A magnetic resonance image acquisition method, system, device, medium and program product
By adopting structured sampling sparseness strategy and multi-level sparse transformation technology in MRI image recovery, the problem of excessive sparseness of low-frequency parts is solved, the image quality and efficiency are significantly improved, and more efficient magnetic resonance image reconstruction is achieved.
Patent Information
- Application Number
- CN202411273492.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-12
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2044-09-12
AI Technical Summary
The existing compression-sensing MRI imaging methods fail to fully utilize the K-space characteristics during image recovery, resulting in limited image recovery accuracy, especially the low-frequency part being too sparse, affecting image quality and contrast.
Using a structured sampling sparseness strategy, the second k spatial sampling mask map is set in the first k spatial mask map, including low-frequency full sampling, low-frequency dense random sampling and high-frequency sparse random sampling, combined with multi-level sparse Transformation technology, optimize the image reconstruction process.
The reconstruction quality of magnetic resonance images is significantly improved, the problem of excessive sparseness of low-frequency components is avoided, and the good balance between image quality and acquisition efficiency is achieved. The image details are presented clearer, and the signal-to-noise ratio and resolution are improved.
Smart Images

Figure CN118864642B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of magnetic resonance imaging technology, and in particular to a magnetic resonance image acquisition method, system, device, medium and program product. Background Art
[0002] Magnetic resonance imaging (MRI) technology is a non-invasive medical imaging technology. The quality of MRI images is closely related to the filling density of K space. K space is the core space for magnetic resonance signal data processing. Its filling method directly determines the image resolution and acquisition time. In order to obtain high-resolution images, traditional technology needs to collect a large number of data points, which directly leads to the extension of acquisition time. Although parallel imaging technology has shortened the scanning time to a certain extent, its performance is still limited by hardware cost and complexity.
[0003] In recent years, compressed sensing (CS) technology has provided new ideas for MRI fast imaging. The compressed sensing theory shows that high-resolution images can be restored from a data volume far less than that required by the Nyquist sampling theorem. The application of this theory in the field of MRI aims to achieve a significant reduction in the amount of data acquisition by optimizing sampling strategies and data reconstruction algorithms while maintaining or improving image quality. However, the existing compressed sensing MRI imaging methods still have many shortcomings in practical applications. First, the global random sampling method fails to fully utilize the K-space characteristics, that is, the image energy is mainly concentrated in the low-frequency center area, resulting in limited image restoration accuracy. In order to restore the image more accurately, it is gradually transformed into a variable density acquisition method with dense random downsampling near the low-frequency center and sparse random downsampling in the high-frequency area. Although the variable density acquisition method improves the image restoration quality to a certain extent, there is still room for optimization.
[0004] In the process of realizing compressed sensing, sparse transformation is one of the key steps. In the sparse transformation link, wavelet transform and curvelet transform are widely used because of their fast calculation speed and suitability for non-stationary signal and image processing. However, when these transformation methods process magnetic resonance images, due to the non-ideal filtering characteristics of multi-stage low-pass filters and the fact that the energy distribution of magnetic resonance images is mainly concentrated in low-frequency components, and the K-space sampling uses a random sampling frequency lower than the Nyquist sampling rate, when the wavelet or curvelet series is too high, the low-frequency part of the finally restored magnetic resonance image will be too sparse, which will affect the overall quality and contrast of the image, especially for magnetic resonance images of anatomical image information, which will affect the image quality and image contrast. Therefore, it is urgent to provide a more efficient and accurate magnetic resonance image acquisition method based on sampling sparsity to solve the above problems. Summary of the invention
[0005] In order to solve the above technical problems, the present invention provides a magnetic resonance image acquisition method, system, device, medium and program product.
[0006] In a first aspect, the present invention provides a method for acquiring a magnetic resonance image, the method comprising the following steps:
[0007] A nuclear magnetic resonance system is provided, wherein the nuclear magnetic resonance system is a multi-coil data acquisition system including a plurality of coils;
[0008] Providing a first k-space mask image; the first k-space mask image includes a first area at the center of the first k-space mask image and a second area around the first area;
[0009] A second k-space sampling mask map is set based on the first k-space mask map; wherein the second k-space sampling mask map includes a first central area based on the location of the first area and smaller in size than the first area, and the first central area is used for low-frequency full sampling of data;
[0010] According to the second k-space sampling mask image, a magnetic resonance signal of each coil is acquired;
[0011] Perform k-space filling according to the acquired magnetic resonance signal of each coil to obtain the k-space of each coil filled with data;
[0012] Based on the k-space of each coil filled with data, a nuclear magnetic resonance block vector of each coil is obtained by transformation;
[0013] The optimal solution of the objective function is jointly solved based on the second k-space sampling mask map, the sparse transformation map of each coil and the nuclear magnetic resonance block vector of each coil; wherein the sparse transformation map of each coil is obtained based on sparse transformation of the nuclear magnetic resonance block vector of each coil in the process of solving the optimal solution of the objective function;
[0014] Obtaining a reconstructed image of a single coil based on the optimal solution;
[0015] Based on performing multi-coil image fusion on the reconstructed image of a single coil of the plurality of coils, a target reconstructed image is obtained.
[0016] In a second aspect, the present invention provides a magnetic resonance image acquisition system, the system comprising:
[0017] An acquisition system acquisition module is used to provide a nuclear magnetic resonance system, wherein the nuclear magnetic resonance system is a multi-coil data acquisition system including a plurality of coils;
[0018] A first mask image acquisition module is used to provide a first k-space mask image; the first k-space mask image includes a first area at the center of the first k-space mask image and a second area around the first area;
[0019] A second mask map acquisition module is used to set a second k-space sampling mask map based on the first k-space mask map; wherein the second k-space sampling mask map includes a first central area based on the location of the first area and smaller in size than the first area, and the first central area is used for low-frequency full sampling of data;
[0020] A magnetic resonance signal acquisition module, used for acquiring a magnetic resonance signal of each coil according to the second k-space sampling mask image;
[0021] A k-space filling module is used to perform k-space filling according to the acquired magnetic resonance signal of each coil to obtain the k-space of each coil filled with data;
[0022] A magnetic resonance image acquisition module, used for transforming and obtaining a nuclear magnetic resonance block vector of each coil based on the k-space of each coil filled with data;
[0023] An optimal solution solving module, used for jointly solving the optimal solution of the objective function based on the second k-space sampling mask map, the sparse transformation map of each coil and the nuclear magnetic resonance block vector of each coil; wherein the sparse transformation map of each coil is obtained based on sparse transformation of the nuclear magnetic resonance block vector of each coil in the process of solving the optimal solution of the objective function;
[0024] An image reconstruction module, used for obtaining a reconstructed image of a single coil based on the optimal solution;
[0025] The image fusion module is used to perform multi-coil image fusion on the reconstructed image of a single coil of the plurality of coils to obtain a target reconstructed image.
