Matrix-elastic-net convolution lower bound regularized accelerated method for magnetic resonance parameter imaging
By employing the matrix elastic mesh convolutional underfimum regularization method, and utilizing the gradient domain model and alternating direction multiplier method to optimize the solution of magnetic resonance parametric imaging, the problems of redundant periods and artifacts in the magnetic resonance parametric imaging process are solved, achieving higher imaging acceleration and accuracy.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHENZHEN INST OF ADVANCED TECH CHINESE ACAD OF SCI
- Filing Date
- 2021-12-16
- Publication Date
- 2026-05-29
AI Technical Summary
In the process of magnetic resonance parametric imaging, the lengthy imaging cycle is prone to introducing motion artifacts, making it difficult to meet the spatiotemporal resolution requirements of clinical diagnosis and treatment tracking. Existing weighted image joint reconstruction methods have shortcomings in terms of accuracy and stability.
The matrix elastic mesh convolution underfimum regularization method is adopted. By using the gradient domain elastic mesh convolution underfimum regularization model and the alternating direction multiplier method, the K-space data of the undersampled weighted image is optimized, the weighted image is reconstructed, and the parameter values are fitted.
While ensuring the stability and accuracy of the solution, the speedup of parametric imaging is increased, the image reconstruction accuracy and parameter estimation accuracy are improved, and the influence of artifacts is reduced.
Smart Images
Figure CN116266350B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of image imaging technology, and in particular to a matrix elastic mesh convolution underfimum regularized accelerated magnetic resonance parametric imaging method, terminal device, and computer storage medium. Background Technology
[0002] Magnetic resonance imaging (MRI) is an extremely important medical imaging diagnostic tool, driving further development in scientific research across disciplines including medicine, neurology, and cognitive science. Among these, parametric MRI, as one of the advantageous techniques in MRI, can effectively characterize human tissues. For example, the T2 parameter can assess the degree of multiple sclerosis and iron overload in patients; T... 1ρ Parametric imaging parameters can assess myocardial fibrosis, joint damage, and degeneration. However, parametric MRI requires acquiring a series of contrast-weighted images at different echo or flip angle times to accurately estimate parameter values. This results in lengthy imaging cycles that are prone to motion artifacts and severely limit spatiotemporal resolution during dynamic quantitative scanning, making it difficult to meet the needs of clinical diagnosis and treatment monitoring. Therefore, researching how to further accelerate the parametric imaging process while maintaining accurate parameter estimation has always been a challenging problem in the field of magnetic resonance imaging.
[0003] Sequential joint reconstruction of parametric images and parameter value fitting estimation is one of the mainstream research directions in the field of accelerated parametric imaging. The core consideration of this direction lies in the design of regularized reconstruction methods for all undersampled weighted images. Regularized reconstruction mostly constrains the low rank of the weighted image matrix in the parameter dimension or the projection sparsity in the transform domain, as well as the combination of constraining the low rank and projection sparsity of the matrix.
[0004] However, the core of the effectiveness of methods that sequentially perform parametric image joint reconstruction and parameter value fitting estimation lies in first providing or designing a powerful weighted image joint reconstruction method. Currently, most weighted image joint reconstructions are based on the low-rank property of the Cascorati matrix, which is presented by the exponential decrease of the image intensity in the parametric dimension. However, this type of method, which directly targets the low-rank property of the image domain matrix, has shortcomings in terms of accuracy and stability characterization. Summary of the Invention
[0005] This application provides a matrix elastic mesh convolution underfimum regularized accelerated magnetic resonance parametric imaging method, a terminal device, and a computer storage medium.
[0006] This application provides a matrix elastic mesh convolution underfimum regularization method for accelerating magnetic resonance parametric imaging.
[0007] The matrix elastic mesh convolution lower bound regularization accelerated magnetic resonance parametric imaging method includes:
[0008] Read the K-space data of a set of undersampled weighted maps;
[0009] The K-space data of the undersampled weighted image is input into a pre-designed gradient domain elastic mesh convolutional infimum regularization model to solve for a set of weighted images;
[0010] Parameter values are fitted based on a set of weighted images obtained from the reconstruction.
[0011] The step of inputting the K-space data of the undersampled weighted graph into a pre-designed gradient domain elastic mesh convolutional infimum regularization model further includes:
[0012] Each weighted image in the K-space data of the undersampled weighted image is divided into several local image blocks;
[0013] Record the index of each local image block in the weighted image matrix;
[0014] The K-space data of the undersampled weighted image and the index of each local image block are input into the gradient domain elastic mesh convolutional underfimum regularization model.
