WO2020006959A1 - 基于平面回波成像的磁共振水脂分离和定量方法及装置 - Google Patents

基于平面回波成像的磁共振水脂分离和定量方法及装置 Download PDF

Info

Publication number
WO2020006959A1
WO2020006959A1 PCT/CN2018/116233 CN2018116233W WO2020006959A1 WO 2020006959 A1 WO2020006959 A1 WO 2020006959A1 CN 2018116233 W CN2018116233 W CN 2018116233W WO 2020006959 A1 WO2020006959 A1 WO 2020006959A1
Authority
WO
WIPO (PCT)
Prior art keywords
fat
water
dimensional
signal
echo
Prior art date
Application number
PCT/CN2018/116233
Other languages
English (en)
French (fr)
Inventor
郭华
胡张选
董子菁
Original Assignee
清华大学
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 清华大学 filed Critical 清华大学
Publication of WO2020006959A1 publication Critical patent/WO2020006959A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N24/00—Investigating or analyzing materials by the use of nuclear magnetic resonance, electron paramagnetic resonance or other spin effects

Definitions

  • the invention relates to the field of imaging technology, in particular to a method and a device for separating and quantifying magnetic resonance water and lipid based on planar echo imaging.
  • the relatively bright fat signal of magnetic resonance imaging will affect the observation of tissue structure and calculation of some important indicators, for example, ADC (Apparent diffusion coefficient), etc.
  • the fat content itself can also be used as a clinical Diagnostic indicators, so the separation or quantification of fat signals has important applications in the clinic.
  • fat-pressing technology has been widely used.
  • SPIR spectral pre-saturation by inversion recovery, selective partial inversion recovery
  • SPAIR spectral attenuated inversion recovery
  • spectrum Attenuation inversion recovery TIR
  • short TI inversion recovery TIR
  • EPI Echo Planar Imaging, planar echo acquisition technology
  • RF Radio Frequency (radio frequency) to complete the acquisition of the entire k-space
  • EPI acquisition also has its own shortcomings, and longer read times will introduce The blurring effect caused by attenuation, and the lower bandwidth of the phase encoding direction will cause severe image deformation and larger chemical shifts of fat signals relative to water signals at the junctions of different tissues with large differences in magnetic media ratio, which will affect important tissue structures. Observations and results of quantitative analysis.
  • MS-EPI multiple excitation EPI
  • iEPI interleaved EPI
  • rsEPI readout-segmented EP
  • PROPELLER-EPI etc.
  • Point spread function (PSF) -based EPI acquisition provides an effective way to solve these problems.
  • the obtained EPI is completely free of distortion and image blur caused by T2 * attenuation.
  • tilted - The acceleration of CAIPI technology greatly improves the time efficiency of PSF-EPI, and has great clinical application value.
  • water-fat separation or quantification technology can also be used for fat quantitative analysis while eliminating the effect of fat signal on the water phase.
  • Dixon method which uses the difference between water signal and fat signal resonance frequency. The resulting phase difference is collected at different echo times, and the water phase and fat phase are calculated.
  • IDEAL iterative decomposition of water, and fat with the echo-asymmetry
  • QPBO quadrature pseudoboolean optimization
  • the present invention aims to solve at least one of the technical problems in the related technology to a certain extent.
  • an object of the present invention is to propose a magnetic resonance water-fat separation and quantification method based on planar echo imaging.
  • the method can effectively improve the applicability of separation and quantification, high collection efficiency, no distortion, no T2 * blur Effect, high signal-to-noise ratio, simple and easy to implement.
  • Another object of the present invention is to propose a magnetic resonance water-fat separation and quantification device based on planar echo imaging.
  • an embodiment of the present invention provides a method for separating and quantifying magnetic resonance fluid and lipid based on planar echo imaging, including the following steps: using a planar echo acquisition imaging sequence PSF encoded by a deformation-free point spread function -EPI performs data collection to obtain multi-channel down-sampled k-space data, and complete three-dimensional k-space data is recovered through tiled-CAIPI according to the k-space data; using the three-dimensional k-space data to perform three-dimensional inverse Fourier transform , And obtain a three-dimensional amplitude image through channel merging; extract a plurality of two-dimensional k-space data from the three-dimensional k-space data, obtain a complex image by two-dimensional inverse Fourier transform, and calculate a corresponding echo offset time; Water-fat separation and quantitative calculation are performed on the complex image and the echo offset time to obtain the water signal in each pixel and the fat signal ratio at each frequency peak; according to the water signal and the fat signal ratio at each
  • the magnetic resonance water-fat separation and quantification method based on planar echo imaging collects deformation-free anatomical structure images or diffusion images through PSF-coded EPI, and combines water-fat separation or quantification technology with it, which may not Additional acquisition data or offset data readout window will obtain the image or fat percentage after water-fat separation, and directly calculate the water-fat composition at the same time, thereby effectively improving the applicability of separation and quantification, high collection efficiency, no deformation, no T2 * Blur effect, high signal-to-noise ratio, easy to implement.
  • the magnetic resonance water-fat separation and quantification method based on planar echo imaging may also have the following additional technical features:
  • the recovering the three-dimensional k-space data according to tilted-CAIPI further includes: reducing the multi-channel k-space data of the phase encoding direction and the point spread function encoding direction. Simultaneously perform reconstruction to obtain the three-dimensional k-space data, where the reconstruction process is:
  • d i (a, b, c) is the data that the i-th coil needs to be interpolated at (a, b, c)
  • d i ′ (a ′, b ′, c ′) is collected in kernelK Data
  • w is the estimated interpolation weight coefficient
  • N c is the number of coils.
  • the three-dimensional amplitude data of the three-dimensional amplitude image is:
  • H (s, y) is the point spread function
  • s and y in the image domain correspond to the k s and k y directions of the k-space domain
  • ⁇ (s) is the image signal strength without deformation
  • ⁇ is the impulse function
  • ⁇ (s) is the shift of the point spread function caused by image distortion.
  • the echo offset time is:
  • n is the number of encoding steps of the extracted two-dimensional k-space data in the phase encoding direction.
  • the water signal in each pixel is:
  • ⁇ W, q and ⁇ F, q respectively represent the water signal and the fat signal component at the pixel q, and f B, q are the field strength offsets there, Respective resonance frequency shifts of the G-type fat signal components with respect to the water signal, and ⁇ g is the proportion of the g-th component in the fat signal.
  • the water phase and each fat phase are:
  • another embodiment of the present invention provides a magnetic resonance water-fat separation and quantification device based on planar echo imaging, including: an acquisition module for planar echoes encoded by a deformation-free point spread function
  • the imaging sequence PSF-EPI is collected for data collection to obtain multi-channel down-sampling k-space data, and according to the k-space data, three-dimensional k-space data is recovered through tilted-CAIPI; a transform and merge module is used for the k-space
  • the data is subjected to a three-dimensional inverse Fourier transform, and a three-dimensional amplitude image is obtained through channel merging; an extraction module is used to extract a plurality of two-dimensional k-space data from the three-dimensional k-space data, and a complex number is obtained through two-dimensional inverse Fourier transform.
  • Image and calculate the corresponding echo offset time a separation and calculation module for using the complex image and the echo offset time for water-fat separation and quantitative calculation to obtain the water signal in each pixel and each frequency peak A fat signal ratio at a location; a calculation module configured to compare the three-dimensional amplitude image in the phase according to the water signal and the fat signal ratio at each frequency peak The bit-coded direction is calculated by weighted square sum to obtain an undistorted water phase and each fatty phase.
  • the magnetic resonance water-fat separation and quantification device based on planar echo imaging collects deformation-free anatomical structure images or diffusion images through PSF-coded EPI, and combines water-fat separation or quantification technology with it, which may not be required. Additional acquisition data or offset data readout window will obtain the image or fat percentage after water-fat separation, and directly calculate the water-fat composition at the same time, thereby effectively improving the applicability of separation and quantification, high collection efficiency, no deformation, no T2 * Blur effect, high signal-to-noise ratio, easy to implement.
  • the magnetic resonance water-fat separation and quantification device based on planar echo imaging may also have the following additional technical features:
  • the acquisition module is further configured to simultaneously reconstruct the multi-channel down-sampled k-space data of the phase encoding direction and the point spread function encoding direction to obtain the three-dimensional k Spatial data, where the reconstruction process is:
  • d i (a, b, c) is the data that the i-th coil needs to be interpolated at (a, b, c)
  • d i ′ (a ′, b ′, c ′) is collected in kernelK Data
  • w is the estimated interpolation weight coefficient
  • N c is the number of coils.
  • the three-dimensional amplitude data of the three-dimensional amplitude image is:
  • H (s, y) is the point spread function
  • s and y in the image domain correspond to the k s and k y directions of the k-space domain
  • ⁇ (s) is the image signal strength without deformation
  • ⁇ is the impulse function
  • ⁇ (s) is the shift of the point spread function caused by image distortion.
  • the echo offset time is:
  • n is the number of encoding steps in the phase encoding direction of the extracted two-dimensional k-space data; the water signal in each pixel is:
  • ⁇ W, q and ⁇ F, q respectively represent the water signal and the fat signal component at the pixel q, and f B, q are the field strength offsets there, Respective resonance frequency offsets of the G-type fat signal components with respect to the water signal, ⁇ g is the proportion of the g-th component in the fat signal; the water phase and each fat phase are:
  • FIG. 1 is a flowchart of a magnetic resonance water-fat separation and quantification method based on planar echo imaging according to an embodiment of the present invention
  • FIG. 2 is a flowchart of a magnetic resonance water-fat separation and quantification method based on planar echo imaging according to a specific embodiment of the present invention
  • FIG. 3 is a schematic diagram of a planar echo acquisition magnetic resonance imaging scan sequence encoded by a point spread function according to an embodiment of the present invention
  • FIG. 4 is a flowchart of obtaining an amplitude image according to an embodiment of the present invention.
  • FIG. 5 is a two-dimensional spatial data extraction flowchart according to an embodiment of the present invention.
  • FIG. 6 is a schematic structural diagram of a magnetic resonance water-fat separation and quantification device based on planar echo imaging according to an embodiment of the present invention.
  • FIG. 1 is a flowchart of a magnetic resonance water-fat separation and quantification method based on planar echo imaging according to an embodiment of the present invention.
  • the magnetic resonance water-fat separation and quantification method based on planar echo imaging includes the following steps:
  • step S101 a plane echo acquisition imaging sequence PSF-EPI encoded with a non-deformed point spread function is used for data acquisition to obtain multi-channel down-sampled k-space data, and the completeness is recovered by tilted-CAIPI according to the k-space data. 3D k-space data.
  • the embodiment of the present invention uses the PSF-encoded EPI imaging sequence for data acquisition, acquires multi-channel down-sampled k-space data, and uses tilted-CAIPI technology to down-sample k-space signals The restoration is performed to obtain complete PSF-encoded 3D k-space data.
  • restoring three-dimensional k-space data according to tilted-CAIPI further includes: simultaneously reconstructing multi-channel down-sampled k-space data of a phase encoding direction and a point spread function encoding direction to obtain three-dimensional k Spatial data, where the reconstruction process is:
  • d i (a, b, c) is the data that the i-th coil needs to be interpolated at (a, b, c)
  • d i ′ (a ′, b ′, c ′) is collected in kernelK Data
  • w is the estimated interpolation weight coefficient
  • N c is the number of coils.
  • the point-spread function-encoded MRI scan sequence is shown in Figure 3 (using spin-echo diffusion magnetic resonance imaging as an example).
  • An additional point is applied before the phase encoding of the traditional single-shot EPI sequence.
  • Diffusion function phase coding The k-space coding interval should be consistent with the basic EPI phase coding. The coding range depends on the required image resolution. Therefore, the k-space data collected will have three dimensions, which are the reading direction k x , EPI phase encoding direction (referred to as EPI-PE) k y , point spread function phase encoding direction (referred to as PSF-PE) k s .
  • a 180 ° echo pulse can be applied to collect two-dimensional navigation echo signals.
  • PSF-PE encoding direction parallel imaging acceleration factors can be used, such as 10 times, 14 times, 18 times, etc.
  • EPI-PE encoding direction parallel imaging acceleration factors can be used, such as 2 times, 4 times, etc.
  • -Partial Fourier acquisition can be implemented in the PE direction, and the embodiment of the present invention does not place restrictions on the imaging acceleration multiple and the partial Fourier acquisition ratio.
  • the EPI imaging data recovery of PSF-encoded high-reduction mining adopts tilted-CAIPI technology.
  • This method simultaneously reconstructs the down-sampling data in the phase encoding and PSF encoding directions.