[0026] In a third aspect, the present invention further provides a computer device, comprising a processor and a memory, wherein the processor is connected to the memory, the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory so that the computer device performs the steps of implementing the above method.
[0027] In a fourth aspect, the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program implements the steps of the above method when executed by a processor.
[0028] In a fifth aspect, the present invention also provides a computer program product, comprising a program executed by at least one processor of a computing device, wherein the program comprises instructions, wherein the execution of the instructions enables the at least one processor to perform the steps of the above method. BRIEF DESCRIPTION OF THE DRAWINGS
[0029] Figure 1 is a schematic flow chart of a method for acquiring a magnetic resonance image provided by some embodiments of the present invention;
[0030] Figure 2 is a schematic diagram of the implementation process of the magnetic resonance image acquisition method provided by some embodiments of the present invention;
[0031] Figure 3 is a first k-space mask image provided by some embodiments of the present invention;
[0032] Figure 4 is a second k-space sampling mask map provided by some embodiments of the present invention;
[0033] Figure 5 is a schematic diagram of a typical three-level DWT processing framework provided by some embodiments of the present invention;
[0034] Figure 6 is a comparison diagram of magnetic resonance image reconstruction results provided by some embodiments of the present invention;
[0035] Figure 7 is a block diagram of a magnetic resonance image acquisition system provided by some embodiments of the present invention;
[0036] Figure 8 is a schematic diagram of the structure of a magnetic resonance device provided by some embodiments of the present invention;
[0037] Fig. 9 It is a schematic diagram of the structure of a computer device provided in some embodiments of the present invention. DETAILED DESCRIPTION
[0038] The following specifically illustrates the implementation mode of the present invention in conjunction with the accompanying drawings. The embodiments are provided for illustrative purposes only and cannot be understood as limiting the present invention. The accompanying drawings are provided for reference and illustration only and do not constitute a limitation on the scope of patent protection of the present invention, because many changes can be made to the present invention without departing from the spirit and scope of the present invention.
[0039] refer to Figure 1 , some embodiments of the present invention provide a method for acquiring a magnetic resonance image, such as Figure 1 As shown, the method comprises the following steps:
[0040] S1. Provide a nuclear magnetic resonance system, which is a multi-coil data acquisition system including a plurality of coils.
[0041] Existing magnetic resonance imaging (MRI) equipment usually uses parallel acquisition coils, that is, by placing multiple coils in different spatial positions, the signals of the target area of the human body can be synchronously captured and integrated, thereby realizing magnetic resonance image reconstruction. Before setting reasonable acquisition parameters and sequences, such as Figure 2 As shown, some embodiments of the present invention need to calculate the sensitivity of each coil, so as to obtain a sensitivity matrix of multiple coils of a magnetic resonance imaging (MRI) device, which is mainly achieved by the following steps:
[0042] 1) Using each coil in a multi-coil system to collect low-resolution images of the entire field of view;
[0043] 2) Calling existing algorithm toolkits, such as BART (Berkeley Advanced Reconstruction Toolbox) or SPM (Statistical Parametric Mapping) and other built-in related calculation functions (functions include filtering, thresholding and point estimation to generate coil sensitivity mapping), etc., to analyze the images collected by multiple coils, and accurately calculate the sensitivity parameter matrix of each coil. These matrices accurately describe the relative weight of each coil to different origin signals in its receiving area in a quantitative form. In this embodiment, the sensitivity matrix of all coils is marked as S = ( , , , …, ), where S is the total set of all coil sensitivity matrices, is the sensitivity matrix of the i-th coil.
[0044] S2. Provide a first k-space mask map, wherein the first k-space mask map includes a first region at the center of the first k-space mask map and a second region around the first region.
[0045] S3. Setting a second k-space sampling mask map based on the first k-space mask map; wherein the second k-space sampling mask map includes a first central region based on the location of the first region and smaller in size than the first region, and the first central region is used for low-frequency full sampling of data.
[0046] S4. According to the second k-space sampling mask image, the magnetic resonance signal of each coil is acquired.
[0047] In some embodiments of the present invention, the first k-space mask diagram intuitively displays the amount and position of data points to be collected in the k-space, thereby achieving precise control of the data collection amount and position. Specifically, the values 0 and 1 displayed on the first k-space mask diagram determine whether the signal at the current phase encoding and frequency encoding position needs to be collected, that is, 0 and 1 on the first k-space mask diagram represent instructions for not collecting and collecting, respectively, guiding the magnetic resonance instrument to capture signals at different positions of phase encoding and frequency encoding.
[0048] like Figure 3 As shown, the conventional k-space sampling mask image presents the principle of dense acquisition in the center of the k-space and random sparse acquisition in the peripheral area, that is, more intensive low-frequency full sampling is performed in the central area of the k-space to ensure the integrity and clarity of the core information of the image; while a random sparse sampling mode is adopted in the peripheral area to reduce the overall data volume requirement while retaining the high-frequency details and edge information of the image as much as possible. However, after the non-ideal low-pass filtering in the process of compressed sensing downsampling and multi-level discrete wavelet transform in the traditional magnetic resonance image reconstruction method, the low-frequency part of the original image is too sparse, thereby affecting the low-frequency amplitude component of the image, thereby reducing the image quality and having an adverse effect on the subsequent image reconstruction accuracy. To solve this problem, some embodiments of the present invention modify the first k-space mask image and the nonlinear objective function to retain as many low-frequency amplitude components of the image as possible, avoid the problem of too sparse low-frequency components of the image after wavelet transform, thereby improving the overall reconstruction quality of the image.
