜平.磁共振成像系统的原理及应用[M].北京:科学出版社.2000:675.
[3]冉启文,谭立英.小波分析与分数傅立叶变换及应用[M].北京:国防工业出版社.2002:15-61.
[4] LI Hong, WANG Hui-Nan, ZHANG Xiao-Guo. Denoising and enhancement in intravascular ultrasound images via multiscale analysis[J].Journal of Sichuan University(Natural Science Edition).Oct.2008,45,5:1105-1110.
[5] T.Abdukirim, S.Takano and K.Niijim. A Construction of spline dyadic wavelet filters[R].Research Report on Information Science and Electrical Engineering of Kyushu University, 2002, 7(1):1-6.
[6]冯前进,陈武凡.基于模糊增强的磁共振成像PROPELLER采样数据优质重建算法[J].南方医科大学学报.2007,27(5):618-620.
[7]冯衍秋.基于PROPELLER采样的磁共振成像运动伪影消除方法研究[D].广州第一军医大学硕士学位论文.2005:10-21.
[8] W.A. Edelstein, P.A. Bottomley, and L.M.Pfeifer, A signal-to-noise calibration procedure for NMR imaging systems. Med.Phys.,1984,11:180-185.
[9] W.A. Edelstein,Golver G,Hardy C. The intrinsic signal-to-noise ratio in NMR imaging. Magnetic Resonance in Medicine,1986,3:604-618.
[10]阮秋琦.数字图像处理.北京:电子工业出版社.2001,183-192。
[11]何宏,唐志航,张细政,杨保安.基于小波多尺度积的图像增强新算法[J].计算机应用与软件.2007,3:163-165.
[12] T.Abdukirim, S.Takano and K.Niijim. A Construction of spline dyadic wavelet filters[R].Research Report on Information Science and Electrical Engineering of Kyushu University, 2008, 7(1):1-6.
[13]郝征科,魏明果.一种基于模糊集的医学图像增强方法[J].计算机与数字工程.2007年35(3):117-118
[14] Mallat, S..A Theory for Multi-resolution Approximation: the Wavelet Approximation. IEEE Trans. PAMI 11 (1989):674-693.
[15] Mallat.S..A Wavelet Tour of Signal Processing.Academic Press, San Diego, 1998.
[16]程正兴.小波分析算法与应用.西安交通大学出版社.1998.
[17]唐晓初.小波分析及其应用.重庆大学出版社.2006.
[18]王新年,梁得群.基于方向信息梯度的图像模糊测度[J].计算机工程与应用.2005,32:45-49.
[19]方青.磁共振部分K空间数据图像重构[D].上海交通大学硕士学位论文.2006.
[20] Keiko Kondo, Miki Haseyama, and Hideo Kitujima. An efficient phase retrieval method using snakes foe image reconstruction[J]. International Conference on Image Processing(ICIP),2004, 2427 -2430.
[21] Lin W,Wehrli F W,Song H K.Correcting bulk in plane motion artifacts in MRI using the point spread function[J].IEEE Transactions on Medical Imaging,2005,24(9):1170-1176.
[22] Hou Zhengsong, Jiang Guiping, Wang Huafeng.An inverse iterative correction algorithm foer rigid translational motion artifacts in MRI[J].Chinese Journal of Bionedical Engineering, 2004, 23(4):353-358.
[23]谭斐,周荷琴等.MRI快速序列的旋转伪影校正[J].计算机辅助设计与图形学学报.2009, 21 (2):143-148.
[24] Michael Lusting, David Donoho, and John M. Pauly. Sparse MRI: The Application of Compressed Sensing for Rapid MR Imaging[J]. Magnetic Resonance in Medicine, Vol.58(6):1182-1195,2007
[25] Sylvain Durand,Jacques Froment.Reconstruction of wavelet coefficients using total variation minimization[J].SLAM J. Sci. Comput. Vol.24:1754-1767,2003.
[26] Huang Min,Guan Jin'an,Huang Li.Comparison of gridding algorithms used in spiral magnetic resonance imaging[J]. Chinese Journal of Magnetic Resonance,2006,23(3):302-311(in Chinese).
[27] Antonin Chambolle. An algorithm for totel variation minimization and applications[J]. Math. Imaging Vis.Vol.20:89-97,2004.
