WO2019153659A1 - 一种新型的非线性并行重建的磁共振成像方法、装置及介质 - Google Patents

一种新型的非线性并行重建的磁共振成像方法、装置及介质 Download PDF

Info

Publication number
WO2019153659A1
WO2019153659A1 PCT/CN2018/096782 CN2018096782W WO2019153659A1 WO 2019153659 A1 WO2019153659 A1 WO 2019153659A1 CN 2018096782 W CN2018096782 W CN 2018096782W WO 2019153659 A1 WO2019153659 A1 WO 2019153659A1
Authority
WO
WIPO (PCT)
Prior art keywords
data
linear
combination
item
coil
Prior art date
Application number
PCT/CN2018/096782
Other languages
English (en)
French (fr)
Inventor
王海峰
梁栋
贾森
刘新
郑海荣
Original Assignee
深圳先进技术研究院
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳先进技术研究院 filed Critical 深圳先进技术研究院
Priority to EP18882276.1A priority Critical patent/EP3751301A4/en
Priority to US16/473,806 priority patent/US11579229B2/en
Publication of WO2019153659A1 publication Critical patent/WO2019153659A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R33/00—Arrangements or instruments for measuring magnetic variables
    • G01R33/20—Arrangements or instruments for measuring magnetic variables involving magnetic resonance
    • G01R33/44—Arrangements or instruments for measuring magnetic variables involving magnetic resonance using nuclear magnetic resonance [NMR]
    • G01R33/48—NMR imaging systems
    • G01R33/54—Signal processing systems, e.g. using pulse sequences ; Generation or control of pulse sequences; Operator console
    • G01R33/56—Image enhancement or correction, e.g. subtraction or averaging techniques, e.g. improvement of signal-to-noise ratio and resolution
    • G01R33/561—Image enhancement or correction, e.g. subtraction or averaging techniques, e.g. improvement of signal-to-noise ratio and resolution by reduction of the scanning time, i.e. fast acquiring systems, e.g. using echo-planar pulse sequences
    • G01R33/5611—Parallel magnetic resonance imaging, e.g. sensitivity encoding [SENSE], simultaneous acquisition of spatial harmonics [SMASH], unaliasing by Fourier encoding of the overlaps using the temporal dimension [UNFOLD], k-t-broad-use linear acquisition speed-up technique [k-t-BLAST], k-t-SENSE
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R33/00—Arrangements or instruments for measuring magnetic variables
    • G01R33/20—Arrangements or instruments for measuring magnetic variables involving magnetic resonance
    • G01R33/44—Arrangements or instruments for measuring magnetic variables involving magnetic resonance using nuclear magnetic resonance [NMR]
    • G01R33/48—NMR imaging systems
    • G01R33/54—Signal processing systems, e.g. using pulse sequences ; Generation or control of pulse sequences; Operator console
    • G01R33/56—Image enhancement or correction, e.g. subtraction or averaging techniques, e.g. improvement of signal-to-noise ratio and resolution
    • G01R33/5608—Data processing and visualization specially adapted for MR, e.g. for feature analysis and pattern recognition on the basis of measured MR data, segmentation of measured MR data, edge contour detection on the basis of measured MR data, for enhancing measured MR data in terms of signal-to-noise ratio by means of noise filtering or apodization, for enhancing measured MR data in terms of resolution by means for deblurring, windowing, zero filling, or generation of gray-scaled images, colour-coded images or images displaying vectors instead of pixels
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R33/00—Arrangements or instruments for measuring magnetic variables
    • G01R33/20—Arrangements or instruments for measuring magnetic variables involving magnetic resonance
    • G01R33/44—Arrangements or instruments for measuring magnetic variables involving magnetic resonance using nuclear magnetic resonance [NMR]
    • G01R33/48—NMR imaging systems
    • G01R33/58—Calibration of imaging systems, e.g. using test probes, Phantoms; Calibration objects or fiducial markers such as active or passive RF coils surrounding an MR active material

