Diffusion magnetic resonance imaging method and system for three-dimensional plane echo imaging

Abstract

The invention provides a diffusion magnetic resonance imaging method and system for three-dimensional plane echo imaging, and the method comprises the steps: constructing a block Hankel matrix through first k space data formed through multiple times of excitation, constructing a target Hankel matrix, and restoring the target Hankel matrix to a k space to form second k space data; converting the second k-space data into first image data; and extracting an average value of the amplitude matrix of the first image data, updating the average value to form second image data, converting the second image data into third k-space data, converting the third k-space data into third image data after data convergence, and reconstructing a magnetic resonance image based on the third image data. DWI imaging excited for multiple times is effectively corrected, and the problem of artifacts caused by phase inconsistency of diffusion magnetic resonance imaging data excited for multiple times is solved; the problems that under the ultralow field condition, the number of channels is small, and the spatial difference of coil sensitivity cannot meet the requirement of parallel imaging image reconstruction are solved, and the imaging quality is improved.

Description

Technical Field

[0001] The present invention relates to the field of magnetic resonance imaging technology, and in particular to a diffusion magnetic resonance imaging method and system for three-dimensional echo plane imaging. Background Art

[0002] Diffusion MRI under ultra-low field conditions is extremely challenging. On the one hand, the signal-to-noise ratio is severely limited under ultra-low field conditions. On the other hand, due to the great difficulty of decoupling between channels under ultra-low fields, the limited number of coil channels, and the fact that the coil sensitivity maps (Coil Sensitivity Maps, CSM) of each channel do not vary significantly in space, it is difficult to provide effective additional spatial encoding through CSM. All of these make it impossible to use traditional parallel imaging-based methods for diffusion MRI. Summary of the invention

[0003] Based on the above content, the present invention provides a diffusion magnetic resonance imaging method and system for three-dimensional echo planar imaging, aiming to solve the technical problems of the prior art such as the limited signal-to-noise ratio of diffusion magnetic resonance imaging under low-field conditions.

[0004] A diffusion magnetic resonance imaging method for three-dimensional echo planar imaging, comprising:

[0005] Step A1, filling the k-space with the magnetic resonance data acquired by multiple excitations to form three-dimensional first k-space data, and constructing a block Hankel matrix according to the first k-space data;

[0006] Step A2, constructing a target Hankel matrix according to the block Hankel matrix, and restoring the target Hankel matrix to the k-space to form second k-space data;

[0007] Step A3, converting the second k-space data into first image data;

[0008] Step A4, extracting the amplitude matrix and phase matrix of the image unit corresponding to each excitation acquisition in the first image data, and calculating the average value of each amplitude matrix to form a first mean matrix;

[0009] Step A5, replacing the amplitude matrix of each image unit with the first mean matrix to update the amplitude of the image unit, and retaining the phase matrix to form second image data;

[0010] Step A6, converting the second image data into k-space to form third k-space data;

[0011] Step A7, judging whether the third k-space data meets the stop condition based on the preset rule:

[0012] If yes, go to step A8;

[0013] If not, executing step A2 after constructing a block Hankel matrix according to the third k-space data;

[0014] Step A8: convert the third k-space data into third image data, and reconstruct a magnetic resonance image based on the third image data.

[0015] Furthermore, in step A1, the magnetic resonance data includes imaging echo data and navigation echo data. During the acquisition of imaging echo data, the intra-layer phase encoding direction adopts a linear traversal method, and the inter-layer phase encoding direction adopts a random traversal method. During the acquisition of navigation echo data, both the intra-layer phase encoding direction and the inter-layer phase encoding direction adopt a linear traversal method.

[0016] Furthermore, in step A1, the first k-space data formed by the imaging echo data is used to construct a first matrix, the first k-space data formed by the navigation echo data is used to construct a second matrix, and the first matrix and the second matrix are spliced ​​to form a block Hankel matrix.