[0049] like Figure 4 As shown, some embodiments of the present invention set a second k-space sampling mask map based on the first k-space mask map, that is, the second k-space sampling mask map includes a first central region based on the position of the first region and smaller in size than the first region, a first peripheral region based on the position of the first region and around the first central region, and a second peripheral region based on the position of the second region; wherein the first central region is used for low-frequency full sampling of data; the first peripheral region is used for low-frequency dense sampling of data; and the second peripheral region is used for high-frequency sparse sampling of data.
[0050] In some embodiments of the present invention, the step of acquiring the magnetic resonance signal of each coil according to the second k-space sampling mask image includes: performing low-frequency full sampling on the first central area with a sampling rate that satisfies the Nyquist sampling theorem, performing low-frequency dense random sampling on the first peripheral area with a sampling rate lower than the Nyquist sampling theorem, and performing high-frequency sparse random sampling on the second peripheral area with a sampling rate lower than the Nyquist sampling theorem.
[0051] Specifically, the sampling rate of the first central area satisfies the Nyquist sampling theorem, ensuring the integrity of the low-frequency information, while the remaining high-frequency parts (the second peripheral area) except the first central area adopt a sparse random sampling method, and the second peripheral area adopts a random or pseudo-random sampling method with a sampling rate less than Nyquist. This variable density and variable sampling frequency method is called structured sampling sparsity. Structured sparsity not only meets the Nyquist sampling rate requirement through full sampling of the central low-frequency area, but also ensures the complete retention of the low-frequency components of the image, which is crucial to image quality. In the subsequent wavelet, curvelet and other multi-level sparse transform processing, this full sampling method of the central low frequency can minimize the impact on the low-frequency area, thereby optimizing the image reconstruction process. Compared with the traditional compressed sensing method, it can significantly improve the quality of the final image.
[0052] In some embodiments of the present invention, the size of the first central area is determined according to the number of sparse transform decomposition layers of the original image signal and the sampling space range of the second k-space sampling mask map, wherein the preset size of the first central area is greater than or equal to the size of the sparse transform map with the lowest frequency after the preset multi-level sparse transform, and the size of the first central area is 10% or less of the size of the second k-space sampling mask map. For example, when the sparse transform adopts wavelet transform, it should be ensured that the size of the first central area is at least equal to the size of the wavelet transform map with the lowest frequency after the multi-level sparse transform. In some embodiments, it does not exceed 10% of the horizontal and vertical axis size of the second k-space sampling mask map to balance the requirements of data volume and image quality. It should be noted that those skilled in the art can adjust this preset percentage according to the specific implementation situation, which is not limited to some embodiments of the present invention.
[0053] S5. Perform k-space filling according to the acquired magnetic resonance signals of each coil to obtain the k-space of each coil filled with data.
[0054] In some embodiments of the present invention, multiple coils and magnetic resonance devices perform parallel magnetic resonance signal acquisition tasks according to the second k-space sampling mask map to obtain magnetic resonance signals of each coil, and then perform k-space filling according to the acquired magnetic resonance signals of each coil. Specifically, the sampling area or sampling points in the second k-space sampling mask map are filled based on a variable density sampling trajectory strategy under Cartesian coordinates. This filling process is divided into the following two modes according to the order of data acquisition:
[0055] Frequency encoding direction priority filling mode: for the frequency encoding direction, all data points on the axis are collected, and variable density random under-sampling is performed on the phase encoding direction and the layer encoding direction. The positions of the under-sampling points are determined by the second k-space sampling mask image in step 102.
[0056] Phase encoding direction priority filling mode: for the phase encoding direction, all data points on the axis are collected, and variable density random under-sampling is performed on the frequency encoding direction and the layer encoding direction. The positions of the under-sampling points are determined by the second k-space sampling mask image in step 102.
[0057] The application of the above two filling modes not only improves the flexibility of data acquisition, but also effectively reduces the amount of data collection while ensuring that key information is not lost, providing strong support for subsequent image reconstruction and optimization.
[0058] S6. Based on the k-space of each coil filled with data, transform and obtain the nuclear magnetic resonance block vector of each coil.
[0059] S7. The optimal solution of the objective function is obtained based on the joint solution of the second k-space sampling mask map, the sparse transformation map of each coil and the nuclear magnetic resonance block vector of each coil; wherein the sparse transformation map of each coil is obtained based on the sparse transformation of the nuclear magnetic resonance block vector of each coil in the process of solving the optimal solution of the objective function.
[0060] Some embodiments of the present invention may use discrete wavelet transform or curvelet transform to perform sparse transform on the magnetic resonance block vector of each coil to obtain a sparse transform map of each coil to facilitate subsequent compressed sensing and image reconstruction.
[0061] In some embodiments of the present invention, this embodiment performs a one-dimensional Fourier transform on the data collected by the coil and fills it into the k-space, then performs an inverse Fourier transform on the K-space of each coil filled with the data, converts the data in the k-space (i.e., frequency domain data) back to the spatial domain, obtains the nuclear magnetic resonance block vector of each coil, and then performs a sparse transform on the nuclear magnetic resonance block vector of each coil in the spatial domain. Discrete wavelet transform is a multi-scale, multi-directional transformation method that can decompose an image into sub-bands of different frequencies and directions. These sub-bands can be used to remove noise, enhance edges, etc. during image reconstruction.
