基于结构张量和各向异性平滑的DTI去噪
|
摘要
扩散张量成像(Diffusion tensor image,DTI)是一种磁共振成像技术,可以提供白质纤维的走行等独特信息,且具有非侵入和不需要造影剂等优点,因此在理论研究和临床应用领域引起了极大的关注.然而在DTI成像过程中,由于受噪声的影响,导致获得的图像边缘信息模糊不清,给病灶的识别带来了难度.为了减少噪声对DTI图像的影响并且有效地保留边缘结构信息,通过结合结构张量和各向异性平滑技术提出一种新型的DTI去噪方法.首先利用结构张量将DTI图像中的像素分成均匀平坦区域和边缘轮廓区域,然后在均匀区域内进行各向同性滤波,而在边缘轮廓区域进行各向异性平滑处理,从而得到去噪后的DTI图像.实验结果表明,基于结构张量和各向异性平滑的DTI去噪方法明显降低了噪声的影响,同时有效地保留了图像的边缘结构信息.
Diffusion tensor image( DTI) is a magnetic resonance imaging technology that can provide unique information such as the white matter tractography,and has the advantages of non-invasive technique and requires no contrast medium. It is all the facts above have stirred great interest toward DTI in theory research and clinical application. However,in the procession of DTI imaging,due to the impact of noise,the edge information of the obtained image is blurred,which brings difficulty to the identification of diseases. In order to reduce the influence of noise on DTI image and preserve the edge information effectively,a newDTI denoising method is proposed by combining structure tensor and anisotropic smoothing technique. Firstly,the pixels of the DTI image are divided into homogeneous areas and edge regions by structure tensor. Then,the isotropic filtering is performed in the homogeneous region,and the anisotropic smoothing is performed in the edge region. Finally,the denoised DTI image is obtained. The experimental results showthat the DTI denoising method based on structure tensor and anisotropic smoothing can reduce the influence of noise significantly,and preserve the edge structure information of the DTI image effectively.
Diffusion tensor image( DTI) is a magnetic resonance imaging technology that can provide unique information such as the white matter tractography,and has the advantages of non-invasive technique and requires no contrast medium. It is all the facts above have stirred great interest toward DTI in theory research and clinical application. However,in the procession of DTI imaging,due to the impact of noise,the edge information of the obtained image is blurred,which brings difficulty to the identification of diseases. In order to reduce the influence of noise on DTI image and preserve the edge information effectively,a newDTI denoising method is proposed by combining structure tensor and anisotropic smoothing technique. Firstly,the pixels of the DTI image are divided into homogeneous areas and edge regions by structure tensor. Then,the isotropic filtering is performed in the homogeneous region,and the anisotropic smoothing is performed in the edge region. Finally,the denoised DTI image is obtained. The experimental results showthat the DTI denoising method based on structure tensor and anisotropic smoothing can reduce the influence of noise significantly,and preserve the edge structure information of the DTI image effectively.
引文
[1]Le Bihan D,Mangin J F,Poupon C,et al.Diffusion tensor imaging:concepts and applications[J].M agnetic Resonance Imaging,2001,13(2):534-546.
[2]Parker G J,Schnabel J A,Symms M R,et al.Non-linear smoothing for reduction of systematic and random errors in diffusion tensor imaging[J].M agnetic Resonance Imaging,2000,11(3):702-710.
[3]Pajevic S,Aldroubi A,Basser P J.A continuous tensor field approximation of discrete DT-M RI data for extracting micro-structural and architectural features of tissue[J].M agnetic Resonance in M edicine,2002,154(2):85-100.
[4]Poupon C,Clark C A,Frouin V,et al.Regularization of diffusion based direction maps for the tracking of brain w hite matter fascicles[J].Neuro Image,2000,12(4):184-195.
[5]Tench C R,Morgan P S,Blumhardt L D,et al.Improved white matter fiber tracking using stochastic labeling[J].M agnetic Resonance in M edicine,2002,48(3):677-683.
[6]Bigun J,Gramlund G H,Wilklund J.Multidimensional orientation estimation w ith application to texture analysis and optic flow[J].IEEE Trans on Pattern Analysis and M achine Intelligence,2001,13(8):775-790.
[7]Brox T,Weickert J,Burget B.Nonlinear structure tensor[J].Image and Vision Computing,2006,24(1):41-45.
[8]Chen B,Hsu E.Noise removal in magnetic resonance diffusion tensor imaging[J].Magnetic Resonance in Medicine,2005,54(2):393-407.
[9]Wang Z,Vemuri B C,Chen Y,et al.A constrained variational principle for direct estimation and smoothing of the diffusion tensor field from complex DWI[J].IEEE Transactions on M edical Imagihg,2004,23(8):930-939.
[10]Pennec X,Fillard P,Ayache N.A riemannian framework for tensor computing[J].Comput Vision,2006,66(1):41-66.
[11]Liu S,Liu M,Li P,et al.SAR image denoising via sparse representation in shearlet domain based on continuous cycle spinning[J].IEEE Transactions on Geoscience&Remote Sensing,2017,55(5):2985-2992.
[12]Wang Jian,Liang Bin,Xue Fei.Method of underwater acoustic signal denoising based on characteristics of empirical mode decomposition of w hite noise[J].Journal of Ordnance Equipment Engineering,2015,36(6):131-134.