[0017] Furthermore, in step A1, the first k-space data formed by the imaging echo data is traversed and vectorized through a three-dimensional sliding window, and each data acquired by the three-dimensional sliding window forms a column vector, and each column vector is arranged in sequence along the row direction to form the first matrix.

[0018] Furthermore, in step A1, the first k-space data formed by the navigation echo data is traversed and vectorized through a three-dimensional sliding window, and each data acquired by the three-dimensional sliding window forms a column vector, and each column vector is arranged in sequence along the row direction to form the second matrix.

[0019] Furthermore, the first matrix and the second matrix are concatenated along the row direction to form a block Hankel matrix.

[0020] Furthermore, the second matrix is ​​multiplied by a predetermined coefficient to obtain a third matrix, and the first matrix and the third matrix are concatenated along a row direction to form a block Hankel matrix.

[0021] Furthermore, in step A8, the average value of the amplitude matrix of the image unit corresponding to each excitation acquisition in the third image data is calculated to obtain a second mean matrix, and the magnetic resonance image is reconstructed according to the second mean matrix.

[0022] A diffusion magnetic resonance imaging system for three-dimensional echo planar imaging, used to perform the diffusion magnetic resonance imaging method for three-dimensional echo planar imaging, comprising:

[0023] A data acquisition module, used for filling the k-space with the magnetic resonance data acquired by multiple excitations to form three-dimensional first k-space data;

[0024] A first construction module, connected to the data acquisition module, is used to construct a block Hankel matrix according to the first k-space data;

[0025] A second building module, connected to the first building module, is used to build a target Hankel matrix according to the block Hankel matrix;

[0026] A space recovery module, connected to the second construction module, used for recovering the target Hankel matrix to the k-space to form second k-space data;

[0027] A first conversion module, connected to the space recovery module, for converting the second k-space data into first image data;

[0028] An information extraction module, connected to the first conversion module, is used to extract the amplitude matrix and phase matrix of the image unit corresponding to each excitation acquisition in the first image data, and calculate the average value of the amplitude matrix to form a first mean matrix;

[0029] An image updating module, connected to the information extraction module, is used to replace the amplitude matrix of each image unit with the first mean matrix to update the amplitude of the image unit, and retain the phase matrix to form second image data;

[0030] A second conversion module, connected to the image update module, is used to convert the second image data into k-space to form third k-space data;

[0031] A stop judgment module, connected to the second conversion module, for judging whether the third k-space data meets the stop condition based on a preset rule, and obtaining a first judgment result;

[0032] The first construction module is also connected to the stop judgment module, and is used to construct a block Hankel matrix according to the third k-space data when the judgment result is that the stop condition is not met;

[0033] The image reconstruction module is connected to the stop judgment module and is used for converting the third k-space data into third image data when the judgment result satisfies the stop condition, and reconstructing the magnetic resonance image based on the third image data.

[0034] Furthermore, the magnetic resonance data includes imaging echo data and navigation echo data; during the imaging echo data acquisition process, the data acquisition module adopts a linear traversal method in the intra-layer phase encoding direction and a random traversal method in the inter-layer phase encoding direction; during the navigation echo data acquisition process, the data acquisition module adopts a linear traversal method in both the intra-layer phase encoding direction and the inter-layer phase encoding direction.

[0035] The beneficial technical effects of the present invention are: by acquiring images based on three-dimensional echo planar imaging technology, building a block Hankel matrix according to k-space data, reconstructing images in k-space, effectively correcting DWI imaging of multiple excitations, and solving the problem of artifacts caused by phase inconsistency of diffusion magnetic resonance imaging data of multiple excitations; solving the problem that the number of channels is small under ultra-low field conditions and the spatial difference of coil sensitivity cannot meet the requirements of parallel imaging image reconstruction, thereby improving image quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 A flowchart of the steps of a diffusion magnetic resonance imaging method for three-dimensional echo planar imaging according to the present invention;

[0037] Figure 2 A schematic diagram of a three-dimensional diffusion magnetic resonance imaging sequence with multiple excitations in a diffusion magnetic resonance imaging method for three-dimensional echo-planar imaging according to the present invention;

[0038] Figure 3 A schematic diagram of constructing a block Hankel matrix in a diffusion magnetic resonance imaging method for three-dimensional echo planar imaging according to the present invention;

[0039] Figure 4 The module schematic diagram of a diffusion magnetic resonance imaging system for three-dimensional echo planar imaging of the present invention. DETAILED DESCRIPTION

[0040] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0042] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.