[0062] Among them, some implementations of the present invention can set the filter level of wavelet transform to 3-6, and set the filter level of curvelet transform to 3-6. The implementation process of wavelet transform and curvelet transform will be described in detail below:
[0063] Wavelet transform refers to the discrete wavelet transform (DWT) of a two-dimensional image. The process of multi-level decomposition of the DWT image can be realized by high-pass and low-pass filters in two directions (horizontal and vertical). Among them, the high-pass filter is used to extract edge features, and the low-pass filter is used for image approximation to retain most of the effective information of the image. A typical three-level DWT processing framework is as follows Figure 5 As shown, in Figure 5 In the figure, x[n] is the input image, g[n] and h[n] represent low-pass filter and high-pass filter respectively, which include horizontal filtering and vertical filtering; 2 represents 2-fold downsampling, and 2-fold downsampling is performed after each filtering step to reduce the image size; Level coefficient is the wavelet transform factor of the corresponding component (high frequency, low frequency) after wavelet transform. After wavelet transform, the sparsity of the image can be determined according to the size of the transform factor, and then the image is solved and reconstructed by using the wavelet inverse transform matrix and compressed sensing algorithm. As the number of wavelet transform levels increases, the sparse representation of the image by the wavelet transform becomes better, but at the same time, the amount of calculation for transformation and reconstruction will increase significantly. Therefore, those skilled in the art can adjust the number of wavelet transform levels according to the specific implementation situation, which is not limited to some embodiments of the present invention.
[0064] Curvelet transform is a multi-scale analysis method, which is an extension of wavelet transform. Compared with the high- and low-pass filters used alternately in the traditional wavelet transform process, the curvelet transform transforms the image to the frequency domain through a bandpass filter, and then performs a wavelet multi-level transform. At the same time, curve fitting is performed at each transformation scale (i.e., the corresponding different wavelet transform factors) to better capture the local features of the image, and then threshold noise processing is performed. After the important local features and details are completely preserved, the noise is filtered out with a threshold, and finally the image can be reconstructed using the curvelet inverse transform matrix and the compressed sensing algorithm. Similarly, the curvelet transform is implemented based on the wavelet transform, so the problems of wavelet transform will also arise. Therefore, technical personnel in this field can adjust the curvelet transform series according to the specific implementation situation, which is not limited to some embodiments of the present invention.
[0065] In some embodiments of the present invention, the step of jointly solving the optimal solution of the objective function based on the second k-space sampling mask image, the sparse transformation image of each coil and the nuclear magnetic resonance block vector of each coil includes: jointly solving the optimal solution of the objective function based on the second k-space sampling mask image, the sparse transformation image of each coil, the nuclear magnetic resonance block vector of each coil, and the diagonal block matrix of each coil.
[0066] In some embodiments of the present invention, the step of jointly solving the optimal solution of the objective function based on the second k-space sampling mask map, the sparse transformation map of each coil and the nuclear magnetic resonance block vector of each coil includes: jointly solving the optimal solution of the objective function based on the second k-space sampling mask map, the sparse transformation map of each coil, the nuclear magnetic resonance block vector of each coil and the low-frequency estimation item of each coil, wherein the low-frequency estimation item of each coil corresponds to the low-frequency full sampling item in the nuclear magnetic resonance block vector of each coil.
[0067] In some embodiments of the present invention, the low-frequency estimation item of each coil is obtained based on the inverse matrix of the diagonal block matrix of each coil, the nuclear magnetic resonance block vector of each coil, the window function and the first central area mask matrix of the second k-space sampling mask image; in some embodiments, the window function is selected from one of the Kaiser window function, the Hamming window function or the Henning window function.
[0068] In some embodiments of the present invention, the step of jointly solving the optimal solution of the objective function based on the second k-space sampling mask image, the sparse transformation image of each coil and the nuclear magnetic resonance block vector of each coil includes: jointly solving the optimal solution of the objective function based on the second k-space sampling mask image, the sparse transformation image of each coil, the nuclear magnetic resonance block vector of each coil, the diagonal block matrix of each coil and the sensitivity matrix of multiple coils in the nuclear magnetic resonance system.
[0069] In some embodiments of the present invention, the step of jointly solving the optimal solution of the objective function based on the second k-space sampling mask image, the sparse transformation image of each coil and the nuclear magnetic resonance block vector of each coil includes: jointly solving the optimal solution of the objective function based on the second k-space sampling mask image, the sparse transformation image of each coil, the nuclear magnetic resonance block vector of each coil, the diagonal block matrix of each coil, the low-frequency estimation item of each coil and the sensitivity matrix of multiple coils in the nuclear magnetic resonance system.
[0070] For ease of understanding, some embodiments of the present invention ensure complete retention of low-frequency information by superimposing the first central region, so that a low-frequency full sampling item is superimposed in the actual sampling data vector. Therefore, a low-frequency estimation item of each coil is introduced into the magnetic resonance image reconstruction model, which corresponds to the low-frequency full sampling item superimposed in the actual sampling data vector due to full sampling, so as to ensure accurate representation of low-frequency information. Therefore, the objective function needs to be:
[0071]
[0072] Where minimize is the minimum value of the objective function; z is the reconstructed image to be solved; M is the mask matrix corresponding to the second k-space sampling mask image; F is the diagonal block Fourier transform matrix; S is the sensitivity matrix of each coil; is the conjugate function of the sparse transformation matrix of each coil; is the low-frequency estimate for each coil; b is the block vector of k-space for each coil filled with data, where , is the k-space block of each coil filled with data obtained by sampling the i-th coil; is the norm operator. When j=2, it is Norm, when j=1, is norm; is the regularization coefficient, and its empirical value range is 0.001-0.1. It needs to be set in advance before solving the optimal solution of the objective function.
[0073] Some embodiments of the present invention may use a fast iterative shrinkage threshold algorithm (FISTA) or a soft threshold iterative shrinkage algorithm to solve the sparse transformation map and the nuclear magnetic resonance block vector of each coil for the optimal solution of the objective function. The optimal solution of the objective function is to obtain the minimum value of the objective function by performing a sparse change solution on the nuclear magnetic resonance block vector of each coil, that is, to obtain the minimum value of the objective function by performing a sparse change solution z on the nuclear magnetic resonance block vector of each coil.
[0074] It is understandable that those skilled in the art may use other relevant algorithms to solve the reconstructed image z to be solved.
[0075] In addition, considering the need to avoid low-frequency estimation items in the sparse transformation process Too sparse, thereby losing key image reconstruction information, therefore, in some embodiments, a second k-space sampling mask is set based on the first k-space mask; wherein the second k-space sampling mask includes a first central area based on the location of the first area and smaller in size than the first area, and the first central area is used for low-frequency full sampling of data, then, the low-frequency estimation item at this time It can be characterized as:
[0076]
[0077] In the formula, is the inverse matrix of the diagonal block Fourier transform matrix; is the window function kernel matrix; is the mask matrix corresponding to the second k-space sampling mask image.