[28] Henry Stark. Theory of convex projection and its application to image restoration[J].Circuits and Systems, Vol.1:936-964,1988.
[29]陈春晓,童超,王世杰,罗立民.MRI刚性平移运动模糊图像建模与恢复[J].中国生物医学工程学报.2007,26(3):368-373.
[30] Lin W, Wehrli FW, Song HK.Correcting bulk in-plane motion artifacts in MRI using the point spread fuction[J]. Medical Imaging, IEEE Transactions on ,24(9):1170-1176, 2005.
[31] Atkinson D, Hill D LG, Stoyle PNR, et al. Automatic correction of motion artifacts in magnetic resonance images using an entropy focus criterion[J].IEEE Transactions on Medical Imaging, 16(6):903-910,1997.
[32]王宏志,张学龙,武杰.核磁共振成像技术实验教程[M].北京:科学出版社,2008,4-6.
[33]陈渊博,陈春晓,周卉芬.c磁共振成像中伪影的研究[J].国际生物医学工程杂志,2006,29(4):214-218.
[34]刘正敏,武海澄,周荷琴.一种基于参考扫描的MRI相位伪影校正方法[J].2007,22 (3):282-286.
[35]黄鑫,陈武凡,卢振泰,冯衍秋.一种并行磁共振成像伪影消除方法[J].电路与系统学报.2008,13(1):18:22.
[36]刘军伟,李传富,吴欢等.改进的熵最小方法用于MRI偏差场的校正[J].中国生物医学工程学报.2008,27(6):867-873.
[37] J.A.Troop and A.C.Gilbert,Signal recovery from random measurements via orthogonal matching pursuit[J].IEEE Trans. Inf. Theory,2007(53)12:4655-4666.
[38]严朝福,李德玉等.超声医学图像非线性滤波算法研究进展[J].生物医学工程学杂志.2006,23(5):1123-1125.
[39]钱惟贤,陈钱,顾国华,管志强.采用方向信息测度和时间迭代技术的红外图像扩散去噪算法[J].红外与激光工程.2010,39(3):561-566.
[40]刘袁缘,严国萍,潘晴,等.一种基于能量差比较算法的差分运动检测方法的改进[J].信号处理.2007,23(6):891-895.
[41]李学玲,齐国清.基于方向信息测度的非线性扩散图像去噪方法[J].中国图像图像学报.2008,13(3):400-405.
[42]焦李成,侯彪,王爽,等.图像多尺度几何分析理论与应用-后小波理论分析与应用[M].西安,西安电子科技大学出版社.2008:3-25.
[43] Atkinson D,Hill D LG,Stoyle PNR, et al.Automatic correction of motion artifacts in magnetic resonance images using an entropy focus criterion[J].IEEE Transactions on Medical Imaging.1997,16(6):903-910.
[44]Castellani,U & M.Cristani & C.Combi & V.Murino & A.Sbarbati & P.Marzola.Visual MRI: Merging information visualization and non-parametric clustering techniques for MRI dataset analysis[J].Artificial Intelligence in Medicine(2008)44,183-199.
[45]Phan T.H. & Md.A.U.Khan. Vessel enhancement filter using directional filter bank[J].Computer Vision an Image Umderstanding 113(2009)101-112.
[46]Chandra S.Y. & S.R. Kothapalli & D.V.G.L.N.Rao.Optical Fourier techniques for medical image processing and pahase contrast imaging[J]. Optical Communications 281(2008)1876-1888.
[47]Anupam Singhal.AssessingMNear-Infrared Quanntum Dots for Deep Tissue,Orgam, and AnimalImaging Applications[R]. Original Report. February 2008,1-3.
[48]Y. Li, et al., Denoising by using multineural networks for medical X-ray imaging applications,Neurocomputing (2009), doi:10.1016/j.neucom.2008.07.019.
[49]X.Lv & X.Gao & H.Zou.Interactive curved planar reformation based on snake model[J].Computerized Medical Imaging and Graphics 32 (2008) 662–669.
[50] J. Yu, J. Tan, Object density-based image segmentation and its applications in biomedical image analysis,Comput. Methods Programs Biomed. (2009), doi:10.1016/j.cmpb.2009.04.013.