Definitions

  • the present invention relates to the field of magnetic resonance image reconstruction, and in particular to a novel nonlinear parallel reconstruction magnetic resonance imaging method, apparatus and medium.
  • Image reconstruction is an important operation in magnetic resonance parallel imaging.
  • high-performance image reconstruction methods especially for parallel fast imaging, play a very important role.
  • image reconstruction methods in multi-channel coil parallel imaging such as: SENSE (Sensitivity encoded) method, GRAPP (Generalized auto calibrating partially parallel acquisitions) method, SMASH (Simultaneous acquisition) Of spatial harmonics, space harmonics, etc.
  • the GRAPPA method has a wide application convenience because it does not need a linear method for calculating the sensitivity of the multi-channel coil.
  • the nonlinear GRAPPA (Nonlinear GRAPPA) reconstruction method is limited by the number of channels of magnetic resonance data, so the reconstructed image quality is poor; based on Virtual Coil Conception (VCC)
  • VCC Virtual Coil Conception
  • the GRAPPA reconstruction method is not ideal for reducing artifacts and noise.
  • the technical problem to be solved by the present invention is to provide an image imaging method that improves the quality of reconstructed images and reduces artifacts and noise of reconstructed images.
  • the present invention firstly discloses a novel nonlinear parallel reconstruction magnetic resonance imaging method, apparatus and medium, and the technical solution is implemented as follows:
  • a novel nonlinear parallel reconstruction magnetic resonance imaging method comprising:
  • step S1 the actual coil data is collected in parallel by the multi-channel coil, and the virtual coil data of the same channel number is expanded according to the actual coil data, and the conjugate symmetry relationship is satisfied between the virtual coil data and the actual coil data;
  • Step S2 obtaining a data combination item according to the combination of the actual coil data and the virtual coil data
  • Step S3 using the completely sampled data in the low frequency region in the sampling space, and combining the data combination items to calibrate the weighting factor;
  • Step S4 reconstructing the missing data in the sampling space according to the calibrated weighting factor to obtain reconstruction data, where the missing data refers to data in the sampling space that is not collected in the high frequency region;
  • Step S5 integrating the completely sampled data in the low frequency region and the reconstructed data in the sampling space to obtain complete sampling space data
  • step S6 the final reconstructed image is obtained according to the multi-channel complete sampling space data.
  • the step S6 is: performing squared summation of the multi-channel complete sampling spatial data to obtain the final reconstructed image, or multiplying the multi-channel complete sampling spatial data by the sensitivity of the multi-channel coil Subsequent summation results in the final reconstructed image.
  • the data combination term comprises a constant, a linear data item and a non-linear data item.
  • the step S3 is to generate a mixed data combination item by using the completely sampled data in the low frequency region, and the data combination item includes a constant, a linear data item and a nonlinear data item.
  • the non-linear data item in the data combination item is a quadratic and above non-linear data item, and/or the non-linear data item in the mixed data combination item is a quadratic and above non-linear data item .
  • the calibration method in step S3 is to fit a set of weighting factors by linear regression to establish a linear combination relationship between the data in the sampling space and its adjacent data.
  • the formula involved in the calibration method is:
  • the present invention also discloses a computer readable medium having a program stored therein for a computer to perform the imaging method.
  • the invention also discloses a magnetic resonance image reconstruction device for using the imaging method
  • the utility model comprises an acquisition module, a data combination module, a calibration module, a reconstruction data module, a fusion module and an imaging module;
  • the acquisition module uses the multi-channel coil to collect the actual coil data in parallel, and expands the virtual coil data of the same channel number according to the actual coil data, and the virtual coil data and the actual coil data satisfy a conjugate symmetry relationship;
  • the data combination module obtains a data combination item according to the actual coil data and the virtual coil data combination
  • the calibration module uses the completely sampled data in the low frequency region in the sampling space, and combines the data combination term to calibrate the weighting factor;
  • the reconstructed data module reconstructs the missing data in the sampling space according to the calibrated weighting factor, where the missing data refers to data in the sampling space that is not collected in the high frequency region;
  • the fusion module combines the completely sampled data in the low frequency region and the reconstructed data in the sampling space to obtain complete sampling space data;
  • the imaging module obtains a final reconstructed image from the multi-channel complete sample space data.
  • the imaging module squares the multi-channel complete sampling space data to obtain the final reconstructed image, or multiplies the multi-channel complete sampling spatial data by the sensitivity of the multi-channel coil. And get the final reconstructed image.
  • the data combination item obtained by the data combination module comprises a constant, a linear data item and a non-linear data item;
  • the calibration module utilizes fully sampled data in the low frequency region to generate a mixed data combination term in combination with a data combination term comprising a constant, a linear mixed data item, and a non-linear data item.
  • the non-linear data item in the data combination item obtained by the calibration module is a quadratic and above non-linear data item, and/or the non-linear data item in the mixed data combination item is quadratic and The above non-linear data items;
  • the calibration module fits a set of weighting factors by linear regression to establish a linear combination relationship between the data in the sampling space and its adjacent data.
  • the calibration method used by the calibration module involves the following formula:
  • the invention improves the image quality of parallel magnetic resonance rapid imaging image reconstruction, and reduces reconstruction artifacts and noise;
  • the present invention does not add additional data and scan time while improving the quality of the reconstructed image.
  • FIG. 1 is a schematic flow chart of the VCC-NL-GRAPPA method
  • Figure 3 is a reconstructed image of the GRAPPA, NL-GRAPPA, and VCC-NL-GRAPPA methods at 5x acceleration (3x net acceleration);
  • Figure 4 is a graph showing the relationship between the mean square error and the external acceleration coefficient of the GRAPPA, NL-GRAPPA, and VCC-NL-GRAPPA methods.
  • the present invention has been made primarily to solve the problems of the corresponding prior art in the field of magnetic resonance image reconstruction, and therefore the present invention is particularly applicable to the subdivision, but does not mean that the present invention
  • the scope of application of the technical solution is thus limited, and those skilled in the art can reasonably implement various specific applications in the field of magnetic resonance imaging as needed.
  • a novel nonlinear parallel reconstruction magnetic resonance imaging method referring to FIG. 1, the steps of the imaging method include:
  • step S1 the actual coil data is collected in parallel by the multi-channel coil, and the virtual coil data of the same channel number is expanded according to the actual coil data, and the conjugate symmetry relationship is satisfied between the virtual coil data and the actual coil data;
  • Step S2 obtaining a data combination item according to the combination of the actual coil data and the virtual coil data
  • Step S3 using the completely sampled data in the low frequency region in the sampling space, and combining the data combination items to calibrate the weighting factor;
  • Step S4 reconstructing the missing data in the sampling space according to the calibrated weighting factor to obtain reconstruction data, where the missing data refers to data in the sampling space that is not collected in the high frequency region;