[0078] In order to reduce the error generated when performing Fourier transform on non-integer periods, some embodiments of the present invention preferably use the Kaiser window function to perform window calculation on the Fourier transform, so as to effectively avoid the spectrum leakage problem, that is, to prevent the energy of a certain frequency component from being improperly dispersed to other frequencies during the Fourier transform process, thereby avoiding errors such as frequency spectrum energy fluctuations and frequency aliasing caused by this. It should be noted that when the window function uses the Kaiser window function, the Kaiser window kernel matrix here is is a constant value to ensure the stability and consistency of the windowing process. The matrix element k of the Kaiser window kernel matrix is:
[0079]
[0080] In the formula, is an element in the window function kernel matrix; n is the element position index of the window function kernel matrix, and its value range is [0, N]; is the first kind of zero-order modified Bessel function; is the spectrum sidelobe attenuation coefficient; is the window length of the window function kernel matrix; it should be noted that the infinite extension characteristic of the traditional inverse Fourier transform in the time domain may cause ambiguous frequency leakage or aliasing. The characteristic of the window function is that it is non-zero in a limited time interval and zero in the rest of the time. The function of the window function is to limit the signal in the time domain to a limited time period, so that the Fourier transform is only performed within the time period without being affected by the infinite extension of the signal in the time domain. In addition to the Kaiser window, some embodiments of the present invention can also use related window functions such as the Hann Window and the Hamming Window as the window function in the low-frequency estimation item, wherein the Hann window function is specifically:
[0081]
[0082] The specific Hamming window function is:
[0083]
[0084] Where N is the window length; n is the position index of the element in the window, and its value range is [0, N].
[0085] Some embodiments of the present invention allow the intermediate variable , and Substitute the following formula:
[0086]
[0087] Then we have:
[0088]
[0089] It should be noted that the diagonal block matrix of each coil is obtained by applying the fast Fourier transform (FFT) to the MRI image X processed by each coil to obtain its corresponding diagonal block matrix , and these diagonal block matrices Placed on the diagonal line, for ease of understanding, the specific acquisition process of the diagonal block matrix of each coil is briefly described below. Some embodiments of the present invention define a Fourier transform matrix F, which is a diagonal block matrix. The Fourier transform matrix F is composed as follows:
[0090]
[0091] The Fourier transform matrix F has non-zero values only on the diagonal, and the values on the diagonal are marked as , is a diagonal block matrix, i ranges from [0, I], and each diagonal block matrix Corresponding to a coil of nuclear magnetic resonance image X, the diagonal block matrix The size corresponds to the size of the processed MRI image X of each coil. Assuming that the size of the processed MRI image X of a coil is m*n, the fast Fourier transform (FFT) algorithm is used to calculate the diagonal block matrix according to the following formula: The effect of X on the MRI image is , for each coil's MRI image X, the corresponding , all calculated Combined together, we get the complete discrete Fourier transform result:
[0092]
[0093]
[0094] In the formula, and is the complex root of unity; j is the imaginary unit; p and j are indices with values ranging from 0 to m-1; q and k are indices with values ranging from 0 to n-1; Y is the discrete Fourier transform corresponding to the input X.
[0095] Based on the above process, the diagonal block matrix of each coil can be obtained and the fast Fourier transform can be completed.
[0096] S8. Obtain a reconstructed image of a single coil based on the optimal solution.
[0097] S9. Perform multi-coil image fusion on the reconstructed image of a single coil of the plurality of coils to obtain a target reconstructed image.
[0098] Considering that the low-frequency estimation items superimposed by each coil are different, compared with the traditional image reconstruction method, some embodiments of the present invention obtain the optimal solution by solving the above objective function, and substitute the optimal solution into ( ) can be used to obtain the reconstructed image of the i-th coil , and then the final reconstructed target image can be obtained through multi-coil fusion. Figure 6 Arranged from left to right are a fully sampled magnetic resonance image, a magnetic resonance image reconstructed using a traditional compressed sensing method, and a target reconstructed image after low-frequency compensation using the method provided in some embodiments of the present invention. By comparing the image reconstruction effects in some embodiments of the present invention, it can be clearly seen that the method proposed in some embodiments of the present invention exhibits significant advantages in magnetic resonance image reconstruction. This method not only enables clearer detail presentation and is closer to the high fidelity of a fully sampled magnetic resonance image, but also retains the efficiency advantage of compressed sensing technology in quickly acquiring data. More importantly, this method effectively overcomes the shortcomings of traditional compressed sensing methods that cause excessive sparseness in low-frequency components of images and affect image quality, and achieves a good balance between image quality and acquisition efficiency.
[0099] The magnetic resonance image acquisition method proposed in some embodiments of the present invention not only makes up for the problem of excessive sparseness of low-frequency components of images caused by traditional compressed sensing technology when applying sparse transformations such as wavelet and curvelet transforms, but also significantly improves the reconstruction quality of magnetic resonance images while maintaining fast reconstruction efficiency. Some embodiments of the present invention set a second k-space sampling mask map based on the first k-space variable density random downsampling strategy of traditional compressed sensing technology, wherein the second k-space sampling mask map includes a first central area based on the position of the first area and smaller in size than the first area, the first central area is used for low-frequency full sampling of data, and at the same time ensures that the first central area is sampled at a low frequency rate that satisfies the Nyquist sampling theorem, and the first peripheral area is sampled at a rate lower than The low-frequency dense random sampling of Nyquist sampling theorem is performed, and the high-frequency sparse random sampling of the second peripheral area with a sampling rate lower than that of Nyquist sampling theorem is performed, thereby forming a structured sparse sampling pattern, which effectively reduces the negative impact of multi-level wavelet and curvelet transform on the low-frequency components of k-space. Compared with conventional compressed sensing reconstruction technology, the magnetic resonance image acquisition method proposed in some embodiments of the present invention sacrifices a small amount of sampling complexity and sampling time cost in exchange for a significant improvement in the reconstruction quality of the magnetic resonance image. Specifically, some embodiments of the present invention greatly shorten the data acquisition time and significantly improve the detection efficiency through downsampling technology. After superimposing the central low-frequency full sampling area and optimizing the reconstruction process, the image quality is significantly improved. It has broad application potential and significant imaging advantages in the field of MRI.