[0043] See also Figure 1 The present invention provides a diffusion magnetic resonance imaging method for three-dimensional echo planar imaging, comprising:

[0044] Step A1, filling the k-space with the magnetic resonance data acquired by multiple excitations to form three-dimensional first k-space data, and constructing a block Hankel matrix according to the first k-space data;

[0045] Step A2, constructing a target Hankel matrix according to the block Hankel matrix, and restoring the target Hankel matrix to the k-space to form second k-space data;

[0046] Step A3, converting the second k-space data into first image data;

[0047] Step A4, extracting the amplitude matrix and phase matrix of the image unit corresponding to each excitation acquisition in the first image data, and calculating the average value of each amplitude matrix to form a first mean matrix;

[0048] Step A5, replacing the amplitude matrix of each image unit with the first mean matrix to update the amplitude of the image unit, and retaining the phase matrix to form second image data;

[0049] Step A6, converting the second image data into k-space to form third k-space data;

[0050] Step A7, judging whether the third k-space data meets the stop condition based on the preset rule:

[0051] If yes, go to step A8;

[0052] If not, executing step A2 after constructing a block Hankel matrix according to the third k-space data;

[0053] Step A8: convert the third k-space data into third image data, and reconstruct a magnetic resonance image based on the third image data.

[0054] Because the signal-to-noise ratio is extremely low under ultra-low field conditions, and the algorithms that perform calculations in k-space are often more stable than those that perform calculations in the image domain, the invention was changed to perform image reconstruction in k-space. By acquiring images based on three-dimensional echo planar imaging technology, building a block Hankel matrix based on k-space data, and reconstructing images in k-space, the DWI imaging of multiple excitations is effectively corrected, and the problem of artifacts caused by inconsistent phases of diffusion magnetic resonance imaging data of multiple excitations is solved; the problem of a small number of channels under ultra-low field conditions and spatial differences in coil sensitivity cannot meet the requirements of parallel imaging image reconstruction, and the image quality is improved.

[0055] See also Figure 2 Furthermore, in step A1, the magnetic resonance data includes imaging echo data and navigation echo data. During the acquisition of imaging echo data, the intra-layer phase encoding direction adopts a linear traversal method, and the inter-layer phase encoding direction adopts a random traversal method. During the acquisition of navigation echo data, both the intra-layer phase encoding direction and the inter-layer phase encoding direction adopt a linear traversal method.

[0056] In the present invention, the kx direction represents the frequency encoding direction, the ky direction represents the intra-layer encoding direction, and the kz direction represents the inter-layer encoding direction. For diffusion magnetic resonance imaging with multiple excitations, a portion of the k space is filled after each excitation. In order to maximize the incoherence between the filled k space positions of different excitations, the present invention abandons the method of linear traversal in the ky-kz directions of the k space in traditional three-dimensional imaging, and instead uses linear traversal in the ky direction, but random traversal in the kz direction; data of different k space positions are collected after different excitations until the entire k space is filled; this can not only ensure the incoherence between different excitations, but also avoid the generation of various artifacts.

[0057] Due to the influence of physiological factors such as cerebrospinal fluid pulsation, there are differences in the phase of diffusion MRI echo data from different excitations. If this phase difference cannot be eliminated during the image reconstruction process, the final reconstructed image will have obvious artifacts. In order to record this phase difference during the imaging echo data acquisition process, it is necessary to introduce the data acquisition part of the navigation echo in the sequence design. The navigation echo is linearly traversed in both the ky and kz directions.