  • Step S5 integrating the completely sampled data in the low frequency region and the reconstructed data in the sampling space to obtain complete sampling space data
  • step S6 the final reconstructed image is obtained according to the multi-channel complete sampling space data.
  • the method disclosed herein utilizes complex conjugate symmetry to expand the number of channels of magnetic resonance data. Specifically, the conjugate transposition operation of the multi-channel complex data obtained by the actual sampling is performed to expand the virtual data of the same channel number without increasing the scanning time and collecting additional data, so that the subsequent reconstruction steps are used. Knowing that the amount of data is doubled, the signal-to-noise ratio and accuracy of subsequent reconstruction results are improved; in addition, when nonlinear data items are combined in image reconstruction, the reconstructed image quality is sufficiently improved, and the artifacts and noise of the reconstructed image are reduced.
  • VCC-NL-GRAPPA non-linear full automatic calibration partial parallel acquisition based on virtual conjugate coil technology
  • edge band also contains part of the high frequency data.
  • the step S6 is: performing square summation of the multi-channel complete sampling spatial data to obtain the final reconstructed image, or the multi-channel complete sampling spatial data and the multi-channel coil. The sensitivity is multiplied and summed to obtain the final reconstructed image.
  • said data combination term comprises a constant, a linear data item and a non-linear data item.
  • the exponential power of the nonlinear data item in the present invention may be two or three or more, and of course, any combination of two or more thereof may be used.
  • the step S3 is: generating a mixed data combination item by using the completely sampled data in the low frequency region, the mixed data combination item including the constant, the linear data item and the nonlinear data item. .
  • the non-linear data item in the data combination item is a quadratic and above non-linear data item, and/or the non-linear data item in the mixed data combination item is quadratic and The above non-linear data items.
  • the calibration method in step S3 is to fit a set of weighting factors by linear regression to establish a linear combination relationship between the data in the sampling space and its adjacent data.
  • the calibration method involves:
  • k is any position in the K space
  • l is the ordinal number of the channel
  • b is the distance between the data to be estimated and the actual sampled data along the under-sampling phase encoding gradient direction in the sampling space;
  • h is the distance between the data to be estimated and the actual sampled data along the full sampling frequency in the sampling space
  • S l is the sampling space data of the lth channel
  • S l * is a conjugate device matrix of S l ;
  • k x , k y are the coordinates of the sampled data points in the two-dimensional sampling space
  • ⁇ k x , ⁇ k y is the minimum interval that adjacent sampling points should satisfy when fully sampling in the sampling space
  • R acceleration multiplier, that is, undersampling multiple
  • the weighting factor in the GRAPPA imaging method can be obtained by calibrating the full sampled data in the low frequency region of the sampling space; where j is the channel number, r is the ordinal number of the channel displacement, and the first term in the superscript (2, 1) The number is the index and the second is the lateral distance, which is for Dot-multiply with a weighting factor with a lateral distance of 1;
  • B 1 , B 2 , H 1 , and H 2 are the number of weighting factors in the imaging method.
  • the present invention also discloses a computer readable medium having a program stored therein for a computer to perform the imaging method.
  • the invention also discloses a magnetic resonance image reconstruction device for using the imaging method
  • the utility model comprises an acquisition module, a data combination module, a calibration module, a reconstruction data module, a fusion module and an imaging module;
  • the acquisition module uses the multi-channel coil to collect the actual coil data in parallel, and expands the virtual coil data of the same channel number according to the actual coil data, and the virtual coil data and the actual coil data satisfy a conjugate symmetry relationship;
  • the data combination module obtains a data combination item according to the actual coil data and the virtual coil data combination
  • the calibration module uses the completely sampled data in the low frequency region in the sampling space, and combines the data combination term to calibrate the weighting factor;
  • the reconstructed data module reconstructs the missing data in the sampling space according to the calibrated weighting factor, where the missing data refers to data in the sampling space that is not collected in the high frequency region;
  • the fusion module combines the completely sampled data in the low frequency region and the reconstructed data in the sampling space to obtain complete sampling space data;
  • the imaging module obtains a final reconstructed image from the multi-channel complete sample space data.
  • the imaging module squares the multi-channel complete sample space data to obtain the final reconstructed image, or compares the multi-channel complete sample space data with multi-channel coils. The degrees are multiplied and summed to obtain the final reconstructed image.
  • the data combination item obtained by the data combination module comprises a constant, a linear data item and a non-linear data item; the calibration module utilizes the completely sampled data in the low frequency region, combined with the data combination item to generate A mixed data combination term comprising a constant, a linear data item, and a non-linear data item.
  • the nonlinear data item in the data combination obtained by the calibration module is a quadratic and above nonlinear data item, and/or nonlinear data in the mixed data combination item.
  • the term is a quadratic and above nonlinear data item; the calibration module uses a linear regression method to fit a set of weighting factors, thereby establishing a linear combination relationship between the data in the sampling space and its adjacent data.
  • the calibration method used by the calibration module involves the following formula:
  • k is any position in the K space
  • l is the ordinal number of the channel
  • b is the distance between the data to be estimated and the actual sampled data along the under-sampling phase encoding gradient direction in the sampling space;
  • h is the distance between the data to be estimated and the actual sampled data along the full sampling frequency in the sampling space
  • S l is the sampling space data of the lth channel
  • S l * is a conjugate device matrix of S l ;
  • k x , k y are the coordinates of the sampled data points in the two-dimensional sampling space
  • ⁇ k x , ⁇ k y is the minimum interval that adjacent sampling points should satisfy when fully sampling in the sampling space
  • R acceleration multiplier, that is, undersampling multiple
  • the weighting factor in the GRAPPA imaging method can be obtained by calibrating the full sampled data in the low frequency region of the sampling space; where j is the channel number, r is the ordinal number of the channel displacement, and the first term in the superscript (2, 1) The number is the index and the second is the lateral distance, which is for Dot-multiply with a weighting factor with a lateral distance of 1;
  • B 1 , B 2 , H 1 , and H 2 are the number of weighting factors in the imaging method.
  • Results 3 and 4 were obtained by simulation and experiment using the method disclosed in the present invention.
  • Figure 3 shows the full automatic calibration partial parallel acquisition (GRAPPA), nonlinear GRAPPA (NL-GRAPPA) and virtual coil nonlinear GRAPPA (VCC-NL-GRAPPA) method at 5 times acceleration (net acceleration 3 times) Reconstruction image comparison
  • Figure 4 is a plot of the mean square error and external acceleration coefficients for the GRAPPA, NL-GRAPPA, and VCC-NL-GRAPPA methods.
  • Diff. represents the difference between the reconstructed image and the reference image
  • ⁇ 5 indicates that the difference is magnified by 5 times
  • MSE indicates the mean square error
  • R indicates the external acceleration coefficient
  • NetR indicates the net acceleration coefficient.
  • the MSE reconstructed by the present invention has a reduced MSE compared to the GRAPPA method and the Nonlinear GRAPPA method, and the artifacts and noise in the image are reduced.