[0100] Some embodiments of the present invention provide a method for acquiring a magnetic resonance image, the method providing a nuclear magnetic resonance system, the nuclear magnetic resonance system being a multi-coil data acquisition system including a plurality of coils; providing a first k-space mask image, the first k-space mask image including a first area at the center of the first k-space mask image and a second area around the first area; setting a second k-space sampling mask image based on the first k-space mask image; wherein the second k-space sampling mask image includes a first central area based on the position of the first area and having a size smaller than the first area, the first central area being used for low-frequency full sampling of data; and The second k-space sampling mask map is used to collect the magnetic resonance signal of each coil; the k-space is filled according to the magnetic resonance signal of each coil collected to obtain the k-space of each coil filled with data; based on the k-space of each coil filled with data, the nuclear magnetic resonance block vector of each coil is transformed; based on the joint solution of the objective function of the second k-space sampling mask map, the sparse transformation map of each coil and the nuclear magnetic resonance block vector of each coil, the optimal solution is obtained; based on the optimal solution, a reconstructed image of a single coil is obtained; based on the reconstructed image of a single coil of the obtained plurality of coils, a multi-coil image fusion is performed to obtain a target reconstructed image. By adopting the magnetic resonance image acquisition method described in some embodiments of the present invention to reconstruct magnetic resonance images, not only the details are presented more clearly, but also the signal-to-noise ratio and resolution are significantly improved, and the quality of the reconstructed image is closer to the full-sampling magnetic resonance image. In addition, while inheriting the advantages of compressed sensing technology in quickly collecting data, the method effectively overcomes the defect that the traditional method is prone to excessive sparseness of the low-frequency components of the image, thereby comprehensively optimizing the quality and practicality of image reconstruction.
[0101] It should be noted that the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.
[0102] In some embodiments of the present invention, Figure 7 As shown, some embodiments of the present invention provide a magnetic resonance image acquisition system, the system comprising:
[0103] The acquisition system acquisition module 101 is used to provide a nuclear magnetic resonance system, wherein the nuclear magnetic resonance system is a multi-coil data acquisition system including a plurality of coils;
[0104] A first mask image acquisition module 102 is used to provide a first k-space mask image, wherein the first k-space mask image includes a first area at the center of the first k-space mask image and a second area around the first area;
[0105] A second mask map acquisition module 103 is used to set a second k-space sampling mask map based on the first k-space mask map; wherein the second k-space sampling mask map includes a first central area based on the location of the first area and smaller in size than the first area, and the first central area is used for low-frequency full sampling of data;
[0106] A magnetic resonance signal acquisition module 104, configured to acquire a magnetic resonance signal of each coil according to the second k-space sampling mask image;
[0107] A k-space filling module 105 is used to perform k-space filling according to the acquired magnetic resonance signals of each coil, so as to obtain the k-space of each coil filled with data;
[0108] A magnetic resonance image acquisition module 106, configured to transform and obtain a nuclear magnetic resonance block vector of each coil based on the k-space of each coil filled with data;
[0109] The optimal solution solving module 107 is used to jointly solve the optimal solution of the objective function based on the second k-space sampling mask map, the sparse transformation map of each coil and the nuclear magnetic resonance block vector of each coil; wherein the sparse transformation map of each coil is obtained based on the sparse transformation of the nuclear magnetic resonance block vector of each coil in the process of solving the optimal solution of the objective function;
[0110] An image reconstruction module 108, configured to obtain a reconstructed image of a single coil based on the optimal solution;
[0111] The image fusion module 109 is used to perform multi-coil image fusion on the reconstructed image of a single coil of the plurality of coils to obtain a target reconstructed image.
[0112] In some embodiments of the present invention, a magnetic resonance image acquisition system can be applied to Figure 8 In the exemplary magnetic resonance electronic device shown, Figure 8In the exemplary magnetic resonance electronic equipment, the instrument integrates a plurality of key components such as an examination bed, a superconducting switch, and a gradient controller to perform the above-mentioned compressed sensing magnetic resonance image reconstruction process. Specifically, the examination bed 1 provides a stable and safe platform for examination; the superconducting switch 2 is used to activate the superconducting coil 5 after the magnetic resonance instrument is started and debugged, so that the superconducting coil 5 enters the working state and is ready for the subsequent magnetic resonance scanning; the gradient controller 3, the radio frequency pulse generator 7, and the radio frequency coil receiver 9 are all precisely controlled by the sequence generator 8, and the sequence can be sent by the host computer 13 The scanning instruction is sent to the sequence generator 8, and the sequence generator 8 automatically sends operation instructions to the gradient controller 3, the RF pulse generator 7, the RF coil receiver 9 and other hierarchical components according to the selected sequence to ensure that they perform scanning, transmission and reception tasks according to the predetermined timing and pulse width. The gradient controller 3 is responsible for driving the gradient coil 4 to generate a three-dimensional (X, Y, Z) gradient field; the RF pulse generator 7 controls the RF coil 6 to emit RF signals of specific energy and frequency to stimulate the magnetic resonance phenomenon of hydrogen atoms in the human body; the RF coil receiver 9 synchronously captures these magnetic resonance signals and transmits them to the data buffer area.
[0113] After the data acquisition device 10 and the K-space filler 11 complete a signal acquisition, the data acquisition device 10 performs Fourier transform on the signal and accurately fills it into the K-space according to the frequency and phase encoding information. When all acquisition tasks are completed, the K-space filler passes the complete K-space data to the image reconstruction module 12. The image reconstruction module 12 has a variety of built-in magnetic resonance image reconstruction algorithms, which can convert the K-space data into high-quality magnetic resonance images. Finally, these images are displayed on the graphical interface of the main computer 13 through the image display 14.
[0114] For the specific definition of a magnetic resonance image acquisition system, please refer to the above-mentioned definition of a magnetic resonance image acquisition method, which will not be repeated here. A person of ordinary skill in the art can appreciate that the various modules and steps described in conjunction with the embodiments disclosed in this application can be implemented in hardware, software, or a combination of both. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.