  • the reconstruction process can be expressed as:
  • d i (a, b, c) is the data that the i-th coil needs to be interpolated at (a, b, c) (corresponding to k x , k y , k s );
  • d i ′ (a ′, b ′, c ′) are the data collected in kernelK, w is the estimated interpolation weight coefficient, and N c is the number of coils.
  • the weight w used in this reconstruction contains the coil sensitivity information and the correlation information (magnetic field nonuniformity information) of the phase encoding and PSF encoding directions. In order to get the weights for training, calibration data needs to be collected.
  • the calibration data needs to be fully sampled in the PSF and phase encoding directions, or restored to full samples, and provide enough data for kernel calculations. Since the collected k-space data has the highest correlation in the diagonal direction, the interpolated kernel Diagonally, as shown in Figure 4A. After the interpolation is completed, three-dimensional partial Fourier reconstruction can be performed using methods such as POCS. Tilted-CAIPI technology can accelerate the PSF-EPI sequence by more than 20 times, and achieve high resolution and no geometric distortion imaging of 4-8 excitations.
  • the EPI-based imaging sequence used in the embodiment of the present invention is not limited to the spin-echo EPI acquisition (SE-EPI) in the embodiment, and gradient echo EPI acquisition (GE-EPI) can also be used to reverse and recover the echo EPI Acquisition (IR-EPI), stimulated echo EPI acquisition (STEAM EPI), etc .; the invention does not limit the imaging contrast, including various T1 weighting, T2 weighting, T2 * weighting, proton density (PD) weighting, diffusion weighting (DWI), etc .;
  • the diffusion preparation sequence of the diffusion imaging that can also be used in the embodiments of the present invention is not limited to the PGSE (pulsed gradient spin echo) in the embodiment, and STE (simulated echo) can also be used.
  • phase correction method of diffusion magnetic resonance imaging in the embodiment of the present invention is not limited to the additional method of collecting navigation echo data in the embodiment, but can also be adopted Air reconstruction (itself reconstructed low resolution image obtained by using the phase information for phase correcting the data collected by the drop), etc., and is not particularly limited.
  • step S102 a three-dimensional inverse Fourier transform is performed on the three-dimensional k-space data, and a three-dimensional amplitude image is obtained through channel merging.
  • an embodiment of the present invention performs three-dimensional inverse Fourier transform on the reconstructed k-space data and uses channel merging to obtain an amplitude image containing PSF information.
  • the three-dimensional amplitude data of the three-dimensional amplitude image is:
  • H (s, y) is the point spread function
  • s and y in the image domain correspond to the k s and k s directions of the k-space domain
  • ⁇ (s) is the intensity of the image signal without deformation
  • ⁇ is the impulse function
  • ⁇ (s) is the shift of the point spread function caused by image distortion.
  • the 3D inverse Fourier transform is performed on the k-space data after mining recovery and channel merge is used. Since the readout direction is independent of the EPI-PE and PSF-PE directions, the obtained 3D amplitude data can be simplified and represented. for:
  • ⁇ (s) is the image signal strength without distortion
  • ⁇ is the impulse function
  • ⁇ (s) is the point spread function deviation caused by image deformation. Shift, as shown in Figure 4B. Integrating the amplitude image along the y direction and s direction according to the following formulas can obtain images without distortion and distortion, as shown in FIG. 4C:
  • the embodiment of the present invention does not limit the channel merging method used, and those skilled in the art may select a corresponding method based on actual conditions.
  • the method for merging multiple-shot multi-channel image domain data mainly includes a square sum SOS method,
  • the optimized signal-to-noise ratio method, the adaptive reconstruction ACC method, the principal component analysis PCA method, and the singular value decomposition SVD method are not specifically limited here.
  • all the multiple-shot multi-channel image domain data can be combined into one image by using, but not limited to, any of the methods described above.
  • step S103 a plurality of two-dimensional k-space data is extracted from the three-dimensional k-space data, and a complex image is obtained through two-dimensional inverse Fourier transform and the corresponding echo offset time is calculated.
  • the echo offset time is:
  • n is the number of encoding steps in the phase encoding direction of the extracted two-dimensional k-space data.
  • n the number of encoding steps and ⁇ k y is the encoding interval
  • N the maximum number of encoding steps
  • the k x -k s plane can be regarded as a traditional two-dimensional k-space data. Means.
  • the sampling interval along its phase encoding direction that is, the PSF-PE direction, is 0, so the image obtained by its inverse Fourier transform does not have distortion and no water signal Chemical shifts with fat signals.
  • the offset time from the time of echo (TE) can be calculated by the following formula:
  • the range of common encoding steps for the selected L two-dimensional k-space data in the PSF-PE direction is:
  • ⁇ W (x) is the water signal component at x
  • ⁇ F (x) is the fat signal component at x
  • f B (x) is the field non-uniformity at that location
  • f F is the relative fat signal
  • t n is different echo offset time
  • s (x; t n ) is an image signal located at x when the echo offset is t n .
  • step S104 water-fat separation and quantitative calculation are performed using a complex image and an echo offset time to obtain a water signal in each pixel and a fat signal ratio at each frequency peak.
  • the water-lipid separation or quantitative method is used to perform water-lipid separation and quantitative calculation by using the complex image obtained in step S103 and the corresponding echo offset time.
  • the water signal in each pixel is:
  • ⁇ W, q and ⁇ F, q respectively represent the water signal and the fat signal component at the pixel q, and f B, q are the field strength offsets there, Respective resonance frequency shifts of the G-type fat signal components with respect to the water signal, and ⁇ g is the proportion of the g-th component in the fat signal.
  • ⁇ W, q and ⁇ F, q respectively represent the water signal and the fat signal component at the pixel q, and f B, q are the field strength offsets there, Respective resonance frequency shifts of G-type fat signal components with respect to water signals, ⁇ g is the proportion of the g-th component in the fat signal, You can use the known statistical value of different fat signal proportions as ⁇ g . At this time, you only need to fit ⁇ W, q , ⁇ F, q and f B, q . The required number of echoes is L ⁇ 3, and you can also treat them as unknown and independent at the same time. Fitting gives ⁇ W, q , And f B, q , the required number of echoes L ⁇ G + 2, and by normalizing the G kinds of fat signal components, we can get the proportion distribution map ⁇ g of different kinds of fat signals.
  • the embodiments of the present invention are not limited to the water-fat separation method used.
  • Common methods for water-fat separation and quantitative calculation include: IDEAL (iterative decomposition of water and fatt with echo asymmetry and least-square-estimation) method, Hierarchical IDEAL method, MAX- IDEAL method, QPBO (quadratic pseudoboolean optimization) method, QPBO-GC (quadratic pseudoboolean optimization graph) method, Hernando-GC (Graph Cut) method, ASR (a safest-first region-growing) method, MRGS (multi-resolution golden section search) and so on.
  • the embodiments of the present invention can perform water-fat separation and quantitative calculation by using, but not limited to, any one of the methods described above.
  • step S105 a weighted sum of squares of the three-dimensional amplitude image in the phase encoding direction is calculated according to the water signal and the fat signal ratio at each frequency peak to obtain an undistorted water phase and each fat phase.
  • the calculated water-fat signal ratio is used to calculate the weighted square sum of the three-dimensional amplitude image in step S102 in the phase encoding direction, thereby obtaining a water phase without distortion and Each fatty phase.
  • the water phase and each fat phase are:
  • the quantitative proportions of the water signal at each pixel and the fat signal at each peak can be calculated as: Using this weight to perform weighted sum of squares (SOS) operation on the three-dimensional amplitude data obtained in step S102 along the y direction according to the following formula, can obtain the high-signal-to-noise non-deformation water phase I W (s) and fat Phase I F (s):
  • the embodiment of the present invention combines the PSF-EPI acquisition technology with the water-fat separation or quantification method, without the need for additional echo acquisition or offset reading window acquisition.
  • the extraction can be used for The water-fat separation or quantitative echo signal is separated and quantified using the aforementioned common techniques.
  • the method of the embodiment of the present invention inherits the advantages of fast acquisition of EPI and the advantages of no deformation, no T2 * blur effect, and high signal-to-noise ratio of PSF-EPI.
  • PSF-EPI various anatomical structures like , Diffusion imaging, etc.
  • various quantitative analyses such as diffusion coefficient ADC and anisotropic FA.
  • a deformation-free anatomical structure image or a diffusion image is acquired through the EPI coded PPI, and the water-fat separation or quantification technology is combined with it.
  • No additional data acquisition or offset data read-out window is required to obtain the water-fat separation image or fat percentage, and directly calculate the water-fat composition at the same time, thereby effectively improving the applicability of separation and quantification, high collection efficiency, no distortion, No T2 * blur effect, high signal-to-noise ratio, easy to implement.
  • FIG. 6 is a schematic structural diagram of a magnetic resonance water-fat separation and quantification device based on planar echo imaging according to an embodiment of the present invention.
  • the magnetic resonance water-fat separation and quantification device 10 based on planar echo imaging includes: an acquisition module 100, a transformation and integration module 200, an extraction module 300, a separation and calculation module 400, and a calculation module 500.
  • the acquisition module 100 is used for data acquisition using a planar echo acquisition imaging sequence PSF-EPI encoded without deformation of a point spread function to obtain multi-channel down-sampled k-space data, and is recovered by tilted-CAIPI according to the k-space data.
  • the transform and merge module 200 is configured to perform three-dimensional inverse Fourier transform on the three-dimensional k-space data, and obtain a three-dimensional amplitude image through channel merging.
  • the extraction module 300 is used to extract a plurality of two-dimensional k-space data from the three-dimensional k-space data, obtain a complex image through a two-dimensional inverse Fourier transform, and calculate a corresponding echo offset time.
  • the separation and calculation module 400 is configured to perform water-fat separation and quantitative calculation using a complex image and an echo offset time to obtain a water signal in each pixel and a fat signal ratio at each frequency peak.
  • the calculation module 500 is configured to calculate a weighted square sum of the three-dimensional amplitude image in the phase encoding direction according to the water signal and the fat signal ratio at each frequency peak to obtain a distortion-free water phase and each fat phase.
  • the device 10 of the embodiment of the present invention collects deformation-free anatomical structure images or diffusion images through PSF-coded EPI, and combines water-fat separation or quantification technology with it, thereby effectively improving the applicability of separation and quantification, high collection efficiency and no deformation No T2 * blur effect, high signal-to-noise ratio, easy to implement.
  • the acquisition module 100 is further configured to simultaneously reconstruct the multi-channel down-sampled k-space data of the phase encoding direction and the point spread function encoding direction to obtain three-dimensional k-space data, wherein the reconstruction process for:
  • d i (a, b, c) is the data that the i-th coil needs to be interpolated at (a, b, c)
  • d i ′ (a ′, b ′, c ′) is collected in kernelK Data
  • w is the estimated interpolation weight coefficient
  • N c is the number of coils.
  • the three-dimensional amplitude data of the three-dimensional amplitude image is:
  • H (s, y) is the point spread function
  • s and y in the image domain correspond to the k s and k y directions of the k-space domain
  • ⁇ (s) is the image signal strength without deformation
  • ⁇ is the impulse function
  • ⁇ (s) is the shift of the point spread function caused by image distortion.
  • the echo offset time is:
  • n is the number of encoding steps in the phase encoding direction of the extracted two-dimensional k-space data; the water signal in each pixel is:
  • ⁇ W, q and ⁇ F, q respectively represent the water signal and the fat signal component at the pixel q, and f B, q are the field strength offsets there, Respective resonance frequency offsets of G-type fat signal components with respect to water signals, ⁇ g is the proportion of the g-th component in the fat signal; the water phase and each fat phase are:
  • a deformation-free anatomical structure image or a diffusion image is acquired through the EPI coded EPI, and the water-fat separation or quantification technology is combined with it to No additional data acquisition or offset data read-out window is required to obtain the water-fat separation image or fat percentage, and directly calculate the water-fat composition at the same time, thereby effectively improving the applicability of separation and quantification, high collection efficiency, no distortion, No T2 * blur effect, high signal-to-noise ratio, easy to implement.
  • first and second are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as “first” and “second” may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "a plurality” is at least two, for example, two, three, etc., unless it is specifically and specifically defined otherwise.