[0058] In this sampling mode, there is no geometric distortion in the kz direction for imaging echo data; this is because the data at different kz positions are collected at different excitations, and as long as the ky positions are consistent, the corresponding echo times are equal, so there is no geometric distortion in the z direction. However, for navigation echo data, there is a certain geometric distortion in the z direction, and it may be necessary to perform geometric distortion correction on the navigation echo data to reduce the impact of the inconsistent geometric distortion between the imaging echo data and the navigation echo data.

[0059] As for the ky direction, both imaging echo data and navigation echo data have the problem of geometric distortion. In the process of collecting navigation echo data, it is a better choice to traverse the ky direction first and then the kz direction. This is because under this scheme, the geometric distortion of imaging echo data and navigation echo data in the ky direction can be consistent.

[0060] Furthermore, in step A1, the first k-space data formed by the imaging echo data is used to construct a first matrix, the first k-space data formed by the navigation echo data is used to construct a second matrix, and the first matrix and the second matrix are spliced ​​to form a block Hankel matrix.

[0061] See also Figure 3 Furthermore, in the step A1, the first k-space data formed by the imaging echo data is traversed and vectorized through a three-dimensional sliding window, and each data acquired by the three-dimensional sliding window forms a column vector, and each column vector is arranged in sequence along the row direction to form the first matrix.

[0062] Specifically, for the two-dimensional k-space data in the frequency encoding direction and the intra-layer phase encoding direction, the data extracted by each two-dimensional sliding window forms a column vector, and the latter column vector is arranged after the former column vector, that is, the column vector formed by the data extracted by the first two-dimensional sliding window is the first column, the column vector formed by the data extracted by the second two-dimensional sliding window is the second column, and the data extracted by the n-th two-dimensional sliding window forms the n-th column. There is an inter-layer encoding direction, so the two-dimensional k-space data of each layer constitutes a single-block matrix, and these single-block matrices are arranged in the column direction, that is, the first layer of two-dimensional space data is arranged in the column direction to form a single-block matrix of several rows, that is, the first matrix is ​​formed.

[0063] Furthermore, in step A1, the first k-space data formed by the navigation echo data is traversed and vectorized through a three-dimensional sliding window, and each data acquired by the three-dimensional sliding window forms a column vector, and each column vector is arranged in sequence along the row direction to form the second matrix.

[0064] The way in which the navigator echo data forms the second matrix is ​​the same as the way in which the imaging echo data constructs the second matrix.

[0065] In the present invention, navigation echo data and imaging echo data are two types of MR signal data (magnetic resonance data) with different purposes and processing methods. Navigation echo is a special technology that is mainly used to monitor and correct image artifacts caused by the subject's movement (such as physiological movements such as breathing and heartbeat). Its working principle is to insert some additional, non-imaging RF pulses and gradients into the imaging sequence to collect a small amount of signals representing specific anatomical locations (usually areas with more obvious movements such as the thorax or abdomen). These signals are called navigation echoes. They are not used directly to generate images, but are used in real time or afterwards to evaluate the subject's movement state during the imaging process. By analyzing the navigation echo data, the system can determine the degree and type of movement, and accordingly make corresponding corrections to the subsequent imaging data, such as adjusting the image data through a motion compensation algorithm to reduce or eliminate motion artifacts and improve image quality.

[0066] In contrast, imaging echo data refers to MR signal data (magnetic resonance data) that is directly used to construct MRI images. In an MRI sequence, when radio frequency pulses excite hydrogen protons in tissues, the protons emit signals that are collected by receiving coils at different echo times to form a series of imaging echoes. These echo signals contain the anatomical structure and physiological information of the tissues, and are converted into the MRI images we see after being processed by Fourier transform and other image reconstruction algorithms. Imaging echo data directly reflects tissue characteristics and is the basis for clinical diagnosis and scientific research.

[0067] Furthermore, the first matrix and the second matrix are concatenated along the row direction to form a block Hankel matrix.

[0068] Furthermore, the second matrix is ​​multiplied by a predetermined coefficient to obtain a third matrix, and the first matrix and the third matrix are concatenated along a row direction to form a block Hankel matrix.