[0115] Some embodiments of the present invention provide a magnetic resonance image acquisition system, wherein the acquisition system acquisition module of the system is used to provide a nuclear magnetic resonance system, wherein the nuclear magnetic resonance system is a multi-coil data acquisition system including a plurality of coils; the first mask image acquisition module is used to provide a first k-space mask image, wherein the first k-space mask image includes a first area at the center of the first k-space mask image and a second area around the first area; the second mask image acquisition module is used to set a second k-space sampling mask image based on the first k-space mask image; wherein the second k-space sampling mask image includes a first central area based on the position of the first area and smaller in size than the first area, and the first central area is used for low-frequency full sampling of data; the magnetic resonance signal acquisition module is used to obtain a second k-space sampling mask image based on the first k-space mask image; The second k-space sampling mask map is used to collect the magnetic resonance signal of each coil; the k-space filling module is used to fill the k-space according to the magnetic resonance signal of each coil collected to obtain the k-space of each coil filled with data; the magnetic resonance image acquisition module is used to transform the nuclear magnetic resonance block vector of each coil based on the k-space of each coil filled with data; the optimal solution solving module is used to jointly solve the optimal solution of the objective function based on the second k-space sampling mask map, the sparse transformation map of each coil and the nuclear magnetic resonance block vector of each coil; the image reconstruction module obtains the reconstructed image of a single coil based on the optimal solution; the image fusion module performs multi-coil image fusion on the reconstructed image of a single coil of the obtained several coils to obtain the target reconstructed image. The magnetic resonance image acquisition system provided by some embodiments of the present invention achieves a significant improvement in both signal-to-noise ratio and resolution, and the quality of the reconstructed image is closer to that of the fully sampled magnetic resonance image. In addition, while inheriting the advantage of compressed sensing technology in quickly collecting data, it effectively overcomes the defect that the traditional technology is prone to excessive sparseness of the low-frequency components of the image, and can reconstruct the magnetic resonance image quickly and with high quality.
[0116] Fig. 9 A computer device is provided in some embodiments of the present invention, comprising a memory, a processor and a transceiver, which are connected via a bus; the memory is used to store a set of computer program instructions and data, and can transmit the stored data to the processor, and the processor can execute the program instructions stored in the memory to perform the steps of the above method.
[0117] The memory may include a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories; the processor may be a central processing unit, a microprocessor, an application-specific integrated circuit, a programmable logic device, or a combination thereof. By way of example but not limitation, the programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic, or any combination thereof.
[0118] Additionally, the memory may be a physically separate unit or may be integrated with the processor.
[0119] It can be understood by those skilled in the art that Fig. 9 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have the same component arrangement.
[0120] In some embodiments of the present invention, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above method are implemented.
[0121] In some embodiments of the present invention, some embodiments of the present invention provide a computer program product, including a program executed by at least one processor of a computing device, the program including instructions, wherein the execution of the instructions causes the at least one processor to perform the steps of the above method.
[0122] Some embodiments of the present invention provide a magnetic resonance image acquisition method, system, device, medium and program product, wherein a second k-space sampling mask diagram proposed in a magnetic resonance image acquisition method is combined with an objective function, which can effectively improve imaging speed and quality, reduce noise, and improve image clarity and resolution, and is suitable for the needs of fast magnetic resonance imaging.
[0123] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in some embodiments of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., an SSD), etc.
[0124] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing related hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods.
[0125] The above-mentioned embodiments only express several preferred implementation modes of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in the technical field, several improvements and substitutions can be made without departing from the technical principles of the present invention, and these improvements and substitutions should also be regarded as the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be based on the protection scope of the claims.
Claims
1. A method for acquiring a magnetic resonance image, characterized in that: The following steps are involved: A nuclear magnetic resonance system is provided, wherein the nuclear magnetic resonance system is a multi-coil data acquisition system including a plurality of coils; Providing a first k-space mask image; the first k-space mask image includes a first area at the center of the first k-space mask image and a second area around the first area; A second k-space sampling mask map is set based on the first k-space mask map; wherein the second k-space sampling mask map includes a first central area based on the location of the first area and smaller in size than the first area, and the first central area is used for low-frequency full sampling of data; According to the second k-space sampling mask image, a magnetic resonance signal of each coil is acquired; Perform k-space filling according to the acquired magnetic resonance signal of each coil to obtain the k-space of each coil filled with data; Based on the k-space of each coil filled with data, a nuclear magnetic resonance block vector of each coil is obtained by transformation; The optimal solution of the objective function is jointly solved based on the second k-space sampling mask map, the sparse transformation map of each coil and the nuclear magnetic resonance block vector of each coil; wherein the sparse transformation map of each coil is obtained based on the sparse transformation of the nuclear magnetic resonance block vector of each coil in the process of solving the optimal solution of the objective function; Obtaining a reconstructed image of a single coil based on the optimal solution; Based on performing multi-coil image fusion on the reconstructed image of a single coil of the plurality of coils, a target reconstructed image is obtained.
2. The magnetic resonance image acquisition method according to claim 1, characterized in that: The sparse transformation method is wavelet transformation or curvelet transformation.
3. The magnetic resonance image acquisition method according to claim 2, characterized in that: The filter order of the wavelet transform is 3-6, and the filter order of the curvelet transform is 3-6.
4. The magnetic resonance image acquisition method according to claim 1, characterized in that: The second k-space sampling mask image also includes a first peripheral area based on the position of the first area around the first central area, and a second peripheral area based on the position of the second area; wherein the first peripheral area is used for low-frequency dense sampling of data, and the second peripheral area is used for high-frequency sparse sampling of data.
5. The magnetic resonance image acquisition method according to claim 4, characterized in that: The step of acquiring the magnetic resonance signal of each coil according to the second k-space sampling mask image comprises: The first central area is subjected to low-frequency full sampling with a sampling rate that satisfies the Nyquist sampling theorem, the first peripheral area is subjected to low-frequency dense random sampling with a sampling rate lower than the Nyquist sampling theorem, and the second peripheral area is subjected to high-frequency sparse random sampling with a sampling rate lower than the Nyquist sampling theorem.