[2]赵喜平.磁共振成像系统的原理及应用[M].北京:科学出版社.2000:675.
[3]冉启文,谭立英.小波分析与分数傅立叶变换及应用[M].北京:国防工业出版社.2002:15-61.
[4] LI Hong, WANG Hui-Nan, ZHANG Xiao-Guo. Denoising and enhancement in intravascular ultrasound images via multiscale analysis[J].Journal of Sichuan University(Natural Science Edition).Oct.2008,45,5:1105-1110.
[5] T.Abdukirim, S.Takano and K.Niijim. A Construction of spline dyadic wavelet filters[R].Research Report on Information Science and Electrical Engineering of Kyushu University, 2002, 7(1):1-6.
[6]冯前进,陈武凡.基于模糊增强的磁共振成像PROPELLER采样数据优质重建算法[J].南方医科大学学报.2007,27(5):618-620.
[7]冯衍秋.基于PROPELLER采样的磁共振成像运动伪影消除方法研究[D].广州第一军医大学硕士学位论文.2005:10-21.
[8] W.A. Edelstein, P.A. Bottomley, and L.M.Pfeifer, A signal-to-noise calibration procedure for NMR imaging systems. Med.Phys.,1984,11:180-185.
[9] W.A. Edelstein,Golver G,Hardy C. The intrinsic signal-to-noise ratio in NMR imaging. Magnetic Resonance in Medicine,1986,3:604-618.
[10]阮秋琦.数字图像处理.北京:电子工业出版社.2001,183-192。
[11]何宏,唐志航,张细政,杨保安.基于小波多尺度积的图像增强新算法[J].计算机应用与软件.2007,3:163-165.
[12] T.Abdukirim, S.Takano and K.Niijim. A Construction of spline dyadic wavelet filters[R].Research Report on Information Science and Electrical Engineering of Kyushu University, 2008, 7(1):1-6.
[13]郝征科,魏明果.一种基于模糊集的医学图像增强方法[J].计算机与数字工程.2007年35(3):117-118
[14] Mallat, S..A Theory for Multi-resolution Approximation: the Wavelet Approximation. IEEE Trans. PAMI 11 (1989):674-693.
[15] Mallat.S..A Wavelet Tour of Signal Processing.Academic Press, San Diego, 1998.
[16]程正兴.小波分析算法与应用.西安交通大学出版社.1998.
[17]唐晓初.小波分析及其应用.重庆大学出版社.2006.
[18]王新年,梁得群.基于方向信息梯度的图像模糊测度[J].计算机工程与应用.2005,32:45-49.
[19]方青.磁共振部分K空间数据图像重构[D].上海交通大学硕士学位论文.2006.
[20] Keiko Kondo, Miki Haseyama, and Hideo Kitujima. An efficient phase retrieval method using snakes foe image reconstruction[J]. International Conference on Image Processing(ICIP),2004, 2427 -2430.
[21] Lin W,Wehrli F W,Song H K.Correcting bulk in plane motion artifacts in MRI using the point spread function[J].IEEE Transactions on Medical Imaging,2005,24(9):1170-1176.
[22] Hou Zhengsong, Jiang Guiping, Wang Huafeng.An inverse iterative correction algorithm foer rigid translational motion artifacts in MRI[J].Chinese Journal of Bionedical Engineering, 2004, 23(4):353-358.
[23]谭斐,周荷琴等.MRI快速序列的旋转伪影校正[J].计算机辅助设计与图形学学报.2009, 21 (2):143-148.
[24] Michael Lusting, David Donoho, and John M. Pauly. Sparse MRI: The Application of Compressed Sensing for Rapid MR Imaging[J]. Magnetic Resonance in Medicine, Vol.58(6):1182-1195,2007
[25] Sylvain Durand,Jacques Froment.Reconstruction of wavelet coefficients using total variation minimization[J].SLAM J. Sci. Comput. Vol.24:1754-1767,2003.
[26] Huang Min,Guan Jin'an,Huang Li.Comparison of gridding algorithms used in spiral magnetic resonance imaging[J]. Chinese Journal of Magnetic Resonance,2006,23(3):302-311(in Chinese).
[27] Antonin Chambolle. An algorithm for totel variation minimization and applications[J]. Math. Imaging Vis.Vol.20:89-97,2004.