Landscapes

  • Physics & Mathematics (AREA)
  • High Energy & Nuclear Physics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Biochemistry (AREA)
  • General Health & Medical Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Immunology (AREA)
  • Pathology (AREA)
  • Magnetic Resonance Imaging Apparatus (AREA)

Abstract

一种基于平面回波成像的磁共振水脂分离和定量方法及装置，其中，方法包括：使用无变形的点扩散函数编码的平面回波成像序列进行数据采集得到多通道降采的k空间数据，并恢复得到完整的三维k空间数据，并进行三维逆傅里叶变换，通过通道合并得到三维幅值图像；从三维k空间数据抽取多个二维k空间数据，通过二维逆傅里叶变换得到复数图像并计算相应回波偏移时间；使用复数图像和回波偏移时间进行水脂分离和定量计算，以得水信号及各个频率峰值处的脂肪信号比例，并对三维幅值图像在相位编码方向进行加权平方和计算，以得到无变形的水相以及各脂肪相。基于平面回波成像的磁共振水脂分离和定量方法适用性强，采集效率高，无变形、无T2*模糊效应、高信噪比，简单易实现。

Description

基于平面回波成像的磁共振水脂分离和定量方法及装置
相关申请的交叉引用
本申请要求清华大学于2018年07月02日提交的、发明名称为“基于平面回波成像的磁共振水脂分离和定量方法及装置”的、中国专利申请号“201810707718.2”的优先权。
技术领域
本发明涉及成像技术领域，特别涉及一种基于平面回波成像的磁共振水脂分离和定量方法及装置。
背景技术
相关技术中，磁共振成像相对高亮的脂肪信号会影响组织结构的观察以及一些重要指标的计算，例如，ADC(Apparent diffusion coefficient，表观扩散系数)等；同时，脂肪含量本身也可以作为临床诊断指标，因此脂肪信号的分离或定量在临床中具有很重要的应用。为了消除脂肪信号的影响，压脂技术得到了广泛应用，临床中常用的压脂技术包括SPIR(spectral pre-saturation by inversion recovery，选择性部分反转恢复法)，SPAIR(spectral attenuated inversion recovery，频谱衰减反转恢复)，STIR(short TI inversion recovery，短TI反转恢复)，这些方法在某些情况下，尤其是在场不均匀性十分严重的区域，例如头颈区域，前额叶等，可能无法彻底的压制脂肪信号，特别地STIR还会造成信噪比低的问题。
发明内容
本申请是基于发明人对以下问题的认识和发现作出的：
EPI(Echo Planar Imaging，平面回波采集技术)的快速采集特点使其具有成像速度快，对运动不敏感等优势，在临床中得到了广泛应用，尤其是单次激发EPI，在一次RF(Radio Frequency，射频)激发后完成整个k空间的采集，在对成像速度要求高的应用中具有非常重要的价值，例如扩散成像、灌注成像、心脏成像以及实时成像等。然而，EPI采集也有它本身的不足，较长的读出时间会引入
Figure PCTCN2018116233-appb-000001
衰减造成的模糊效应，相位编码方向的较低带宽会导致在磁介质率相差较大的不同组织交界处产生严重的图像变形以及脂肪信号相对水信号的更大的化学位移，从而影响重要组织结构的观察以及量化分析的结果。
单次激发EPI与并行采集技术的结合可以减少读出窗的长度以及ESP(effective echo spacing，有效回波间隔)，减少
Figure PCTCN2018116233-appb-000002
模糊效应以及图像变形和化学位移现象，但是依然受限于加速倍数，同时会降低信噪比。多次激发EPI(MS-EPI)技术，例如iEPI(interleaved EPI)， 读出分段EPI(readout-segmented EP，rsEPI)，PROPELLER-EPI等将整个k空间采集分为若干部分，可以在保持信噪比的条件下减少上述问题，但是依然无法完全消除EPI特有的变形伪影。基于点扩散函数编码(Point spread function，PSF)的EPI采集(PSF-EPI)为解决这些问题提供了一种有效方式，所得到的EPI完全无变形无T2*衰减引起的图像模糊，同时，tilted-CAIPI技术的采集加速大大提高了PSF-EPI的时间效率，具有很大的临床应用价值。
与一般压脂技术不同，水脂分离或定量技术在消除脂肪信号对水相的影响的同时，还可以用于脂肪定量分析，常用如Dixon方法，该方法利用水信号与脂肪信号共振频率的不同所导致的相位差异，采集不同回波时刻的图像，进而计算得出水相以及脂肪相。在此基础上发展出了多种水脂分离或定量方法，例如多点Dixon方法，迭代重建的IDEAL(iterative decomposition of water and fat with echo asymmetry and least-square estimation)方法，Hernando-GC(Hernnado-GraphCut)方法，QPBO(quadratic pseudo boolean optimization)方法等。通过多次回波采集(如FFE(fast field echo，多回波的快速梯度回波))或者多次偏移读出窗口采集(如FSE(fast spin echo，多采集的快速自旋回波))，计算得出水脂信号成分。
本发明旨在至少在一定程度上解决相关技术中的技术问题之一。
为此，本发明的一个目的在于提出一种基于平面回波成像的磁共振水脂分离和定量方法，该方法可以有效提高分离和定量的适用性，采集效率高，无变形、无T2*模糊效应、高信噪比，简单易实现。
本发明的另一个目的在于提出一种基于平面回波成像的磁共振水脂分离和定量装置。
为达到上述目的，本发明一方面实施例提出了一种基于平面回波成像的磁共振水脂分离和定量方法，包括以下步骤：使用无变形的点扩散函数编码的平面回波采集成像序列PSF-EPI进行数据采集，以得到多通道降采的k空间数据，根据所述k空间数据通过tilted-CAIPI恢复得到完整的三维k空间数据；使用所述三维k空间数据进行三维逆傅里叶变换，并通过通道合并得到三维幅值图像；从所述三维k空间数据抽取多个二维k空间数据，通过二维逆傅里叶变换得到复数图像并计算相应回波偏移时间；对所述复数图像和回波偏移时间进行水脂分离和定量计算，以得到每个像素中水信号和每个频率峰值处的脂肪信号比例；根据所述水信号及每个频率峰值处的脂肪信号比例对所述三维幅值图像在相位编码方向进行加权平方和计算，以得到无变形的水相以及各脂肪相。
本发明实施例的基于平面回波成像的磁共振水脂分离和定量方法，通过PSF编码的EPI采集无变形解剖结构图像或扩散图像，并将水脂分离或定量技术与其相结合，可以不需要额外采集数据或偏移数据读出窗口，便得到水脂分离后的图像或者脂肪百分比，直接同时计算得出水脂成分，从而有效提高分离和定量的适用性，采集效率高，无变形、无T2*模糊效应、高信噪比，简单易实现。
另外，根据本发明上述实施例的基于平面回波成像的磁共振水脂分离和定量方法还可以具有以下附加的技术特征：
进一步地，在本发明的一个实施例中，所述根据tilted-CAIPI恢复三维k空间数据，进一步包括：将相位编码方向和所述点扩散函数编码方向的所述多通道降采的k空间数据同时进行重建得到所述三维k空间数据，其中，重建过程为：
Figure PCTCN2018116233-appb-000003
其中，d i(a,b,c)是在(a,b,c)处第i个线圈需要被插值的数据，d i′(a′,b′,c′)是在kernelK中采集到的数据，w为估计得到的插值权重系数，N c是线圈数目。
进一步地，在本发明的一个实施例中，所述三维幅值图像的三维幅值数据为：
I(s,y)＝ρ(s)H(s,y)，
Figure PCTCN2018116233-appb-000004
其中，H(s,y)为点扩散函数，图像域内s，y分别对应k空间域的k s，k y方向，ρ(s)为无变形的图像信号强度，δ为冲激函数，Δ(s)为图像变形所导致的点扩散函数偏移。
进一步地，在本发明的一个实施例中，所述回波偏移时间为：
Figure PCTCN2018116233-appb-000005
其中，
Figure PCTCN2018116233-appb-000006
为基础的单次激发EPI序列的有效回波间隔，n为所抽取的二维k空间数据在相位编码方向的编码步数。
进一步地，在本发明的一个实施例中，所述每个像素中水信号为：
Figure PCTCN2018116233-appb-000007
其中，ρ W,q以及ρ F,q分别代表像素点q处的水信号以及脂肪信号成分，f B,q为该处的场强偏移量，