[0069] That is to say, in order to reduce the distortion problem of imaging echo data and navigation echo data in the kz direction, the first k-space data formed by the navigation echo data can be multiplied by a predetermined coefficient less than 1 to reduce the influence of the distortion. Further, as a preferred manner, in step A1, the first k-space data formed by the navigation echo data is subjected to distortion correction to form a second matrix, and then the first matrix and the second matrix are spliced ​​along the row direction to form a block Hankel matrix.

[0070] That is to say, due to the inconsistency of geometric distortion in the z direction between the imaging echo data and the navigation echo data (i.e., the imaging echo data has no geometric distortion, while the navigation echo data has certain geometric distortion). Although image reconstruction in k-space is relatively insensitive to such inconsistency of geometric distortion, in order to further improve the quality of reconstructed images, the navigation echo data can be calibrated for geometric distortion, or when constructing the block Hankel matrix, it can be multiplied by a coefficient less than 1 to reduce the impact of geometric distortion on the reconstruction process.

[0071] In step A2, the process of constructing a target Hankel matrix according to the block Hankel matrix includes: performing singular value decomposition on the block Hankel matrix, then performing rank truncation, and forming the target Hankel matrix according to the diagonal matrix obtained by rank truncation.

[0072] In step A7, the process of constructing a block Hankel matrix according to the third k-space data is the same as the process of constructing a target Hankel matrix according to the block Hankel matrix in step A2, except that the block Hankel matrix is ​​constructed according to the third k-space data.

[0073] In step A6, after the second image data is converted into k-space, the data at the position corresponding to the k-space formed by the conversion of the second image data is replaced with the data actually collected in the first k-space data to form third k-space data. The actual data is replaced to ensure that the actually collected data remains unchanged.

[0074] In step A3, the second k-space data is converted into first image data by inverse Fourier transform.

[0075] In step A6, the second image data is converted into k-space by Fourier transform to form third k-space data;

[0076] In step A8 , the third k-space data is converted into third image data by Fourier transformation.

[0077] Furthermore, in step A8, the average value of the amplitude matrix of the image unit corresponding to each excitation acquisition in the third image data is calculated to obtain a second mean matrix, and the magnetic resonance image is reconstructed according to the second mean matrix.

[0078] See also Figure 4 The present invention also provides a diffusion magnetic resonance imaging system for three-dimensional echo planar imaging, which is used to perform the diffusion magnetic resonance imaging method for three-dimensional echo planar imaging, comprising:

[0079] A data acquisition module (1) is used to fill the k-space with magnetic resonance data acquired through multiple excitations to form three-dimensional first k-space data;

[0080] A first construction module (2), connected to the data acquisition module (1), is used to construct a block Hankel matrix according to the first k-space data;

[0081] A second building module (3), connected to the first building module (2), is used to build a target Hankel matrix according to the block Hankel matrix;

[0082] A space recovery module (4), connected to the second construction module (3), used for recovering the target Hankel matrix to the k-space to form second k-space data;

[0083] A first conversion module (5), connected to the space recovery module (4), is used to convert the second k-space data into first image data;

[0084] An information extraction module (6), connected to the first conversion module (5), is used to extract the amplitude matrix and phase matrix of the image unit corresponding to each excitation acquisition in the first image data, and calculate the average value of the amplitude matrix to form a first mean matrix;

[0085] An image updating module (7) is connected to the information extraction module (6) and is used to replace the amplitude matrix of each image unit with the first mean matrix to update the amplitude of the image unit and retain the phase matrix to form second image data;

[0086] A second conversion module (8), connected to the image updating module (7), is used to convert the second image data into k-space to form third k-space data;

[0087] A stop judgment module (9), connected to the second conversion module (8), is used to judge whether the third k-space data meets the stop condition based on a preset rule, and obtain a judgment result;

[0088] The first construction module (2) is also connected to the stop judgment module (9) and is used to construct a block Hankel matrix according to the third k-space data when the judgment result is that the stop condition is not met;

[0089] The image reconstruction module (10) is connected to the stop judgment module (9) and is used to convert the third k-space data into third image data when the judgment result satisfies the stop condition, and reconstruct the magnetic resonance image based on the third image data.