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • High Energy & Nuclear Physics (AREA)
  • Condensed Matter Physics & Semiconductors (AREA)
  • General Physics & Mathematics (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • General Health & Medical Sciences (AREA)
  • Radiology & Medical Imaging (AREA)
  • Health & Medical Sciences (AREA)
  • Signal Processing (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Magnetic Resonance Imaging Apparatus (AREA)

Abstract

一种基于多通道线圈数据的复数共轭对称性和非线性GRAPPA图像重建的并行快速成像方法、装置及介质。它主要分为五个步骤：从实际多通道线圈数据扩展出虚拟共轭线圈数据；从实际和虚拟多通道线圈数据组合出线性和非线性数据项；用混合的低频全采样数据(其边缘带有部分高频数据)校准线性和非线性数据项的加权因子；根据校准的加权因子重建出高频欠采样的数据；融合低频全采样的数据和高频重建的数据。其有益效果主要有：提高了并行磁共振快速成像图像重建的图像质量，减小了重建伪影和噪音；在提高重建的图像质量的同时，不增加额外的数据和扫描时间。

Description

一种新型的非线性并行重建的磁共振成像方法、装置及介质 技术领域
本发明涉及磁共振图像重建领域，特别涉及一种新型的非线性并行重建的磁共振成像方法、装置及介质。
背景技术
在磁共振并行成像中，图像重建是一个重要操作。在实际临床研究中，高性能的图像重建方法，尤其是实现并行快速成像有着非常重要的作用。目前,多通道线圈并行成像中的图像重建方法有很多,比如:SENSE(Sensitivity encoded，敏感度编码)法,GRAPPA(Generalized auto calibrating partially parallel acquisitions，全面自动校准部分并行采集)法,SMASH(Simultaneous acquisition of spatial harmonics，空间谐波)法等。其中，GRAPPA法因不需要计算多通道线圈敏感度的线性方法，而具有广泛的应用便利性。
基于GRAPPA法发展而来的方法中，非线性的GRAPPA(Nonlinear GRAPPA)重建方法受到磁共振数据的通道数量限制，所以重建图像质量较差；基于虚拟共轭线圈技术(Virtual Coil Conception，VCC)的GRAPPA重建方法则在削弱伪影和噪音方面的效果并不理想。
发明内容
本发明要解决的技术问题是提供一种提高重建图像质量，减小重建图像的伪影和噪音的图像成像方法。
为了解决上述技术问题，本发明首先披露了一种新型的非线性并行重建的磁共振成像方法、装置及介质，其技术方案是这样实施的：
一种新型的非线性并行重建的磁共振成像方法，所述成像方法的步骤包括：
步骤S1，用多通道线圈并行采集实际线圈数据，根据所述实际线圈数据扩展出相同通道数的虚拟线圈数据，所述虚拟线圈数据和所述实际线圈数据之间满足共轭对称关系；
步骤S2，根据所述实际线圈数据和所述虚拟线圈数据组合得到数据组合项；
步骤S3，利用采样空间中，低频区域内完全采样的数据，结合数据组合项校准加权因子；
步骤S4，根据校准的所述加权因子，重建采样空间中的缺失数据得到重建数据，所述缺失数据指采样空间中，高频区域没有采集的数据；
步骤S5，融合采样空间中所述低频区域内完全采样数据和所述重建数据得到完整的采样空间数据；
步骤S6，根据多通道完整的采样空间数据得到最终重建图像。
优选地，所述步骤S6为，将所述多通道完整的采样空间数据进行平方求和得到所述最终重建图像，或者将所述多通道完整的采样空间数据与多通道线圈的敏感度相乘后求和得到所述最终重建图像。
优选地，在所述步骤S2中，所述数据组合项包括常数、线性数据项和非线性数据项。
优选地，所述步骤S3为：利用低频区域内完全采样的数据，结合数据组合项产生混合数据组合项，所述混合数据组合项包括常数、线性数据项和非线性数据项。
优选地，所述数据组合项中的非线性数据项为二次及其以上非线性数据项，和/或所述混合数据组合项中的非线性数据项为二次及其以上非线性数据项。
优选地，所述步骤S3中的校准方法为，用线性回归法拟合出一组加权因子，进而建立采样空间中的数据与其相邻数据之间的一组线性组合关系。
优选地，所述步骤S3中，校准方法所涉及的公式为：
其中，j＝1,2,……2·L；
r≠b·R；
Figure PCTCN2018096782-appb-000002
本发明还公开了一种计算机可读介质，该计算机可读介质具有存储在其中的程序，该程序用于计算机执行所述的成像方法。
本发明还公开了一种用于使用所述成像方法的磁共振图像重建装置，
包括采集模块、数据组合模块、校准模块、重建数据模块、融合模块、成像模块；
所述采集模块利用多通道线圈并行采集实际线圈数据，根据所述实际线圈数据扩展出相同通道数的虚拟线圈数据，所述虚拟线圈数据和所述实际线圈数据之间满足共轭对称关系；
所述数据组合模块根据所述实际线圈数据和所述虚拟线圈数据组合得到数据组合项；
所述校准模块利用采样空间中，低频区域内完全采样的数据，结合数据组合项校准加权因子；