Figure PCTCN2018116233-appb-000008
分别为G种脂肪信号成分相对水信号的共振频率偏移，α g为第g种成分在脂肪信号中的比重。
进一步地，在本发明的一个实施例中，所述水相以及各脂肪相为：
Figure PCTCN2018116233-appb-000009
Figure PCTCN2018116233-appb-000010
其中，*代表共轭操作，不同峰值处的脂肪相依次计算为
Figure PCTCN2018116233-appb-000011
为达到上述目的，本发明另一方面实施例提出了一种基于平面回波成像的磁共振水脂分离和定量装置，包括：采集模块，用于采用无变形的点扩散函数编码的平面回波采集成像序列PSF-EPI进行数据采集，以得到多通道降采的k空间数据，并根据所述k空间数据通过tilted-CAIPI恢复得到三维k空间数据；变换合并模块，用于对所述k空间数据进行三维逆傅里叶变换，并通过通道合并得到三维幅值图像；抽取模块，用于从所述三维k空间数据抽取多个二维k空间数据，通过二维逆傅里叶变换得到复数图像并计算相应的回波偏移时间；分离与计算模块，用于使用所述复数图像和回波偏移时间进行水脂分离和定量计算，以得到每个像素中水信号和每个频率峰值处的脂肪信号比例；计算模块，用于根据所述水信号及每个频率峰值处的脂肪信号比例对所述三维幅值图像在相位编码方向进行加权平方和计算，以得到无变形的水相以及各脂肪相。
本发明实施例的基于平面回波成像的磁共振水脂分离和定量装置，通过PSF编码的EPI采集无变形解剖结构图像或扩散图像，并将水脂分离或定量技术与其相结合，可以不需要额外采集数据或偏移数据读出窗口，便得到水脂分离后的图像或者脂肪百分比，直接同时计算得出水脂成分，从而有效提高分离和定量的适用性，采集效率高，无变形、无T2*模糊效应、高信噪比，简单易实现。
另外，根据本发明上述实施例的基于平面回波成像的磁共振水脂分离和定量装置还可以具有以下附加的技术特征：
进一步地，在本发明的一个实施例中，所述采集模块进一步用于将相位编码方向和所述点扩散函数编码方向的所述多通道降采的k空间数据同时进行重建得到所述三维k空间数据，其中，重建过程为：
Figure PCTCN2018116233-appb-000012
其中，d i(a,b,c)是在(a,b,c)处第i个线圈需要被插值的数据，d i′(a′,b′,c′)是在kernelK中采集到的数据，w为估计得到的插值权重系数，N c是线圈数目。
进一步地，在本发明的一个实施例中，所述三维幅值图像的三维幅值数据为：
I(s,y)＝ρ(s)H(s,y)，
Figure PCTCN2018116233-appb-000013
其中，H(s,y)为点扩散函数，图像域内s，y分别对应k空间域的k s，k y方向，ρ(s)为无变形的图像信号强度，δ为冲激函数，Δ(s)为图像变形所导致的点扩散函数偏移。
进一步地，在本发明的一个实施例中，其中，所述回波偏移时间为：
Figure PCTCN2018116233-appb-000014
其中，
Figure PCTCN2018116233-appb-000015
为基础的单次激发EPI序列的有效回波间隔，n为所抽取的二维k空间数据 在相位编码方向上的编码步数；所述每个像素中水信号为：
Figure PCTCN2018116233-appb-000016
其中，ρ W,q以及ρ F,q分别代表像素点q处的水信号以及脂肪信号成分，f B,q为该处的场强偏移量，
Figure PCTCN2018116233-appb-000017
分别为G种脂肪信号成分相对水信号的共振频率偏移，α g为第g种成分在脂肪信号中的比重；所述水相以及各脂肪相为：
Figure PCTCN2018116233-appb-000018
Figure PCTCN2018116233-appb-000019
其中，*代表共轭操作，不同峰值处的脂肪相依次计算为
Figure PCTCN2018116233-appb-000020
本发明附加的方面和优点将在下面的描述中部分给出，部分将从下面的描述中变得明显，或通过本发明的实践了解到。
附图说明
本发明上述的和/或附加的方面和优点从下面结合附图对实施例的描述中将变得明显和容易理解，其中：
图1为根据本发明一个实施例的基于平面回波成像的磁共振水脂分离和定量方法的流程图；
图2为根据本发明一个具体实施例的基于平面回波成像的磁共振水脂分离和定量方法的流程图；
图3为根据本发明一个实施例的点扩散函数编码的平面回波采集磁共振成像扫描序列示意图；
图4为根据本发明一个实施例的幅值图像获取流程图；
图5为根据本发明一个实施例的二维空间数据抽取流程图；
图6为根据本发明一个实施例的基于平面回波成像的磁共振水脂分离和定量装置的结构示意图。
具体实施方式
下面详细描述本发明的实施例，所述实施例的示例在附图中示出，其中自始至终相同或类似的标号表示相同或类似的元件或具有相同或类似功能的元件。下面通过参考附图描述的实施例是示例性的，旨在用于解释本发明，而不能理解为对本发明的限制。
下面参照附图描述根据本发明实施例提出的基于平面回波成像的磁共振水脂分离和定量方法及装置，首先将参照附图描述根据本发明实施例提出的基于平面回波成像的磁共振水脂分离和定量方法。
图1是本发明一个实施例的基于平面回波成像的磁共振水脂分离和定量方法的流程图。
如图1所示，该基于平面回波成像的磁共振水脂分离和定量方法包括以下步骤：
在步骤S101中，使用无变形的点扩散函数编码的平面回波采集成像序列PSF-EPI进行数据采集，以得到多通道降采的k空间数据，并根据k空间数据通过tilted-CAIPI恢复得到完整的三维k空间数据。
可以理解的是，如图2所示，本发明实施例使用基于PSF编码的EPI成像序列进行数据采集，获取多通道降采的k空间数据，并使用tilted-CAIPI技术对降采样的k空间信号进行恢复，获得完整的PSF编码的三维k空间数据。
进一步地，在本发明的一个实施例中，根据tilted-CAIPI恢复三维k空间数据，进一步包括：将相位编码方向和点扩散函数编码方向的多通道降采的k空间数据同时进行重建得到三维k空间数据，其中，重建过程为：
Figure PCTCN2018116233-appb-000021
其中，d i(a,b,c)是在(a,b,c)处第i个线圈需要被插值的数据，d i′(a′,b′,c′)是在kernelK中采集到的数据，w为估计得到的插值权重系数，N c是线圈数目。
具体而言，点扩散函数编码的磁共振成像扫描序列如图3所示(以自旋回波的扩散磁共振成像举例)，在传统的单次激发EPI序列的相位编码之前会施加一个额外的点扩散函数相位编码，其k空间编码间隔应与基础的EPI相位编码保持一致，编码范围取决于需求的图像分辨率，从而采集得到的k空间数据将存在三个维度，分别为读出方向k x，EPI相位编码方向(简称EPI-PE)k y，点扩散函数相位编码方向(简称PSF-PE)k s。针对扩散磁共振成像，为了进行重建过程中的相位矫正，在成像回波之后，可施加180°回聚脉冲以采集二维导航回波信号。PSF-PE编码方向的并行成像加速倍数可以使用如10倍、14倍、18倍等，EPI-PE编码方向的并行成像加速倍数可以使用如2倍、4倍等，同时在PSF-PE及EPI-PE方向均可以实施部分傅里叶采集，本发明实施例对成像加速倍数及部分傅里叶采集比例不加限制。
高倍降采的PSF编码的EPI成像数据恢复采用tilted-CAIPI技术。该方法将相位编码和PSF编码方向的降采数据同时进行重建，如图4A所示，重建过程可以被表示为：
Figure PCTCN2018116233-appb-000022
该式中d i(a,b,c)是在(a,b,c)(对应k x,k y,k s)处第i个线圈需要被插值的数据； d i′(a′,b′,c′)是在kernelK中采集到的数据，w为估计得到的插值权重系数，N c是线圈数目。在此重建中使用的权重w包含了线圈敏感度信息，以及相位编码和PSF编码方向的相关性信息(磁场不均匀性信息)。为了训练得到该权重，需要采集校准数据。校准数据需要在PSF和相位编码方向满采样，或者恢复为满采样，并提供足够的数据来进行kernel计算，由于所采集的k空间数据在对角线方向具有最高的相关性，所以插值的kernel沿对角线方向，如图4A所示。插值完成后，可以使用如POCS方法进行三维部分傅里叶重建。Tilted-CAIPI技术能够将PSF-EPI序列加速20倍以上，实现4-8次激发的高分辨率无几何失真成像。