[0090] Because the signal-to-noise ratio is extremely low under ultra-low field conditions, and the algorithms that perform calculations in k-space are often more stable than those that perform calculations in the image domain, the invention was changed to perform image reconstruction in k-space. By acquiring images based on three-dimensional echo planar imaging technology, building a block Hankel matrix based on k-space data, and reconstructing images in k-space, the DWI imaging of multiple excitations is effectively corrected, and the problem of artifacts caused by inconsistent phases of diffusion magnetic resonance imaging data of multiple excitations is solved; the problem of a small number of channels under ultra-low field conditions and spatial differences in coil sensitivity cannot meet the requirements of parallel imaging image reconstruction, and the image quality is improved.

[0091] Furthermore, the magnetic resonance data includes imaging echo data and navigation echo data; during the imaging echo data acquisition process, the data acquisition module (1) uses a linear traversal method for the intra-layer phase encoding direction and a random traversal method for the inter-layer phase encoding direction; during the navigation echo data acquisition process, the data acquisition module uses a linear traversal method for both the intra-layer phase encoding direction and the inter-layer phase encoding direction.

[0092] In the present invention, the kx direction represents the frequency encoding direction, the ky direction represents the intra-layer encoding direction, and the kz direction represents the inter-layer encoding direction. For diffusion magnetic resonance imaging with multiple excitations, a portion of the k space is filled after each excitation. In order to maximize the incoherence between the filled k space positions of different excitations, the present invention abandons the method of linear traversal in the ky-kz directions of the k space in traditional three-dimensional imaging, and instead uses linear traversal in the ky direction, but random traversal in the kz direction; data of different k space positions are collected after different excitations until the entire k space is filled; this can not only ensure the incoherence between different excitations, but also avoid the generation of various artifacts.

[0093] The above are only preferred embodiments of the present invention, and are not intended to limit the implementation methods and protection scope of the present invention. Those skilled in the art should be aware that all solutions obtained by equivalent substitutions and obvious changes made using the description and illustrations of the present invention should be included in the protection scope of the present invention.

Claims

1. A diffusion magnetic resonance imaging method for three-dimensional echo planar imaging, characterized in that: include: Step A1, filling k-space with magnetic resonance data acquired by multiple excitations to form three-dimensional first k-space data, and constructing a block Hankel matrix according to the first k-space data; Step A2, constructing a target Hankel matrix according to the block Hankel matrix, and restoring the target Hankel matrix to k-space to form second k-space data; Step A3, converting the second k-space data into first image data; Step A4, extracting the amplitude matrix and phase matrix of the image unit corresponding to each excitation acquisition in the first image data, and calculating the average value of each amplitude matrix to form a first mean matrix; Step A5, replacing the amplitude matrix of each image unit with the first mean matrix to update the amplitude of the image unit, and retaining the phase matrix to form second image data; Step A6, converting the second image data into k-space to form third k-space data; Step A7, judging whether the third k-space data meets the stop condition based on a preset rule: If yes, go to step A8; If not, construct a block Hankel matrix according to the third k-space data and then execute step A2; Step A8: convert the third k-space data into third image data, and reconstruct a magnetic resonance image based on the third image data.

2. A diffusion magnetic resonance imaging method for three-dimensional echo planar imaging as claimed in claim 1, characterized in that: In step A1, the magnetic resonance data includes imaging echo data and navigation echo data. During the acquisition of the imaging echo data, the intra-layer phase encoding direction adopts a linear traversal method, and the inter-layer phase encoding direction adopts a random traversal method. During the acquisition of the navigation echo data, the intra-layer phase encoding direction and the inter-layer phase encoding direction both adopt a linear traversal method.