[13]Zheng Yu-hui,Pan Yu,Wang Ping-an.A nonlinear tensor based on trace[J].Journal of Computer-Aided Design and Graphics,2008,20(2):259-266.
[14]Haussecker H,Spies H.Handbook of computer vision and applications[M].Beijing:Academic Press,1999.
[15]Durand F,Dorsey J.Fast bilateral filtering for the display of high dynamic range images[J].ACM Transactions on Graphics,2002,21(3):257-266.
[16]Cohen L.The history of noise[On the 100th anniversary of its birth][J].IEEE Singal Processing Magazine,2005,22(6):20-45.
[17]Liu S Q,Li P F,Liu M,et al.DTI image denoising based on complex shearlet domain and complex diffusion anisotropic filtering[C].The 6th International Conference on Communications,Signal Processing,and Systems(CSPS),Harbin,China,2017:267-273.
[18]Qing X,Anderson A W,Gore J C,et al.Diffusion tensor image smoothing using efficient and effective anisotropic filtering[C].IEEE 11th International Conference on Computer Vision(ICCV),Rio de Janeiro,Brazil,2007:1-6.
[12]王剑,梁斌,薛飞.一种基于白噪声EMD分解统计特性的水声信号去噪方法[J].兵器装备工程学报,2015,36(6):131-134.
[13]郑钰辉,潘瑜,王平安.基于迹的非线性结构张量[J].计算机辅助设计与图形学学报,2008,20(2):259-266.
[2]Parker G J,Schnabel J A,Symms M R,et al.Non-linear smoothing for reduction of systematic and random errors in diffusion tensor imaging[J].M agnetic Resonance Imaging,2000,11(3):702-710.
[3]Pajevic S,Aldroubi A,Basser P J.A continuous tensor field approximation of discrete DT-M RI data for extracting micro-structural and architectural features of tissue[J].M agnetic Resonance in M edicine,2002,154(2):85-100.
[4]Poupon C,Clark C A,Frouin V,et al.Regularization of diffusion based direction maps for the tracking of brain w hite matter fascicles[J].Neuro Image,2000,12(4):184-195.
[5]Tench C R,Morgan P S,Blumhardt L D,et al.Improved white matter fiber tracking using stochastic labeling[J].M agnetic Resonance in M edicine,2002,48(3):677-683.
[6]Bigun J,Gramlund G H,Wilklund J.Multidimensional orientation estimation w ith application to texture analysis and optic flow[J].IEEE Trans on Pattern Analysis and M achine Intelligence,2001,13(8):775-790.
[7]Brox T,Weickert J,Burget B.Nonlinear structure tensor[J].Image and Vision Computing,2006,24(1):41-45.
[8]Chen B,Hsu E.Noise removal in magnetic resonance diffusion tensor imaging[J].Magnetic Resonance in Medicine,2005,54(2):393-407.
[9]Wang Z,Vemuri B C,Chen Y,et al.A constrained variational principle for direct estimation and smoothing of the diffusion tensor field from complex DWI[J].IEEE Transactions on M edical Imagihg,2004,23(8):930-939.
[10]Pennec X,Fillard P,Ayache N.A riemannian framework for tensor computing[J].Comput Vision,2006,66(1):41-66.
[11]Liu S,Liu M,Li P,et al.SAR image denoising via sparse representation in shearlet domain based on continuous cycle spinning[J].IEEE Transactions on Geoscience&Remote Sensing,2017,55(5):2985-2992.
[12]Wang Jian,Liang Bin,Xue Fei.Method of underwater acoustic signal denoising based on characteristics of empirical mode decomposition of w hite noise[J].Journal of Ordnance Equipment Engineering,2015,36(6):131-134.
[13]Zheng Yu-hui,Pan Yu,Wang Ping-an.A nonlinear tensor based on trace[J].Journal of Computer-Aided Design and Graphics,2008,20(2):259-266.
[14]Haussecker H,Spies H.Handbook of computer vision and applications[M].Beijing:Academic Press,1999.
[15]Durand F,Dorsey J.Fast bilateral filtering for the display of high dynamic range images[J].ACM Transactions on Graphics,2002,21(3):257-266.
[16]Cohen L.The history of noise[On the 100th anniversary of its birth][J].IEEE Singal Processing Magazine,2005,22(6):20-45.
[17]Liu S Q,Li P F,Liu M,et al.DTI image denoising based on complex shearlet domain and complex diffusion anisotropic filtering[C].The 6th International Conference on Communications,Signal Processing,and Systems(CSPS),Harbin,China,2017:267-273.
[18]Qing X,Anderson A W,Gore J C,et al.Diffusion tensor image smoothing using efficient and effective anisotropic filtering[C].IEEE 11th International Conference on Computer Vision(ICCV),Rio de Janeiro,Brazil,2007:1-6.
[12]王剑,梁斌,薛飞.一种基于白噪声EMD分解统计特性的水声信号去噪方法[J].兵器装备工程学报,2015,36(6):131-134.
[13]郑钰辉,潘瑜,王平安.基于迹的非线性结构张量[J].计算机辅助设计与图形学学报,2008,20(2):259-266.