本发明实施例所采用的基于EPI的成像序列不局限于实施例中的自旋回波EPI采集(SE-EPI)，亦可以使用梯度回波EPI采集(GE-EPI)，反转恢复回波EPI采集(IR-EPI)，受激回波EPI采集(STEAM EPI)等；本发明对成像对比度不加限制，包括各种T1加权、T2加权、T2 *加权、质子密度(PD)加权、扩散加权(DWI)等；本发明实施例也可使用的扩散成像的扩散准备序列不局限于实施例中的PGSE(pulsed gradient spin echo，脉冲梯度自旋回波序列)，亦可以使用STE(simulated echo，受激回波扩散准备序列)，OGSE(oscillating gradient spin echo，振荡梯度自旋回波扩散准备序列)，DDE(double diffusion encoding，双重扩散编码扩散准备序列)，CODE(convex optimized diffusion encoding，凸优化扩散编码扩散准备序列)等；本发明实施例中扩散磁共振成像的相位矫正的方式不局限于实施例中的额外采集导航回波数据方法，亦可以采用自导航的重建方式(由降采数据自身重建得到低分辨率图像利用其相位信息用于相位矫正)等，在此不做具体限定。
在步骤S102中，对三维k空间数据进行三维逆傅里叶变换，并通过通道合并得到三维幅值图像。
可以理解的是，如图2所示，本发明实施例对重建得到的k空间数据进行三维逆傅里叶变换并使用通道合并获取包含PSF信息的幅值图像。
进一步地，在本发明的一个实施例中，三维幅值图像的三维幅值数据为：
I(s,y)＝ρ(s)H(s,y)，
Figure PCTCN2018116233-appb-000023
其中，H(s,y)为点扩散函数，图像域内s，y分别对应k空间域的k s，k s方向，ρ(s)为无变形的图像信号强度，δ为冲激函数，Δ(s)为图像变形所导致的点扩散函数偏移。
具体而言，对降采恢复之后的k空间数据进行三维逆傅里叶变换并使用通道合并，由于读出方向独立于EPI-PE及PSF-PE方向，所得到的三维幅值数据可以简化表示为：
I(s,y)＝ρ(s)H(s,y)，[2]
其中：
Figure PCTCN2018116233-appb-000024
即为点扩散函数。图像域内s，y分别对应k空间域的k s，k y方向，ρ(s)为无变形的图像信号强度，δ为冲激函数，Δ(s)为图像变形所导致的点扩散函数偏移，如图4B所示。对该幅值图像按下式分别沿y方向和s方向进行积分可以得到无变形及有变形的图像，如图4C所示：
I(s)＝∫I(s,y)dy，[4]
I(y)＝∫I(s,y)ds，[5]
本发明实施例不限制所采用的通道合并方法，本领域技术人员可以根基实际情况选择相应的方法，例如，对多次激发的多通道的图像域数据进行合并的方法主要包括平方和SOS方法、最优化信噪比方法、自适应重建ACC方法、主成分分析PCA方法和奇异值分解SVD方法等，在此不做具体限定。本发明实施例可通过但并不限于上述任意一种方法将所有多次激发的多通道的图像域数据合并为一幅图像。
在步骤S103中，从三维k空间数据抽取多个二维k空间数据，并通过二维逆傅里叶变换得到复数图像并计算相应回波偏移时间。
可以理解的是，如图2所示，从恢复的三维k空间数据中使用全部或抽取若干用于水脂分离或定量的二维k空间数据(例如，5或7个)，通过二维逆傅里叶变换获得相应的复数图像并计算对应的回波偏移时间。
进一步地，在本发明的一个实施例中，回波偏移时间为：
Figure PCTCN2018116233-appb-000025
其中，
Figure PCTCN2018116233-appb-000026
为基础的单次激发EPI序列的有效回波间隔，n为抽取的二维k空间数据在相位编码方向上的编码步数。
具体而言，考虑重建恢复的三维k空间数据，对于每个EPI相位编码步骤nΔk y而言，其中n为编码步数，Δk y为编码间隔(如图5A所示，以最大编码步数N为6举例，n的取值范围为-6至6)，k x-k s平面可以看成一个传统的二维k空间数据，以
Figure PCTCN2018116233-appb-000027
表示。同时，对于该二维k空间而言，沿着其相位编码方向即PSF-PE方向的采样间隔为0，因此由其经过逆傅里叶变换得到的图像不存在变形，且不会发生水信号与脂肪信号的化学位移。对于EPI-PE编码步骤为n处的二维k空间数据，其相对于回波时刻(TE)的偏移时间可以由下式计算：
Figure PCTCN2018116233-appb-000028
其中，
Figure PCTCN2018116233-appb-000029
为基础的单次激发EPI序列的有效回波间隔(echo spacing)，可以抽取若干不同n处的二维k空间数据，利用其回波时间偏移进行水脂分离或定量。与传统的水脂分离或定量过程相同，本实施例仅需要有限个回波时刻的图像，例如5或7个， 记为L。注意到不同EPI-PE编码处的二维k空间数据在PSF-PE方向的编码范围不同，因此需要对所选用的二维k空间数据进行处理。考虑L个二维k空间数据分别位于EPI-PE编码步数为
Figure PCTCN2018116233-appb-000030
处，可记为
Figure PCTCN2018116233-appb-000031
PSF-PE方向的相位编码可以表示为mΔk s，其中Δk s为该方向的编码间隔，应与Δk y相同，m为编码步数(如图5A所示，以最大编码步数M为6举例，m的取值范围为-6至6)。因此，对于EPI-PE编码步数为n处的二维k空间数据，PSF-PE方向的k空间中心处于m＝-n处，如图5A中黑色实心点所示，实际编码步数范围为n-M到n+M。所选用的L个二维k空间数据在PSF-PE方向的共同编码步数范围为：
Figure PCTCN2018116233-appb-000032
从
Figure PCTCN2018116233-appb-000033
中分别提取该编码部分的数据并通过填零操作实现与设计的PSF EPI成像相同的分辨率，结果可记为
Figure PCTCN2018116233-appb-000034
如图5B所示为例，在图5A中选用EPI-PE编码步数分别为0，1，2的三个回波时刻的k空间数据(虚线框内)，经过上述操作之后的
Figure PCTCN2018116233-appb-000035
如图5B中实线框所表示，其中纹理点表示经过填零的k空间数据。对
Figure PCTCN2018116233-appb-000036
进行二维逆傅里叶变换可得到图像域数据
Figure PCTCN2018116233-appb-000037
相应的回波时刻偏移分别为
Figure PCTCN2018116233-appb-000038
如前，
Figure PCTCN2018116233-appb-000039
均为无变形且不存在化学位移的图像，同时在扩散成像场合中的不同回波时刻之间扩散磁共振成像中运动所导致的相位误差已经在(1)中的tilted-CAIPI重建过程中消除，从而可以简化水脂分离或定量算法中的信号模型，如下式表示：
Figure PCTCN2018116233-appb-000040
其中，ρ W(x)为位于x处的水信号成分，ρ F(x)为位于x处的脂肪信号成分，f B(x)为该处的场不均匀性，f F为脂肪信号相对水信号的共振频率偏移，t n为不同的回波偏移时间，s(x；t n)为回波偏移为t n时位于x处的图像信号。
在步骤S104中，使用复数图像和回波偏移时间进行水脂分离和定量计算，以得到每个像素中水信号和每个频率峰值处的脂肪信号比例。
可以理解的是，如图2所示，本发明实施例使用步骤S103中获得的复数图像以及对应的回波偏移时间采用水脂分离或定量方法进行水脂分离以及定量计算，计算得到每个像素中水信号和各频率峰值处的脂肪信号的各自比例。
进一步地，在本发明的一个实施例中，每个像素中水信号为：
Figure PCTCN2018116233-appb-000041
其中，ρ W,q以及ρ F,q分别代表像素点q处的水信号以及脂肪信号成分，f B,q为该处的场强 偏移量，
Figure PCTCN2018116233-appb-000042
分别为G种脂肪信号成分相对水信号的共振频率偏移，α g为第g种成分在脂肪信号中的比重。
具体而言，如上述介绍可以得到L个位于不同回波时刻的图像
Figure PCTCN2018116233-appb-000043
及相应的回波时刻偏移量
Figure PCTCN2018116233-appb-000044
即可以使用一般的水脂分离与定量方法进行处理。同时，考虑到实际情况中脂肪信号的共振谱具有多个峰值(常用3个或6个，以G表示)，并进一步考虑信号的
Figure PCTCN2018116233-appb-000045
衰减，公式8中的信号模型可以得到扩充，对某一回波时刻
Figure PCTCN2018116233-appb-000046
任一像素点处的信号表示为：
Figure PCTCN2018116233-appb-000047
其中，ρ W,q以及ρ F,q分别代表像素点q处的水信号以及脂肪信号成分，f B,q为该处的场强偏移量，
Figure PCTCN2018116233-appb-000048
分别为G种脂肪信号成分相对水信号的共振频率偏移，α g为第g种成分在脂肪信号中的比重，
Figure PCTCN2018116233-appb-000049