所述重建数据模块根据校准的所述加权因子，重建采样空间中的缺失数据得到重建数据，所述缺失数据指采样空间中，高频区域没有采集的数据；
所述融合模块融合采样空间中所述低频区域内完全采样数据和所述重建数据得到完整的采样空间数据；
所述成像模块根据多通道完整的采样空间数据得到最终重建图像。
优选地，所述成像模块将所述多通道完整的采样空间数据进行平方求和得到所述最终重建图像，或者将所述多通道完整的采样空间数据与多通道线圈的敏感度相乘后求和得到所述最终重建图像。
优选地，所述数据组合模块得到的所述数据组合项包括常数、线性数据项和非线性数据项；
所述校准模块利用低频区域内完全采样的数据，结合数据组合项产生混合数据组合项，所述混合数据组合项包括常数、线性混合数据项和非线性数据项。
优选地，所述校准模块得到的所述数据组合项中的非线性数据项为二次及其以上非线性数据项，和/或所述混合数据组合项中的非线性数据项为二次及其以上非线性数据项；
所述校准模块用线性回归法拟合出一组加权因子，进而建立采样空间中的数据与其相邻数据之间的一组线性组合关系。
优选地，所述校准模块使用的校准方法所涉及的公式为：
Figure PCTCN2018096782-appb-000003
其中，j＝1,2,……2·L；
r≠b·R；
Figure PCTCN2018096782-appb-000004
实施本发明的有益效果主要有：
1、本发明提高了并行磁共振快速成像图像重建的图像质量，减小了重建伪影和噪音；
2、本发明在提高重建的图像质量的同时，不增加额外的数据和扫描时间。
附图说明
为更好地理解本发明的技术方案，可参考下列的、用于对现有技术或实施例进行说明的附图。这些附图将对部分实施例或现有技术涉及的产品或方法进行简要的展示。这些附图的基本信息如下：
图1为所述VCC-NL-GRAPPA方法的流程示意图；
图2为所述VCC-NL-GRAPPA方法在采样空间中进行加速的欠采样方案；
图3为GRAPPA、NL-GRAPPA和VCC-NL-GRAPPA方法在5倍加速(净加速3倍)时候的重建图像；
图4为GRAPPA、NL-GRAPPA和VCC-NL-GRAPPA方法的均方差和外部加速系数的关系图。
具体实施方式
现在对本发明实施例中的技术方案或有益效果作进一步的展开描述，显然，所描述的实施例仅是本发明的部分实施方式，而并非全部。
需要指出的是，本发明创造的提出，主要是为了解决磁共振图像重建领域内，相应的现有技术存在的问题，所以本发明创造特别适用于该细分领域，但并非意味本发明创造的技术方案所可应用的范围因此受限，本领域技术人员可根据需要，在磁共振成像领域下的各种具体应用场合进行合理地实施。
一种新型的非线性并行重建的磁共振成像方法，参考图1，所述成像方法的步骤包括：
步骤S1，用多通道线圈并行采集实际线圈数据，根据所述实际线圈数据扩展出相同通道数的虚拟线圈数据，所述虚拟线圈数据和所述实际线圈数据之间满足共轭对称关系；
步骤S2，根据所述实际线圈数据和所述虚拟线圈数据组合得到数据组合项；
步骤S3，利用采样空间中，低频区域内完全采样的数据，结合数据组合项校准加权因子；
步骤S4，根据校准的所述加权因子，重建采样空间中的缺失数据得到重建数据，所述缺失数据指采样空间中，高频区域没有采集的数据；
步骤S5，融合采样空间中所述低频区域内完全采样数据和所述重建数据得到完整的采样空间数据；
步骤S6，根据多通道完整的采样空间数据得到最终重建图像。
本发明所公开的方法，利用复数共轭对称性来扩充磁共振数据的通道数量。具体的为，在不增加扫描时间以及采集额外数据的情况下，对实际采样获得的多通道复数数据进行共轭转置操作扩展出相同通道数的虚拟数据，从而使得用于后续重建步骤的已知数据量加倍，提高后续重建结果的信噪比和准确性；此外，在图像重建时候组合出非线性数据项，则充分提高重建图像质量，减小重建图像的伪影和噪音。
VCC-NL-GRAPPA(基于虚拟共轭线圈技术的非线性全面自动校准部分并行采集)方法在采样空间中进行加速的欠采样的方案如图2所示。在方形的采样区域内，其中的黑色区域代表欠采样区域，灰色区域代表采样区域。
另外，需要注意的是，对于低频区域内完全采样的数据，在一些情况下其边缘带也包含有部分高频数据。
在一个优选的实施例中，所述步骤S6为，将所述多通道完整的采样空间数据进行平方求和得到所述最终重建图像，或者将所述多通道完整的采样空间数据与多通道线圈的敏感度相乘后求和得到所述最终重建图像。
在一个优选的实施例中，在所述步骤S2中，所述数据组合项包括常数、线性数据项和非线性数据项。
本发明中的非线性数据项的指数次幂可以是二次，也可以是三次以上，当然也可以是其中任意两种以上的组合。
在一个优选的实施例中，所述步骤S3为：利用低频区域内完全采样的数据，结合数据组合项产生混合数据组合项，所述混合数据组合项包括常数、线性数据项和非线性数据项。
在一个优选的实施例中，所述数据组合项中的非线性数据项为二次及其以上非线性数据项，和/或所述混合数据组合项中的非线性数据项为二次及其以上非线性数据项。
从实际的真实和虚拟的多通道线圈数据中，组合出常数、线性和非线性数据项。因为非线性数据项很多，为了在不过多的降低最后成像质量的条件下，加速运算速度，一般只截取二次非线性数据组合项，高于二次数据组合项在实施时候可以被忽略不计。
在一个优选的实施例中，所述步骤S3中的校准方法为，用线性回归法拟合出一组加权因子，进而建立采样空间中的数据与其相邻数据之间的一组线性组合关系。
为重建采样空间高频区域中缺失的采样数据，需要首先用采样空间低频区域内完全采样的部分数据进行校准，也即用线性回归法拟合出一组加权因子，建立起采样空间中数据和相邻数据之间的一组线性组合关系。