6. The magnetic resonance image acquisition method according to claim 3, characterized in that: The size of the first central area is preset to be greater than or equal to the size of the sparse transformation map with the lowest frequency after the preset multi-level sparse transformation.
7. The magnetic resonance image acquisition method according to claim 1, characterized in that: The size of the first central area is preset to be 10% or less of the size of the second k-space sampling mask image.
8. The magnetic resonance image acquisition method according to claim 1, characterized in that: The step of jointly solving the optimal solution of the objective function based on the second k-space sampling mask map, the sparse transformation map of each coil and the nuclear magnetic resonance block vector of each coil comprises: The optimal solution of the objective function is obtained based on the joint solution of the second k-space sampling mask image, the sparse transformation image of each coil, the nuclear magnetic resonance block vector of each coil and the low-frequency estimation item of each coil, wherein the low-frequency estimation item of each coil corresponds to the low-frequency full sampling item in the nuclear magnetic resonance block vector of each coil.
9. The magnetic resonance image acquisition method according to claim 8, characterized in that: The low frequency estimation item of each coil is obtained based on the inverse matrix of the diagonal blocking matrix of each coil, the nuclear magnetic resonance blocking vector of each coil, a window function and a first central area mask matrix of the second k-space sampling mask image.
10. The magnetic resonance image acquisition method according to claim 9, characterized in that: The window function is selected from one of a Kaiser window function, a Hamming window function or a Hening window function.
11. The magnetic resonance image acquisition method according to claim 1, characterized in that: The step of jointly solving the optimal solution of the objective function based on the second k-space sampling mask image, the sparse transformation image of each coil and the nuclear magnetic resonance block vector of each coil includes: jointly solving the optimal solution of the objective function based on the second k-space sampling mask image, the sparse transformation image of each coil, the nuclear magnetic resonance block vector of each coil and the diagonal block matrix of each coil.
12. The magnetic resonance image acquisition method according to claim 1, characterized in that: The step of jointly solving the optimal solution of the objective function based on the second k-space sampling mask image, the sparse transformation image of each coil and the nuclear magnetic resonance block vector of each coil includes: jointly solving the optimal solution of the objective function based on the second k-space sampling mask image, the sparse transformation image of each coil, the nuclear magnetic resonance block vector of each coil, the diagonal block matrix of each coil and the sensitivity matrix of multiple coils in the nuclear magnetic resonance system.
13. The magnetic resonance image acquisition method according to claim 1, characterized in that: The step of jointly solving the optimal solution of the objective function based on the second k-space sampling mask image, the sparse transformation image of each coil and the nuclear magnetic resonance block vector of each coil includes: jointly solving the optimal solution of the objective function based on the second k-space sampling mask image, the sparse transformation image of each coil, the nuclear magnetic resonance block vector of each coil, the diagonal block matrix of each coil, the low-frequency estimation item of each coil and the sensitivity matrix of multiple coils in the nuclear magnetic resonance system.
14. The magnetic resonance image acquisition method according to claim 1, characterized in that: The step of jointly solving the optimal solution of the objective function based on the second k-space sampling mask image, the sparse transformation image of each coil and the nuclear magnetic resonance block vector of each coil includes: solving the optimal solution of the objective function for the sparse transformation image and the nuclear magnetic resonance block vector of each coil based on a fast iterative threshold shrinkage algorithm.
15. The magnetic resonance image acquisition method according to claim 1, characterized in that: The optimal solution of the objective function is to solve the minimum value of the objective function by performing sparse changes based on the nuclear magnetic resonance block vector of each coil.
16. A magnetic resonance image acquisition system, characterized in that: The system comprises: An acquisition system acquisition module is used to provide a nuclear magnetic resonance system, wherein the nuclear magnetic resonance system is a multi-coil data acquisition system including a plurality of coils; A first mask image acquisition module is used to provide a first k-space mask image; the first k-space mask image includes a first area at the center of the first k-space mask image and a second area around the first area; A second mask map acquisition module is used to set a second k-space sampling mask map based on the first k-space mask map; wherein the second k-space sampling mask map includes a first central area based on the location of the first area and smaller in size than the first area, and the first central area is used for low-frequency full sampling of data; A magnetic resonance signal acquisition module, used for acquiring a magnetic resonance signal of each coil according to the second k-space sampling mask image; A k-space filling module is used to perform k-space filling according to the acquired magnetic resonance signal of each coil to obtain the k-space of each coil filled with data; A magnetic resonance image acquisition module, used for transforming and obtaining a nuclear magnetic resonance block vector of each coil based on the k-space of each coil filled with data; An optimal solution solving module, used for jointly solving the optimal solution of the objective function based on the second k-space sampling mask map, the sparse transformation map of each coil and the nuclear magnetic resonance block vector of each coil; wherein the sparse transformation map of each coil is obtained based on sparse transformation of the nuclear magnetic resonance block vector of each coil in the process of solving the optimal solution of the objective function; An image reconstruction module, used for obtaining a reconstructed image of a single coil based on the optimal solution; The image fusion module is used to perform multi-coil image fusion on the reconstructed image of a single coil of the plurality of coils to obtain a target reconstructed image.
17. A computer device, characterized in that: The computer device comprises a processor and a memory, wherein the processor is connected to the memory, the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the computer device executes the method according to any one of claims 1 to 15.
18. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed, the method according to any one of claims 1 to 15 is implemented.
19. A computer program product, characterized in that: The method comprises a program executed by at least one processor of a computing device, the program comprising instructions, wherein execution of the instructions causes the at least one processor to perform the method according to any one of claims 1 to 15.
Citation Information
Patent Citations
Image processing method based on sparse sampling magnetic resonance imaging
CN103064046A
Optical spectrum based ultrasonic displacement difference imaging method
CN108714021A
Magnetic resonance imaging method and device and computer equipment
CN111812571A
Magnetic resonance imaging apparatus and generating method for magnetic resonance image thereof
KR1020150125543A
Magnetic resonance imaging device and method for generating magnetic resonance image
WO2015167307A1