可以使用已知的不同脂肪信号比例统计值作为α g，此时仅需拟合ρ W,q，ρ F,q和f B,q，需求的回波数L≥3，亦可以当成未知数同时独立拟合得到ρ W,q，
Figure PCTCN2018116233-appb-000050
和f B,q，需求的回波数L≥G+2，通过归一化G种脂肪信号成分，可以得出不同种类脂肪信号的比例分布图α g。
本发明实施例不限制所采用的水脂分离方法，常用的水脂分离及定量计算方法有：IDEAL(iterative decomposition of water and fat with echo asymmetry and least-square estimation)方法，Hierarchical IDEAL方法，MAX-IDEAL方法，QPBO(quadratic pseudoboolean optimization)方法，QPBO-GC(quadratic pseudo boolean optimization graph cut)方法，Hernando-GC(Graph Cut)方法，ASR(a safest-first region-growing)方法，MRGS(multi-resolution golden section search)等。本发明实施例可通过但并不限于上述任意一种方法进行水脂分离以及定量计算。
在步骤S105中，根据水信号及每个频率峰值处的脂肪信号比例对三维幅值图像在相位编码方向进行加权平方和计算，以得到无变形的水相以及各脂肪相。
可以理解的是，如图2所示，本发明实施例利用计算得到的水脂信号比例对步骤S102中的三维幅值图像在相位编码方向进行加权平方和计算，从而得到无变形的水相以及各脂肪相。
进一步地，在本发明的一个实施例中，水相以及各脂肪相为：
Figure PCTCN2018116233-appb-000051
Figure PCTCN2018116233-appb-000052
其中，*代表共轭操作，不同峰值处的脂肪相依次计算为
Figure PCTCN2018116233-appb-000053
具体而言，为了进一步提高图像的信噪比，按照步骤S104中计算得出水、脂信号成分后，可以计算出各像素点水信号与各峰值处脂肪信号的定量比重，分别为
Figure PCTCN2018116233-appb-000054
使用该权重分别对步骤S102中所得到的三维幅值数据沿y方向按下式进行加权平方求和(SOS)操作，可以得到高信噪比的无变形的水相I W(s)以及脂肪相I F(s)：
Figure PCTCN2018116233-appb-000055
Figure PCTCN2018116233-appb-000056
其中，*代表共轭操作，不同峰值处的脂肪相可以依次计算为：
Figure PCTCN2018116233-appb-000057
综上，本发明实施例将PSF-EPI采集技术与水脂分离或定量方法相结合，无需额外的回波采集或偏移读出窗口采集，利用所采集的PSF-EPI数据本身，提取可用于水脂分离或定量的回波信号，使用前述的常用技术进行水脂分离和定量。本发明实施例方法继承了EPI的快速采集优势以及PSF-EPI的无变形、无T2*模糊效应、高信噪比等优点，可以在PSF-EPI的各种成像场合(如各种解剖结构像、扩散成像等)直接计算得出各水脂成分，用于脂肪定量分析或者去除脂肪信号对各种定量分析的影响，如扩散系数ADC和各向异性FA等。
根据本发明实施例提出的基于平面回波成像的磁共振水脂分离和定量方法，通过PSF编码的EPI采集无变形解剖结构图像或扩散图像，并将水脂分离或定量技术与其相结合，可以不需要额外采集数据或偏移数据读出窗口，便得到水脂分离后的图像或者脂肪百分比，直接同时计算得出水脂成分，从而有效提高分离和定量的适用性，采集效率高，无变形、无T2*模糊效应、高信噪比，简单易实现。
其次参照附图描述根据本发明实施例提出的基于平面回波成像的磁共振水脂分离和定量装置。
图6是本发明一个实施例的基于平面回波成像的磁共振水脂分离和定量装置的结构示意图。
如图6所示，该基于平面回波成像的磁共振水脂分离和定量装置10包括：采集模块100、 变换合并模块200、抽取模块300、分离与计算模块400和计算模块500。
其中，采集模块100用于采用无变形的点扩散函数编码的平面回波采集成像序列PSF-EPI进行数据采集，以得到多通道降采的k空间数据，并根据k空间数据通过tilted-CAIPI恢复得到三维k空间数据。变换合并模块200用于对三维k空间数据进行三维逆傅里叶变换，并通过通道合并得到三维幅值图像。抽取模块300用于从三维k空间数据抽取多个二维k空间数据，通过二维逆傅里叶变换得到复数图像并计算相应的回波偏移时间。分离与计算模块400用于使用复数图像和回波偏移时间进行水脂分离和定量计算，以得到每个像素中水信号和每个频率峰值处的脂肪信号比例。计算模块500用于根据水信号及每个频率峰值处的脂肪信号比例对三维幅值图像在相位编码方向进行加权平方和计算，以得到无变形的水相以及各脂肪相。本发明实施例的装置10通过PSF编码的EPI采集无变形解剖结构图像或扩散图像，并将水脂分离或定量技术与其相结合，从而有效提高分离和定量的适用性，采集效率高，无变形、无T2*模糊效应、高信噪比，简单易实现。
进一步地，在本发明的一个实施例中，采集模块100进一步用于将相位编码方向和点扩散函数编码方向的多通道降采的k空间数据同时进行重建得到三维k空间数据，其中，重建过程为：
Figure PCTCN2018116233-appb-000058
其中，d i(a,b,c)是在(a,b,c)处第i个线圈需要被插值的数据，d i′(a′,b′,c′)是在kernelK中采集到的数据，w为估计得到的插值权重系数，N c是线圈数目。
进一步地，在本发明的一个实施例中，三维幅值图像的三维幅值数据为：
I(s,y)＝ρ(s)H(s,y)，
Figure PCTCN2018116233-appb-000059
其中，H(s,y)为点扩散函数，图像域内s，y分别对应k空间域的k s，k y方向，ρ(s)为无变形的图像信号强度，δ为冲激函数，Δ(s)为图像变形所导致的点扩散函数偏移。
进一步地，在本发明的一个实施例中，其中，回波偏移时间为：
Figure PCTCN2018116233-appb-000060
其中，
Figure PCTCN2018116233-appb-000061
为基础的单次激发EPI序列的有效回波间隔，n为抽取的二维k空间数据在相位编码方向上的编码步数；每个像素中水信号为：
Figure PCTCN2018116233-appb-000062
其中，ρ W,q以及ρ F,q分别代表像素点q处的水信号以及脂肪信号成分，f B,q为该处的场强 偏移量，
Figure PCTCN2018116233-appb-000063
分别为G种脂肪信号成分相对水信号的共振频率偏移，α g为第g种成分在脂肪信号中的比重；水相以及各脂肪相为：
Figure PCTCN2018116233-appb-000064
Figure PCTCN2018116233-appb-000065
其中，*代表共轭操作，不同峰值处的脂肪相依次计算为
Figure PCTCN2018116233-appb-000066
需要说明的是，前述对基于平面回波成像的磁共振水脂分离和定量方法实施例的解释说明也适用于该实施例的基于平面回波成像的磁共振水脂分离和定量装置，此处不再赘述。
根据本发明实施例提出的基于平面回波成像的磁共振水脂分离和定量装置，通过PSF编码的EPI采集无变形解剖结构图像或扩散图像，并将水脂分离或定量技术与其相结合，可以不需要额外采集数据或偏移数据读出窗口，便得到水脂分离后的图像或者脂肪百分比，直接同时计算得出水脂成分，从而有效提高分离和定量的适用性，采集效率高，无变形、无T2*模糊效应、高信噪比，简单易实现。
此外，术语“第一”、“第二”仅用于描述目的，而不能理解为指示或暗示相对重要性或者隐含指明所指示的技术特征的数量。由此，限定有“第一”、“第二”的特征可以明示或者隐含地包括至少一个该特征。在本发明的描述中，“多个”的含义是至少两个，例如两个，三个等，除非另有明确具体的限定。
在本说明书的描述中，参考术语“一个实施例”、“一些实施例”、“示例”、“具体示例”、或“一些示例”等的描述意指结合该实施例或示例描述的具体特征、结构、材料或者特点包含于本发明的至少一个实施例或示例中。在本说明书中，对上述术语的示意性表述不必须针对的是相同的实施例或示例。而且，描述的具体特征、结构、材料或者特点可以在任一个或多个实施例或示例中以合适的方式结合。此外，在不相互矛盾的情况下，本领域的技术人员可以将本说明书中描述的不同实施例或示例以及不同实施例或示例的特征进行结合和组合。
尽管上面已经示出和描述了本发明的实施例，可以理解的是，上述实施例是示例性的，不能理解为对本发明的限制，本领域的普通技术人员在本发明的范围内可以对上述实施例进行变化、修改、替换和变型。