在一个优选的实施例中，所述步骤S3中，校准方法所涉及的公式为：
Figure PCTCN2018096782-appb-000005
其中，j＝1,2,……2·L；
r≠b·R；
Figure PCTCN2018096782-appb-000006
k为K空间中的任意位置；
l为通道的序数；
b为采样空间中沿欠采样相位编码梯度方向，待估计数据和实际采样数据之间的距离；
h为采样空间中沿全采样频率编码梯度方向，待估计数据和实际采样数据之间的距离；
S l为第l个通道的采样空间数据；
S l *为S l的共轭装置矩阵；
k x，k y为采样数据点在二维采样空间的坐标；
Δk x，Δk y为采样空间内完全采样时相邻采样点应满足的最小间隔；
R:加速倍数，也即欠采样倍数；
Figure PCTCN2018096782-appb-000007
为GRAPPA成像方法中的加权因子，可通过对采样空间低频区域中全采样数据进行校准获得；其中，j为通道序数，r为通道位移的序数，上标(2，1)内的第一项数字为指数，第二项为横向距离，也就是
Figure PCTCN2018096782-appb-000008
为
Figure PCTCN2018096782-appb-000009
与距其横向距离为1的加权因子进行点乘；
B 1，B 2，H 1，H 2为成像方法中的加权因子的个数。
本发明还公开了一种计算机可读介质，该计算机可读介质具有存储在其中的程序，该程序用于计算机执行所述的成像方法。
本发明还公开了一种用于使用所述成像方法的磁共振图像重建装置，
包括采集模块、数据组合模块、校准模块、重建数据模块、融合模块、成像模块；
所述采集模块利用多通道线圈并行采集实际线圈数据，根据所述实际线圈数据扩展出相同通道数的虚拟线圈数据，所述虚拟线圈数据和所述实际线圈数据之间满足共轭对称关系；
所述数据组合模块根据所述实际线圈数据和所述虚拟线圈数据组合得到数据组合项；
所述校准模块利用采样空间中，低频区域内完全采样的数据，结合数据组合项校准加权因子；
所述重建数据模块根据校准的所述加权因子，重建采样空间中的缺失数据得到重建数据，所述缺失数据指采样空间中，高频区域没有采集的数据；
所述融合模块融合采样空间中所述低频区域内完全采样数据和所述重建数据得到完整的采样空间数据；
所述成像模块根据多通道完整的采样空间数据得到最终重建图像。
在一个优选的实施例中，所述成像模块将所述多通道完整的采样空间数据进行平方求和得到所述最终重建图像，或者将所述多通道完整的采样空间数据与多通道线圈的敏感度相乘后求和得到所述最终重建图像。
在一个优选的实施例中，所述数据组合模块得到的所述数据组合项包括常数、线性数据项和非线性数据项；所述校准模块利用低频区域内完全采样的数据，结合数据组合项产生混合数据组合项，所述混合数据组合项包括常数、线性数据项和非线性数据项。
在一个优选的实施例中，所述校准模块得到的所述数据组合项中的非线性数据项为二次及其以上非线性数据项，和/或所述混合数据组合项中的非线性数据项为二次及其以上非线性数据项；所述校准模块用线性回归法拟合出一组加权因子，进而建立采样空间中的数据与其相邻数据之间的一组线性组合关系。
在一个优选的实施例中，所述校准模块使用的校准方法所涉及的公式为：
Figure PCTCN2018096782-appb-000010
其中，j＝1,2,……2·L；
r≠b·R；
Figure PCTCN2018096782-appb-000011
k为K空间中的任意位置；
l为通道的序数；
b为采样空间中沿欠采样相位编码梯度方向，待估计数据和实际采样数据之间的距离；
h为采样空间中沿全采样频率编码梯度方向，待估计数据和实际采样数据之间的距离；
S l为第l个通道的采样空间数据；
S l *为S l的共轭装置矩阵；
k x，k y为采样数据点在二维采样空间的坐标；
Δk x，Δk y为采样空间内完全采样时相邻采样点应满足的最小间隔；
R:加速倍数，也即欠采样倍数；
Figure PCTCN2018096782-appb-000012
为GRAPPA成像方法中的加权因子，可通过对采样空间低频区域中全采样数据进行校准获得；其中，j为通道序数，r为通道位移的序数，上标(2，1)内的第一项数字为指数，第二项为横向距离，也就是
Figure PCTCN2018096782-appb-000013
为
Figure PCTCN2018096782-appb-000014
与距其横向距离为1的加权因子进行点乘；
B 1，B 2，H 1，H 2为成像方法中的加权因子的个数。
利用本发明公开的方法，经过模拟和实验得到结果图3和图4。其中，图3为全面自动校准部分并行采集(GRAPPA)、非线性GRAPPA(NL-GRAPPA)和虚拟线圈的非线性GRAPPA(VCC-NL-GRAPPA)方法在5倍加速(净加速3倍)时候的重建图像比较；图4为GRAPPA、NL-GRAPPA和VCC-NL-GRAPPA方法的均方差和外部加速系数的关系图。其中的Diff.表示重建图像与参考图像的差值，×5表示差值放大5倍显示，MSE表示均方差，R表示外部加速系数，NetR表示净加速系数。
由图可知，通过本发明重建的图像，其MSE相对于GRAPPA法和Nonlinear GRAPPA法均有所减小，同时图像中伪影和噪音均有所减少。
最后需要指出的是，上文所列举的实施例，为本发明较为典型的、较佳实施例，仅用于详细说明、解释本发明的技术方案，以便于读者理解，并不用以限制本发明的保护范围或者应用。因此，在本发明的精神和原则之内所作的任何修改、等同替换、改进等而获得的技术方案，都应被涵盖在本发明的保护范围之内。