[0015] The matrix elastic mesh convolutional underfimum regularized accelerated magnetic resonance parametric imaging method further includes:
[0016] Obtain an initialized elastic mesh model, wherein the initialized elastic mesh model is for the complete image;
[0017] A local elastic mesh model is constructed based on the initialized elastic mesh model, wherein the local elastic mesh model is for local patches of the image;
[0018] A regularization term is introduced into the local elastic network model to form the gradient domain elastic network convolutional fimum regularization model.
[0019] The regularization terms include time-domain regularization terms and spatial-domain regularization terms.
[0020] The step of introducing a regularization term into the local elastic network model to form the gradient domain elastic network convolutional fimum regularization model includes:
[0021] Regularization terms and feature operators are introduced into the local elastic network model to form the gradient domain elastic network convolutional fimum regularization model;
[0022] The feature operator is a gradient domain operator.
[0023] The matrix elastic mesh convolutional underfimum regularized accelerated magnetic resonance parametric imaging method further includes:
[0024] An iterative solution algorithm is constructed based on the alternating direction multiplier method to solve the gradient domain elastic mesh convolutional fimum regularization model.
[0025] The weighted image matrix is optimized and solved using the iterative solution algorithm based on the gradient domain elastic mesh convolution underfimum regularization model to obtain a set of weighted images.
[0026] The step of optimizing the gradient domain elastic mesh convolutional underfimum regularization model based on the iterative solution algorithm to obtain a set of weighted images includes:
[0027] An augmented Lagrangian function is constructed based on a gradient-domain elastic mesh convolutional fimum regularization model, wherein the augmented Lagrangian function includes several operators;
[0028] The problem function for solving each operator is derived from the augmented Lagrange function;
[0029] Input the K-space data of the undersampled weighted graph, and iteratively solve the problem function of solving the plurality of operators to obtain the values of the plurality of operators;
[0030] Image reconstruction is performed based on the values of the aforementioned operators to obtain a set of weighted images.
[0031] The step of reading a set of K-space data of an undersampled weighted map includes:
[0032] K-space data of the undersampled weighted map of the weighted image matrix to be reconstructed is obtained based on the K-space undersampling mechanism;
[0033] The K-space data of the undersampled weighted graph is input into the gradient domain elastic mesh convolutional underfimum regularization model.
[0034] This application also provides a terminal device, the terminal device including a memory and a processor, wherein the memory is coupled to the processor;
[0035] The memory is used to store program data, and the processor is used to execute the program data to implement the above-described matrix elastic mesh convolution underfimum regularization accelerated magnetic resonance parametric imaging method.
[0036] This application also provides a computer storage medium for storing program data, which, when executed by a processor, is used to implement the above-described matrix elastic mesh convolution underfimum regularized accelerated magnetic resonance parametric imaging method.
[0037] The beneficial effects of this application are as follows: the terminal device reads K-space data of a set of undersampled weighted images; the K-space data of the undersampled weighted images are input into a pre-designed gradient domain elastic mesh convolutional underfimum regularization model, and the ADMM algorithm is used to optimize the solution to obtain a set of weighted images; parameter values are fitted based on the reconstructed set of weighted images. Through the above method, the matrix elastic mesh convolutional underfimum regularization accelerated magnetic resonance imaging method of this application further improves the acceleration of parametric imaging by designing a gradient domain elastic mesh convolutional underfimum regularization model, while ensuring the stability and accuracy of the solution to the inverse problem. Attached Figure Description
[0038] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:
[0039] Figure 1 This is a flowchart illustrating an embodiment of the matrix elastic mesh convolutional underfimum regularization accelerated magnetic resonance parametric imaging method provided in this application;
[0040] Figure 2 This is a schematic diagram of the framework of an embodiment of the matrix elastic mesh convolutional underfimum regularization accelerated magnetic resonance parametric imaging method provided in this application;
[0041] Figure 3 This is a schematic flowchart of another embodiment of the matrix elastic mesh convolutional underfimum regularization accelerated magnetic resonance parametric imaging method provided in this application;
[0042] Figure 4 This is a flowchart illustrating another embodiment of the matrix elastic mesh convolutional underfimum regularization accelerated magnetic resonance parametric imaging method provided in this application;
[0043] Figure 5 This is a schematic diagram of the "gold standard" weighted images at different acquisition times provided in this application;
[0044] Figure 6 This is a schematic diagram of an imaging method using a traditional local low-rank regularization model in the image domain, as provided in this application.