Claims (10)

  1. 一种基于平面回波成像的磁共振水脂分离和定量方法，其特征在于，包括以下步骤：
    使用无变形的点扩散函数编码的平面回波成像序列PSF-EPI进行数据采集，以得到多通道降采的k空间数据，并通过tilted-CAIPI恢复得到完整的三维k空间数据；
    对所述三维k空间数据进行三维逆傅里叶变换，并通过通道合并得到三维幅值图像；
    从所述三维k空间数据抽取多个二维k空间数据，通过二维逆傅里叶变换得到复数图像并计算相应回波偏移时间；
    使用所述复数图像和回波偏移时间进行水脂分离和定量计算，以得到每个像素中水信号和每个频率峰值处的脂肪信号比例；以及
    根据所述水信号及每个频率峰值处的脂肪信号比例对所述三维幅值图像在相位编码方向进行加权平方和计算，以得到无变形的水相以及各脂肪相。
  2. 根据权利要求1所述的基于平面回波成像的磁共振水脂分离和定量方法，其特征在于，所述根据tilted-CAIPI恢复三维k空间数据，进一步包括：
    将相位编码方向和所述点扩散函数编码方向的所述多通道降采的k空间数据同时进行重建得到所述三维k空间数据，其中，重建过程为：
    Figure PCTCN2018116233-appb-100001
    其中，d i(a,b,c)是在(a,b,c)处第i个线圈需要被插值的数据，d i′(a′,b′,c′)是在kernelK中采集到的数据，w为估计得到的插值权重系数，N c是线圈数目。
  3. 根据权利要求1所述的基于平面回波成像的磁共振水脂分离和定量方法，其特征在于，所述三维幅值图像的三维幅值数据为：
    I(s,y)＝ρ(s)H(s,y)，
    Figure PCTCN2018116233-appb-100002
    其中，H(s,y)为点扩散函数，图像域内s，y分别对应k空间域的k s，k y方向，ρ(s)为无变形的图像信号强度，δ为冲激函数，Δ(s)为图像变形所导致的点扩散函数偏移。
  4. 根据权利要求1所述的基于平面回波成像的磁共振水脂分离和定量方法，其特征在于，所述回波偏移时间为：
    t n＝n×t yesp，
    其中，t yesp为基础的单次激发EPI序列的有效回波间隔，n为所抽取的二维k空间数据在相位编码方向的编码步数。
  5. 根据权利要求1所述的基于平面回波成像的磁共振水脂分离和定量方法，其特征在 于，所述每个像素中水信号为：
    Figure PCTCN2018116233-appb-100003
    其中，ρ W,q以及ρ F,q分别代表像素点q处的水信号以及脂肪信号成分，f B,q为该处的场强偏移量，
    Figure PCTCN2018116233-appb-100004
    分别为G种脂肪信号成分相对水信号的共振频率偏移，α g为第g种成分在脂肪信号中的比重。
  6. 根据权利要求1所述的基于平面回波成像的磁共振水脂分离和定量方法，其特征在于，所述水相以及各脂肪相为：
    Figure PCTCN2018116233-appb-100005
    Figure PCTCN2018116233-appb-100006
    其中，*代表共轭操作，不同峰值处的脂肪相依次计算为
    Figure PCTCN2018116233-appb-100007
  7. 一种基于平面回波成像的磁共振水脂分离和定量装置，其特征在于，包括：
    采集模块，用于根据采用无变形的点扩散函数编码的平面回波采集成像序列PSF-EPI进行数据采集，以得到多通道降采的k空间数据，并根据所述k空间数据通过tilted-CAIPI恢复得到三维k空间数据；
    变换合并模块，用于对所述三维k空间数据进行三维逆傅里叶变换，并通过通道合并得到三维幅值图像；
    抽取模块，用于从所述三维k空间数据抽取多个二维k空间数据，通过二维逆傅里叶变换得到复数图像并计算相应回波偏移时间；
    分离与计算模块，用于使用所述复数图像和回波偏移时间进行水脂分离和定量计算，以得到每个像素中水信号和每个频率峰值处的脂肪信号比例；以及
    计算模块，用于根据所述水信号及每个频率峰值处的脂肪信号比例对所述三维幅值图像在相位编码方向进行加权平方和计算，以得到无变形的水相以及各脂肪相。
  8. 根据权利要求7所述的基于平面回波成像的磁共振水脂分离和定量装置，其特征在于，所述采集模块进一步用于将相位编码方向和所述点扩散函数编码方向的所述多通道降采的k空间数据同时进行重建得到所述三维k空间数据，其中，重建过程为：
    Figure PCTCN2018116233-appb-100008
    其中，d i(a,b,c)是在(a,b,c)处第i个线圈需要被插值的数据，d i′(a′,b′,c′)是在kernelK中采集到的数据，w为估计得到的插值权重系数，N c是线圈数目。
  9. 根据权利要求7所述的基于平面回波成像的磁共振水脂分离和定量装置，其特征在于，所述三维幅值图像的三维幅值数据为：
    I(s,y)＝ρ(s)H(s,y)，
    Figure PCTCN2018116233-appb-100009
    其中，H(s,y)为点扩散函数，图像域内s，y分别对应k空间域的k s，k y方向，ρ(s)为无变形的图像信号强度，δ为冲激函数，Δ(s)为图像变形所导致的点扩散函数偏移。
  10. 根据权利要求7所述的基于平面回波成像的磁共振水脂分离和定量装置，其特征在于，其中，
    所述回波偏移时间为：
    t n＝n×t yesp，
    其中，t yesp为基础的单次激发EPI序列的有效回波间隔，n为所抽取的二维k空间数据在相位编码方向的编码步数；
    所述每个像素中水信号为：
    Figure PCTCN2018116233-appb-100010
    其中，ρ W,q以及ρ F,q分别代表像素点q处的水信号以及脂肪信号成分，f B,q为该处的场强偏移量，
    Figure PCTCN2018116233-appb-100011
    分别为G种脂肪信号成分相对水信号的共振频率偏移，α g为第g种成分在脂肪信号中的比重；
    所述水相以及各脂肪相为：
    Figure PCTCN2018116233-appb-100012
    Figure PCTCN2018116233-appb-100013
    其中，*代表共轭操作，不同峰值处的脂肪相依次计算为
    Figure PCTCN2018116233-appb-100014
PCT/CN2018/116233 2018-07-02 2018-11-19 基于平面回波成像的磁共振水脂分离和定量方法及装置 WO2020006959A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201810707718.2A CN109115820B (zh) 2018-07-02 2018-07-02 基于平面回波成像的磁共振水脂分离和定量方法及装置
CN201810707718.2 2018-07-02