[28] Henry Stark. Theory of convex projection and its application to image restoration[J].Circuits and Systems, Vol.1:936-964,1988.
[29]陈春晓,童超,王世杰,罗立民.MRI刚性平移运动模糊图像建模与恢复[J].中国生物医学工程学报.2007,26(3):368-373.
[30] Lin W, Wehrli FW, Song HK.Correcting bulk in-plane motion artifacts in MRI using the point spread fuction[J]. Medical Imaging, IEEE Transactions on ,24(9):1170-1176, 2005.
[31] Atkinson D, Hill D LG, Stoyle PNR, et al. Automatic correction of motion artifacts in magnetic resonance images using an entropy focus criterion[J].IEEE Transactions on Medical Imaging, 16(6):903-910,1997.
[32]王宏志,张学龙,武杰.核磁共振成像技术实验教程[M].北京:科学出版社,2008,4-6.
[33]陈渊博,陈春晓,周卉芬.c磁共振成像中伪影的研究[J].国际生物医学工程杂志,2006,29(4):214-218.
[34]刘正敏,武海澄,周荷琴.一种基于参考扫描的MRI相位伪影校正方法[J].2007,22 (3):282-286.
[35]黄鑫,陈武凡,卢振泰,冯衍秋.一种并行磁共振成像伪影消除方法[J].电路与系统学报.2008,13(1):18:22.
[36]刘军伟,李传富,吴欢等.改进的熵最小方法用于MRI偏差场的校正[J].中国生物医学工程学报.2008,27(6):867-873.
[37] J.A.Troop and A.C.Gilbert,Signal recovery from random measurements via orthogonal matching pursuit[J].IEEE Trans. Inf. Theory,2007(53)12:4655-4666.
[38]严朝福,李德玉等.超声医学图像非线性滤波算法研究进展[J].生物医学工程学杂志.2006,23(5):1123-1125.
[39]钱惟贤,陈钱,顾国华,管志强.采用方向信息测度和时间迭代技术的红外图像扩散去噪算法[J].红外与激光工程.2010,39(3):561-566.
[40]刘袁缘,严国萍,潘晴,等.一种基于能量差比较算法的差分运动检测方法的改进[J].信号处理.2007,23(6):891-895.
[41]李学玲,齐国清.基于方向信息测度的非线性扩散图像去噪方法[J].中国图像图像学报.2008,13(3):400-405.
[42]焦李成,侯彪,王爽,等.图像多尺度几何分析理论与应用-后小波理论分析与应用[M].西安,西安电子科技大学出版社.2008:3-25.
[43] Atkinson D,Hill D LG,Stoyle PNR, et al.Automatic correction of motion artifacts in magnetic resonance images using an entropy focus criterion[J].IEEE Transactions on Medical Imaging.1997,16(6):903-910.
[44]Castellani,U & M.Cristani & C.Combi & V.Murino & A.Sbarbati & P.Marzola.Visual MRI: Merging information visualization and non-parametric clustering techniques for MRI dataset analysis[J].Artificial Intelligence in Medicine(2008)44,183-199.
[45]Phan T.H. & Md.A.U.Khan. Vessel enhancement filter using directional filter bank[J].Computer Vision an Image Umderstanding 113(2009)101-112.
[46]Chandra S.Y. & S.R. Kothapalli & D.V.G.L.N.Rao.Optical Fourier techniques for medical image processing and pahase contrast imaging[J]. Optical Communications 281(2008)1876-1888.
[47]Anupam Singhal.AssessingMNear-Infrared Quanntum Dots for Deep Tissue,Orgam, and AnimalImaging Applications[R]. Original Report. February 2008,1-3.
[48]Y. Li, et al., Denoising by using multineural networks for medical X-ray imaging applications,Neurocomputing (2009), doi:10.1016/j.neucom.2008.07.019.
[49]X.Lv & X.Gao & H.Zou.Interactive curved planar reformation based on snake model[J].Computerized Medical Imaging and Graphics 32 (2008) 662–669.
[50] J. Yu, J. Tan, Object density-based image segmentation and its applications in biomedical image analysis,Comput. Methods Programs Biomed. (2009), doi:10.1016/j.cmpb.2009.04.013.