[0045] Figure 7 This is a schematic diagram of the imaging using the gradient domain elastic mesh convolutional fimum regularization model provided in this application;
[0046] Figure 8 This is a schematic diagram comparing the parameter estimation of the traditional method provided in this application with the matrix elastic mesh convolutional underfimum regularized accelerated magnetic resonance parametric imaging method of this application;
[0047] Figure 9 This is a schematic diagram of the structure of an embodiment of the terminal device provided in this application;
[0048] Figure 10 This is a schematic diagram of the structure of an embodiment of the computer storage medium provided in this application. Detailed Implementation
[0049] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.
[0050] The traditional regularized accelerated magnetic resonance imaging model will be introduced below:
[0051] Specifically, the magnetic resonance imaging data acquisition process based on the K-space undersampling mechanism can be discretized as follows:
[0052] b=Ax+ξ
[0053] in, The weighted image matrix to be reconstructed is represented by each column of the weighted image matrix, where each column represents a weighted image acquired at time t. For the corresponding K-space undersampled data, This is the undersampled Fourier encoding matrix. Let be complex noise that is assumed to follow a Gaussian distribution.
[0054] By exploring priors about the weighted image, the inverse problem described above is typically transformed into an unconstrained optimization problem:
[0055]
[0056] In the above cost function, the first term is the data fitting term, and the second term is the regularization term; λ∈R + This is a regularization parameter used to control the strength of regularization.
[0057] Within the framework of compressed sensing theory or matrix filling theory, many existing methods currently consider the low-rank property of the weighted image matrix and the projection sparsity of the time base. Taking the low-rank constraint of the matrix as an example:
[0058]
[0059] Where, ||x|| * As a convex relaxation constraint for low-rank matrices, namely the nuclear norm constraint (singular value summation).
[0060] Based on the above, this application provides a gradient-domain-based elastic mesh convolutional underfimum regularization method for accelerating magnetic resonance parametric imaging. Please refer to [link to details]. Figure 1 and Figure 2 , Figure 1 This is a flowchart illustrating an embodiment of the matrix elastic mesh convolutional underfimum regularization accelerated magnetic resonance parametric imaging method provided in this application. Figure 2 This is a schematic diagram of the framework of an embodiment of the matrix elastic mesh convolution underfimum regularization accelerated magnetic resonance parametric imaging method provided in this application.
[0061] The matrix elastic mesh convolutional underfimum regularized accelerated magnetic resonance parametric imaging method of this application is applied to a terminal device. This terminal device can be a server or a system in which the server and terminal device cooperate. Accordingly, the various parts of the terminal device, such as units, subunits, modules, and submodules, can be all located in the server, or they can be located separately in the server and the terminal device.
[0062] Furthermore, the aforementioned server can be either hardware or software. When the server is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When the server is software, it can be implemented as multiple software programs or software modules, such as software or software modules used to provide distributed servers, or as a single software program or software module; no specific limitation is made here. In some possible implementations, the matrix elastic mesh convolutional underfimum regularized accelerated magnetic resonance parametric imaging method of this application embodiment can be implemented by a processor calling computer-readable instructions stored in memory.
[0063] It should be noted that the matrix elastic mesh convolution underfimum regularization accelerated magnetic resonance parametric imaging method provided in this application can be applied to any parametric imaging scenario, such as other imaging scenarios with redundancy in the time dimension. Specifically, the matrix elastic mesh convolution underfimum regularization accelerated magnetic resonance parametric imaging method provided in this application can be used in high-quality magnetic resonance parametric imaging acceleration tasks, and will be described in detail in the following description using a magnetic resonance imaging scenario as an example.
[0064] Specifically, such as Figure 1 As shown, the matrix elastic mesh convolutional underfimum regularized accelerated magnetic resonance parametric imaging method of this application specifically includes the following steps:
[0065] Step S11: Read the K-space data of a set of undersampled weighted maps.
[0066] In this embodiment of the application, the terminal device acquires and reads a set of weighted images through a magnetic resonance imaging device to form a weighted image matrix to be reconstructed.