3. A diffusion magnetic resonance imaging method for three-dimensional echo planar imaging as claimed in claim 2, characterized in that: In the step A1, a first matrix is ​​constructed using the first k-space data formed by the imaging echo data, a second matrix is ​​constructed using the first k-space data formed by the navigation echo data, and the first matrix and the second matrix are spliced ​​to form the block Hankel matrix.

4. A diffusion magnetic resonance imaging method for three-dimensional echo planar imaging as claimed in claim 3, characterized in that: In step A1, the first k-space data formed by the imaging echo data is traversed and vectorized through a three-dimensional sliding window, and each data acquired by the three-dimensional sliding window forms a column vector, and each column vector is arranged in sequence along the row direction to form the first matrix.

5. A diffusion magnetic resonance imaging method for three-dimensional echo planar imaging as claimed in claim 3, characterized in that: In step A1, the first k-space data formed by the navigation echo data is traversed and vectorized through a three-dimensional sliding window, and each data acquired by the three-dimensional sliding window forms a column vector, and each column vector is arranged in sequence along the row direction to form the second matrix.

6. A diffusion magnetic resonance imaging method for three-dimensional echo planar imaging as claimed in claim 3, characterized in that: In the step A1, the first matrix and the second matrix are concatenated along the row direction to form the block Hankel matrix.

7. A diffusion magnetic resonance imaging method for three-dimensional echo planar imaging as claimed in claim 3, characterized in that: In the step A1, the second matrix is ​​multiplied by a predetermined coefficient to obtain a third matrix, and the first matrix and the third matrix are concatenated along the row direction to form the block Hankel matrix.

8. A diffusion magnetic resonance imaging method for three-dimensional echo planar imaging as claimed in claim 3, characterized in that: In step A8, the average value of the amplitude matrix of the image unit corresponding to each excitation acquisition in the third image data is calculated to obtain a second mean matrix, and the magnetic resonance image is reconstructed according to the second mean matrix.

9. A diffusion magnetic resonance imaging system for three-dimensional echo planar imaging, characterized in that: A diffusion magnetic resonance imaging method for performing three-dimensional echo planar imaging as claimed in any one of claims 1 to 3, comprising: A data acquisition module, used for filling the k-space with the magnetic resonance data acquired by multiple excitations to form three-dimensional first k-space data; A first construction module, connected to the data acquisition module, for constructing a block Hankel matrix according to the first k-space data; a second construction module, connected to the first construction module, for constructing a target Hankel matrix according to the block Hankel matrix; A space recovery module, connected to the second construction module, used for recovering the target Hankel matrix to the k-space to form second k-space data; A first conversion module, connected to the space recovery module, for converting the second k-space data into first image data; an information extraction module, connected to the first conversion module, for extracting the amplitude matrix and the phase matrix of the image unit corresponding to each excitation acquisition in the first image data, and calculating the average value of the amplitude matrix to form a first mean matrix; an image updating module, connected to the information extraction module, and configured to replace the amplitude matrix of each image unit with the first mean matrix to update the amplitude of the image unit, and retain the phase matrix to form second image data; A second conversion module, connected to the image update module, for converting the second image data into k-space to form third k-space data; a stop judgment module, connected to the second conversion module, configured to judge whether the third k-space data meets the stop condition based on a preset rule, and obtain a judgment result; The first construction module is also connected to the stop judgment module, and is used to construct a block Hankel matrix according to the third k-space data when the judgment result is that the stop condition is not met; The image reconstruction module is connected to the stop judgment module and is used for converting the third k-space data into third image data when the judgment result satisfies the stop condition, and reconstructing the magnetic resonance image based on the third image data.

10. A diffusion magnetic resonance imaging system for three-dimensional echo planar imaging as claimed in claim 9, characterized in that: The magnetic resonance data includes imaging echo data and navigation echo data; During the imaging echo data acquisition process, the data acquisition module uses a linear traversal method for the intra-layer phase encoding direction and a random traversal method for the inter-layer phase encoding direction; during the navigation echo data acquisition process, the data acquisition module uses a linear traversal method for both the intra-layer phase encoding direction and the inter-layer phase encoding direction.

Images