Claims (13)

  1. 一种新型的非线性并行重建的磁共振成像方法，其特征在于：
    所述成像方法的步骤包括：
    步骤S1，用多通道线圈并行采集实际线圈数据，根据所述实际线圈数据扩展出相同通道数的虚拟线圈数据，所述虚拟线圈数据和所述实际线圈数据之间满足共轭对称关系；
    步骤S2，根据所述实际线圈数据和所述虚拟线圈数据组合得到数据组合项；
    步骤S3，利用采样空间中，低频区域内完全采样的数据，结合数据组合项校准加权因子；
    步骤S4，根据校准的所述加权因子，重建采样空间中的缺失数据得到重建数据，所述缺失数据指采样空间中，高频区域没有采集的数据；
    步骤S5，融合采样空间中所述低频区域内完全采样数据和所述重建数据得到完整的采样空间数据；
    步骤S6，根据多通道完整的采样空间数据得到最终重建图像。
  2. 根据权利要求1所述的成像方法，其特征在于：
    所述步骤S6为，将所述多通道完整的采样空间数据进行平方求和得到所述最终重建图像，或者将所述多通道完整的采样空间数据与多通道线圈的敏感度相乘后求和得到所述最终重建图像。
  3. 根据权利要求2所述的成像方法，其特征在于：
    在所述步骤S2中，所述数据组合项包括常数、线性数据项和非线性数据项。
  4. 根据权利要求3所述的成像方法，其特征在于：
    所述步骤S3为：利用低频区域内完全采样的数据，结合数据组合项产生混合数据组合项，所述混合数据组合项包括常数、线性数据项和非线性数据项。
  5. 根据权利要求4所述的成像方法，其特征在于：
    所述数据组合项中的非线性数据项为二次及其以上非线性数据项，和/或所述混合数据组合项中的非线性数据项为二次及其以上非线性数据项。
  6. 根据权利要求5所述的成像方法，其特征在于：
    所述步骤S3中的校准方法为，用线性回归法拟合出一组加权因子，进而建立采样空间中的数据与其相邻数据之间的一组线性组合关系。
  7. 根据权利要求6所述的成像方法，其特征在于：
    所述步骤S3中，校准方法所涉及的公式为：
    Figure PCTCN2018096782-appb-100001
    其中，j＝1,2,……2·L；
    r≠b·R；
    Figure PCTCN2018096782-appb-100002
  8. 一种计算机可读介质，该计算机可读介质具有存储在其中的程序，该程序用于计算机执行权利要求1～7中任一项所述的成像方法。
  9. 一种用于使用权利要求1～7中任一项所述成像方法的磁共振图像重建装置，其特征在于：
    包括采集模块、数据组合模块、校准模块、重建数据模块、融合模块、成像模块；
    所述采集模块利用多通道线圈并行采集实际线圈数据，根据所述实际线圈数据扩展出相同通道数的虚拟线圈数据，所述虚拟线圈数据和所述实际线圈数据之间满足共轭对称关系；
    所述数据组合模块根据所述实际线圈数据和所述虚拟线圈数据组合得到数据组合项；
    所述校准模块利用采样空间中，低频区域内完全采样的数据，结合数据组合项校准加权因子；
    所述重建数据模块根据校准的所述加权因子，重建采样空间中的缺失数据得到重建数据，所述缺失数据指采样空间中，高频区域没有采集的数据；
    所述融合模块融合采样空间中所述低频区域内完全采样数据和所述重建数据得到完整的采样空间数据；
    所述成像模块根据多通道完整的采样空间数据得到最终重建图像。
  10. 根据权利要求9所述的装置，其特征在于：
    所述成像模块将所述多通道完整的采样空间数据进行平方求和得到所述最终重建图像，或者将所述多通道完整的采样空间数据与多通道线圈的敏感度相乘后求和得到所述最终重建图像。
  11. 根据权利要求9所述的装置，其特征在于：
    所述数据组合模块得到的所述数据组合项包括常数、线性数据项和非线性数据项；
    所述校准模块利用低频区域内完全采样的数据，结合数据组合项产生混合数据组合项，所述混合数据组合项包括常数、线性数据项和非线性数据项。
  12. 根据权利要求9所述的装置，其特征在于：
    所述校准模块得到的所述数据组合项中的非线性数据项为二次及其以上非线性数据项，和/或所述混合数据组合项中的非线性数据项为二次及其以上非线性数据项；
    所述校准模块用线性回归法拟合出一组加权因子，进而建立采样空间中的数据与其相邻数据之间的一组线性组合关系。
  13. 根据权利要求9所述的装置，其特征在于：
    所述校准模块使用的校准方法所涉及的公式为：
    Figure PCTCN2018096782-appb-100003
    其中，j＝1,2,……2·L；
    r≠b·R；
    Figure PCTCN2018096782-appb-100004
PCT/CN2018/096782 2018-02-09 2018-07-24 一种新型的非线性并行重建的磁共振成像方法、装置及介质 WO2019153659A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP18882276.1A EP3751301A4 (en) 2018-02-09 2018-07-24 NEW METHOD, DEVICE AND SUPPORT FOR MAGNETIC RESONANCE IMAGING OF NON-LINEAR PARALLEL RECONSTRUCTION
US16/473,806 US11579229B2 (en) 2018-02-09 2018-07-24 Imaging method and device for nonlinear parallel magnetic resonance image reconstruction, and medium

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201810132444.9A CN110133557B (zh) 2018-02-09 2018-02-09 一种新型的非线性并行重建的磁共振成像方法、装置及介质
CN201810132444.9 2018-02-09

Publications (1)

Publication Number Publication Date
WO2019153659A1 true WO2019153659A1 (zh) 2019-08-15

Family

ID=67549146

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/096782 WO2019153659A1 (zh) 2018-02-09 2018-07-24 一种新型的非线性并行重建的磁共振成像方法、装置及介质

Country Status (4)

Country Link
US (1) US11579229B2 (zh)
EP (1) EP3751301A4 (zh)
CN (1) CN110133557B (zh)
WO (1) WO2019153659A1 (zh)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111103562B (zh) * 2019-12-09 2022-03-04 中国科学院深圳先进技术研究院 一种同时多个片层成像的重建方法及装置
CN112557981B (zh) * 2020-12-03 2023-06-13 川北医学院 一种并行磁共振成像的改进算法
CN112526423B (zh) * 2020-12-10 2023-06-06 川北医学院 基于共轭和层间信息的并行磁共振成像算法
US20240046533A1 (en) * 2020-12-17 2024-02-08 University Of Southern California Region-optimzed virtual (rovir) coils
US12105169B2 (en) * 2021-01-26 2024-10-01 Ohio State Innovation Foundation High-dimensional fast convolutional framework (HICU) for calibrationless MRI
CN114565690B (zh) * 2021-12-15 2024-09-13 中国科学院深圳先进技术研究院 基于波浪梯度编码场和深度学习模型的磁共振成像方法