[0067] Furthermore, before inputting the weighted image matrix to be reconstructed into the gradient domain elastic mesh convolutional underfimum regularization model, the terminal device can also undersample the weighted images in the weighted image matrix to be reconstructed. The undersampling process is basically the same as the image processing process described above, and the specific steps are as follows: obtain the K-space undersampled data of the weighted image matrix to be reconstructed based on the K-space undersampling mechanism; input the K-space undersampled data into the gradient domain elastic mesh convolutional underfimum regularization model.
[0068] Step S12: Input the K-space data of the undersampled weighted image into the pre-designed gradient domain elastic mesh convolutional underfimum regularization model, and solve to obtain a set of weighted images.
[0069] In this embodiment, the terminal device uses a pre-designed gradient domain elastic mesh convolutional underfimum regularization model to reconstruct the K-space data of the undersampled weighted image, thereby reconstructing the relevant weighted image.
[0070] Furthermore, to improve the accuracy of image reconstruction, the terminal device can further subdivide the weighted images in the undersampled weighted image before inputting the K-space data of the undersampled weighted image into the model. Specifically, the terminal device divides each weighted image in the undersampled weighted image into several local image blocks and records the index of each local image block in the undersampled weighted image, which reflects the position of the local image block in the undersampled weighted image. Finally, the terminal device inputs the K-space data of the undersampled weighted image and the index of each local image block into a pre-designed gradient domain elastic net convolutional infimum regularization model for image reconstruction.
[0071] For the design process of the gradient-domain elastic mesh convolutional hypothallic regularization model, please refer to [link to relevant documentation]. Figure 3 , Figure 3 This is a schematic flowchart of another embodiment of the matrix elastic mesh convolution underfimum regularization accelerated magnetic resonance parametric imaging method provided in this application.
[0072] Specifically, such as Figure 3 As shown, the matrix elastic mesh convolutional underfimum regularized accelerated magnetic resonance parametric imaging method of this application specifically includes the following steps:
[0073] Step S21: Obtain the initialized elastic mesh model, wherein the initialized elastic mesh model is for the complete image.
[0074] In this embodiment of the application, the terminal device defines a positive number α≥0, and defines a matrix elastic network model as:
[0075]
[0076] Where x is the weighted image matrix.
[0077] Step S22: Build a local elastic mesh model based on the initialized elastic mesh model, wherein the local elastic mesh model is for local blocks of the image.
[0078] In this embodiment, the elastic net model can be used to improve the stability of the solution to the inverse problem, while the introduction of a local elastic net model can achieve better low-rank performance while ensuring solution robustness. Therefore, the terminal device can build a local elastic net model based on the above elastic net model. The local elastic net model is specifically defined as follows:
[0079]
[0080] Where b represents the index of the local image block into which the weighted image x is divided.
[0081] Step S23: Introduce a regularization term into the local elastic network model to form a gradient domain elastic network convolutional fimum regularization model.
[0082] In this embodiment, to achieve better problem-solving accuracy, a regularization method using Infimal Convolutional (IC) is designed for the local matrix elastic network model. This regularization ensures that the adaptive balanced weighted image is regularized in both the temporal and spatial domains. The convolutional infimum is expressed as follows:
[0083]
[0084] Furthermore, if we only consider modeling in the feature domain and enhance the feature reconstruction of the weighted image, then the final regularized reconstruction model of this application embodiment, namely the gradient domain elastic mesh convolutional fimum regularized model, can be defined as:
[0085]
[0086] Where φ is a feature operator; if the feature domain of the image is chosen as the gradient domain, φ is a gradient domain operator used to characterize gradient features; if the feature domain of the image is chosen as the wavelet domain, φ is a wavelet domain operator used to characterize wavelet features; λ∈R + and γ∈R + These are all regularization parameters used to balance the strength of the regularization terms.
[0087] Furthermore, this application proposes an algorithm to solve the gradient domain elastic mesh convolutional underfimum regularization model designed using the aforementioned model design method. Please refer to [link to details]. Figure 4 , Figure 4 This is a flowchart illustrating another embodiment of the matrix elastic mesh convolution underfimum regularization accelerated magnetic resonance parametric imaging method provided in this application.
[0088] Specifically, such as Figure 4 As shown, the matrix elastic mesh convolutional underfimum regularized accelerated magnetic resonance parametric imaging method of this application specifically includes the following steps:
[0089] Step S31: Construct an iterative solution algorithm for the gradient domain elastic network convolutional underfimum regularization model based on the alternating direction multiplier method.