Publications (1)

Publication Number Publication Date
WO2020006959A1 true WO2020006959A1 (zh) 2020-01-09

Family

ID=64822021

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/116233 WO2020006959A1 (zh) 2018-07-02 2018-11-19 基于平面回波成像的磁共振水脂分离和定量方法及装置

Country Status (2)

Country Link
CN (1) CN109115820B (zh)
WO (1) WO2020006959A1 (zh)

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111025210B (zh) * 2019-12-20 2022-02-01 东软医疗系统股份有限公司 磁共振成像方法、装置、电子设备、存储介质
CN111352054B (zh) * 2020-03-31 2020-10-13 浙江大学 一种振荡梯度准备的3d梯度自旋回波成像方法及设备
CN111990997B (zh) * 2020-08-25 2022-10-28 李友 一种基于反转恢复实图重建的水脂分离方法及系统
CN112156383B (zh) * 2020-08-31 2022-10-11 上海沈德医疗器械科技有限公司 一种基于k空间能谱分析的磁共振温度修正方法及系统
CN112763955B (zh) * 2020-12-31 2022-12-20 苏州朗润医疗系统有限公司 一种基于图割算法的磁共振图像水脂分离方法
CN115469254A (zh) * 2021-05-18 2022-12-13 上海联影医疗科技股份有限公司 一种磁共振定量成像方法
CN117233676B (zh) * 2023-11-15 2024-03-26 之江实验室 一种回波时间依赖的磁共振扩散成像信号生成方法和装置

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4739266A (en) * 1985-12-12 1988-04-19 U.S. Philips Corporation MR tomography method and apparatus for performing the method
US5187439A (en) * 1989-11-10 1993-02-16 U.S. Philips Corp. Magnetic resonance tomography method for generating separate fat and water images, and apparatus for performing the method
CN1363043A (zh) * 2000-01-27 2002-08-07 Ge医疗系统环球技术有限公司 水和脂肪的分离图像的形成方法,磁共振成像设备,参考峰的相位测定方法和参考峰的位置测定方法
US20040056660A1 (en) * 2000-12-28 2004-03-25 Yumiko Yatsui Magnetic resonance imaging device and method
US20130107006A1 (en) * 2011-10-28 2013-05-02 New York University Constructing a 3-dimensional image from a 2-dimensional image and compressing a 3-dimensional image to a 2-dimensional image
CN103505210A (zh) * 2012-06-28 2014-01-15 西门子(深圳)磁共振有限公司 一种实现水脂分离的磁共振成像方法和装置
CN103608693A (zh) * 2011-04-21 2014-02-26 皇家飞利浦有限公司 具有用于脂肪抑制的化学位移编码的对比度增强磁共振血管造影

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101427918A (zh) * 2007-11-06 2009-05-13 中国人民解放军第二军医大学 一种磁共振成像方法及其应用
KR101773617B1 (ko) * 2011-10-28 2017-09-01 가천의과학대학교 산학협력단 Ｅｐｉ 영상의 왜곡 보정 방법 및 그를 이용한 ｍｒｉ 장치
CN102928796B (zh) * 2012-09-28 2014-12-24 清华大学 快速扩散磁共振成像和重建方法
CN104597420B (zh) * 2015-02-02 2018-01-16 清华大学 基于多次激发的磁共振扩散成像方法
CN105548927B (zh) * 2015-08-06 2018-11-09 清华大学 基于多层同时激发的多次激发的磁共振扩散成像方法
CN106780643B (zh) * 2016-11-21 2019-07-26 清华大学 磁共振多次激发扩散成像运动矫正方法
US10520573B2 (en) * 2016-11-23 2019-12-31 General Electric Company System and method for performing wave-encoded magnetic resonance imaging of an object
CN107064843B (zh) * 2017-04-25 2019-12-03 上海联影医疗科技有限公司 一种磁共振成像方法及磁共振成像设备

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4739266A (en) * 1985-12-12 1988-04-19 U.S. Philips Corporation MR tomography method and apparatus for performing the method
US5187439A (en) * 1989-11-10 1993-02-16 U.S. Philips Corp. Magnetic resonance tomography method for generating separate fat and water images, and apparatus for performing the method
CN1363043A (zh) * 2000-01-27 2002-08-07 Ge医疗系统环球技术有限公司 水和脂肪的分离图像的形成方法,磁共振成像设备,参考峰的相位测定方法和参考峰的位置测定方法
US20040056660A1 (en) * 2000-12-28 2004-03-25 Yumiko Yatsui Magnetic resonance imaging device and method
CN103608693A (zh) * 2011-04-21 2014-02-26 皇家飞利浦有限公司 具有用于脂肪抑制的化学位移编码的对比度增强磁共振血管造影
US20130107006A1 (en) * 2011-10-28 2013-05-02 New York University Constructing a 3-dimensional image from a 2-dimensional image and compressing a 3-dimensional image to a 2-dimensional image
CN103505210A (zh) * 2012-06-28 2014-01-15 西门子(深圳)磁共振有限公司 一种实现水脂分离的磁共振成像方法和装置

Also Published As

Publication number Publication date
CN109115820A (zh) 2019-01-01
CN109115820B (zh) 2020-03-24

Similar Documents

Publication Publication Date Title
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
US7348776B1 (en) Motion corrected magnetic resonance imaging
Küstner et al. MR image reconstruction using a combination of compressed sensing and partial Fourier acquisition: ESPReSSo
US20230360179A1 (en) Machine learning processing of contiguous slice image data
US20140376794A1 (en) Correlation Imaging For Multi-Scan MRI With Multi-Channel Data Acquisition
US20190041479A1 (en) Dynamic magnetic resonance imaging
Ertürk et al. Denoising MRI using spectral subtraction
Dar et al. Synergistic reconstruction and synthesis via generative adversarial networks for accelerated multi-contrast MRI
CA2964945A1 (en) System and method for producing distortion free magnetic resonance images using dual-echo echo-planar imaging
WO2022183988A1 (en) Systems and methods for magnetic resonance image reconstruction with denoising
US20240036141A1 (en) Joint k-space and image-space reconstruction imaging method and device
Li et al. An adaptive directional Haar framelet-based reconstruction algorithm for parallel magnetic resonance imaging
Zhang et al. Deep learning based multiplexed sensitivity-encoding (DL-MUSE) for high-resolution multi-shot DWI
US20150042329A1 (en) Dynamic contrast enhanced magnetic resonance imaging with high spatial-temporal resolution
Chiou et al. A simple simultaneous geometric and intensity correction method for echo-planar imaging by EPI-based phase modulation
CN110286344B (zh) 一种快速磁共振可变分辨率成像方法、系统和可读介质
EP2793040A1 (en) Method for processing phase information, particularly for susceptibility weighted imaging, obtained from PROPELLER measurements
Zhong et al. Understanding aliasing effects and their removal in SPEN MRI: A k‐space perspective
WO2022027419A1 (zh) 磁共振成像方法、装置及计算机存储介质
Akçakaya et al. Subject-specific convolutional neural networks for accelerated magnetic resonance imaging
He et al. Compressed sensing-based simultaneous recovery of magnitude and phase MR images via dual trigonometric sparsity
Wang et al. Multiple B-value model-based residual network (MORN) for accelerated high-resolution diffusion-weighted imaging
Bhattacharya et al. Denoising and deinterleaving of EPSI data using structured low-rank matrix recovery
Xu et al. Integrating parallel imaging with generalized series for accelerated dynamic imaging
Poot APIR-Net: Autocalibrated Parallel Imaging Reconstruction Using a Neural Network

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: 18925393

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18925393

Country of ref document: EP

Kind code of ref document: A1