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1799498A (zh) * 2004-12-31 2006-07-12 西门子（中国）有限公司 磁共振成像快速广义自校准并行采集图像重建算法
CN1827039A (zh) * 2005-02-28 2006-09-06 西门子（中国）有限公司 频域灵敏度编码磁共振并行成像方法
US20140340083A1 (en) * 2013-05-17 2014-11-20 Shanghai United Imaging Healthcare Co., Ltd. Parallel acquisition image reconstruction method and device for magnetic resonance imaging
CN104749538A (zh) * 2015-04-30 2015-07-01 郑州轻工业学院 一种并行磁共振成像相位处理方法

Family Cites Families (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2008053436A1 (en) * 2006-10-31 2008-05-08 Philips Intellectual Property & Standards Gmbh Mri rf encoding using multiple transmit coils
US20140163664A1 (en) * 2006-11-21 2014-06-12 David S. Goldsmith Integrated system for the ballistic and nonballistic infixion and retrieval of implants with or without drug targeting
US8076938B2 (en) * 2009-03-31 2011-12-13 General Electric Company System and method of parallel imaging with calibration to a virtual coil
US9035003B2 (en) * 2010-08-31 2015-05-19 The University Of Kansas High capacity magnetic nanoparticles as supports for reagents and catalysts
CN102519597B (zh) * 2011-12-14 2013-12-25 中国电子科技集团公司第四十一研究所 一种傅里叶变换光谱仪相位校正切趾方法
CN103185878B (zh) * 2011-12-27 2015-04-15 上海联影医疗科技有限公司 磁共振图像并行采集以及图像重建方法
CN102937706B (zh) * 2012-10-31 2016-08-03 中国科学院深圳先进技术研究院 一种磁共振并行成像方法及磁共振成像仪
US9581671B2 (en) * 2014-02-27 2017-02-28 Toshiba Medical Systems Corporation Magnetic resonance imaging with consistent geometries
CN104434108B (zh) * 2014-12-09 2017-02-01 中国科学院武汉物理与数学研究所 一种基于相邻层信息和欠采样的多层磁共振快速成像方法
WO2017113205A1 (zh) * 2015-12-30 2017-07-06 中国科学院深圳先进技术研究院 一种基于深度卷积神经网络的快速磁共振成像方法及装置
CN106772167B (zh) * 2016-12-01 2019-05-07 中国科学院深圳先进技术研究院 核磁共振成像方法及装置
CN107507148B (zh) * 2017-08-30 2018-12-18 南方医科大学 基于卷积神经网络去除磁共振图像降采样伪影的方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1799498A (zh) * 2004-12-31 2006-07-12 西门子（中国）有限公司 磁共振成像快速广义自校准并行采集图像重建算法
CN1827039A (zh) * 2005-02-28 2006-09-06 西门子（中国）有限公司 频域灵敏度编码磁共振并行成像方法
US20140340083A1 (en) * 2013-05-17 2014-11-20 Shanghai United Imaging Healthcare Co., Ltd. Parallel acquisition image reconstruction method and device for magnetic resonance imaging
CN104749538A (zh) * 2015-04-30 2015-07-01 郑州轻工业学院 一种并行磁共振成像相位处理方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3751301A4 *

Also Published As

Publication number Publication date
US11579229B2 (en) 2023-02-14
CN110133557A (zh) 2019-08-16
EP3751301A4 (en) 2021-11-17
CN110133557B (zh) 2021-08-31
EP3751301A1 (en) 2020-12-16
US20210356545A1 (en) 2021-11-18

Similar Documents

Publication Publication Date Title
WO2019153659A1 (zh) 一种新型的非线性并行重建的磁共振成像方法、装置及介质
US10393842B1 (en) Highly-scalable image reconstruction using deep convolutional neural networks with bandpass filtering
Chen et al. A primal–dual fixed point algorithm for convex separable minimization with applications to image restoration
Cheng et al. Highly scalable image reconstruction using deep neural networks with bandpass filtering
Ye et al. Projection reconstruction MR imaging using FOCUSS
US8879852B2 (en) Non-contrast-enhanced 4D MRA using compressed sensing reconstruction
CN107991636B (zh) 一种基于适应性结构低秩矩阵的快速磁共振图像重建方法
US8379951B2 (en) Auto calibration parallel imaging reconstruction method from arbitrary k-space sampling
Holme et al. ENLIVE: an efficient nonlinear method for calibrationless and robust parallel imaging
CN100578546C (zh) 基于复二维奇异谱分析的磁共振部分k数据图像重建方法
US10089713B2 (en) Systems and methods for registration of images
RU2626184C2 (ru) Способ, устройство и система для реконструкции магнитно-резонансного изображения
CN106772167B (zh) 核磁共振成像方法及装置
US11710261B2 (en) Scan-specific recurrent neural network for image reconstruction
WO2020006959A1 (zh) 基于平面回波成像的磁共振水脂分离和定量方法及装置
Dar et al. Synergistic reconstruction and synthesis via generative adversarial networks for accelerated multi-contrast MRI
Ma et al. Shearlet-based compressed sensing for fast 3D cardiac MR imaging using iterative reweighting
US6560353B1 (en) Method of MRI image reconstruction from partially acquired data in two or more dimensions using a multidimensional inverse transform technique
CN105184755B (zh) 基于自一致性的含联合全变分的并行磁共振成像高质量重构方法
Zhang et al. Alternating direction method of multipliers for nonconvex log total variation image restoration
CN105678822A (zh) 一种基于Split Bregman迭代的三正则磁共振图像重构方法
CN117233674A (zh) 基于单激发buda采集和联合约束重建的mr成像方法及系统
Hu et al. SPICE: Self-supervised learning for MRI with automatic coil sensitivity estimation
CN102579045B (zh) 磁共振超欠采样k数据的稀疏化成像方法
CN115311135A (zh) 一种基于3dcnn的各向同性mri分辨率重建方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18882276

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2018882276

Country of ref document: EP

Effective date: 20200909