[0090] In this embodiment, the terminal device relies on the Alternating Direction of Method of Multipliers (ADMM) to construct the iterative solution algorithm for the regularization model provided in this application. Specifically, the terminal device constructs the augmented Lagrangian function based on the gradient domain elastic mesh convolutional fimum regularization reconstruction model:
[0091]
[0092] The problem of minimizing the augmented Lagrange function is broken down into solving multiple subproblems. The terminal device can solve the corresponding subproblems according to different variables. The subproblem x outputs the reconstructed weighted image.
[0093] Step S32: Optimize the gradient domain elastic mesh convolutional underfimum regularization model based on the iterative solution algorithm to obtain a set of weighted images.
[0094] In this embodiment of the application, the problem of solving the above-mentioned operators can be expressed as follows:
[0095] The solution to the x operator problem can be expressed as follows:
[0096]
[0097] The solution to the v operator problem can be expressed as follows:
[0098]
[0099] The problem of solving the x and v operators described above is a quadratic programming problem, which can be solved using the conjugate gradient descent method.
[0100] The solution to the s operator problem can be expressed as follows:
[0101]
[0102] The solution to the w operator problem can be expressed as follows:
[0103]
[0104] The above problems concerning the s operator and the w operator both have closed-form solutions:
[0105]
[0106] The solution to the l-operator problem can be expressed as follows:
[0107] l = l + μ1(φx - vs)
[0108] The solution to the m-operator problem can be expressed as follows:
[0109] m = m + μ2(vw)
[0110] The terminal device inputs the K-space data of the undersampled weighted image into the problem of solving the above-mentioned operators. After iterative solving using the above-mentioned solution method, the values of the operators are obtained. Among them, the value of the x operator is used to reconstruct the image to obtain the reconstructed weighted image.
[0111] It should be noted that, apart from the original dual-class algorithm of the above embodiments, other algorithms based on proximal gradients are also applicable to the algorithm design and processing of the gradient domain elastic mesh convolutional underfimum regularization model in this application, which will not be described in detail here.
[0112] Step S13: Fit parameter values based on a set of weighted images obtained from the reconstruction.
[0113] In this embodiment, the terminal device reads K-space data of a set of undersampled weighted images; inputs the K-space data of the undersampled weighted images into a pre-designed gradient domain elastic mesh convolutional underfimum regularization model, and uses the ADMM algorithm to optimize the solution to obtain a set of weighted images; and fits parameter values based on the reconstructed set of weighted images. Through the above method, the matrix elastic mesh convolutional underfimum regularization accelerated magnetic resonance imaging method of this application, by designing a gradient domain elastic mesh convolutional underfimum regularization model, further improves the acceleration of parameter imaging while ensuring the stability and accuracy of the solution to the inverse problem. Through the above method, the matrix elastic mesh convolutional underfimum regularization accelerated magnetic resonance imaging method of this application, by designing a gradient domain elastic mesh convolutional underfimum regularization model, improves the solution stability of low-rank solutions, thereby obtaining better solution accuracy. This application proposes a gradient-feature-based matrix elastic mesh convolutional underfimum regularization method, which is effectively used to accelerate magnetic resonance parametric imaging. It also proposes a local matrix low-rank characterization mechanism based on gradient domain features, which, compared to the traditional global matrix low-rank characterization in the image domain, has better weighted image detail reconstruction capabilities during the low-rank optimization process. Furthermore, it proposes a convolutional underfimum regularization method for local matrix elastic meshes, balancing the regularization effects of the spatial and temporal domains to improve reconstruction accuracy and thus improve parameter estimation.
[0114] Under the condition of meeting basic clinical application requirements, the matrix elastic mesh convolutional underfimum regularization accelerated magnetic resonance parametric imaging method provided in this application can obtain a higher undersampling rate, that is, a higher acceleration imaging capability; the stability of the elastic mesh model for the solution makes the reconstruction of the weighted image stable, even in noisy conditions; the design of convolutional underfimum regularization can adaptively separate spatiotemporal constraints, and the image reconstruction accuracy and the subsequent parameter estimation accuracy dependent on the weighted image are both improved.
[0115] Furthermore, this application also uses T 1ρ Retrospective sampling simulation of the parametric imaging dataset verified the effectiveness of the matrix elastic mesh convolution underfimum regularization accelerated magnetic resonance parametric imaging method protected by the above embodiments.
[0116] Specifically, in one-dimensional undersampling mode (acceleration rate R = 5.98), the matrix elastic mesh convolutional underfimum regularized accelerated magnetic resonance parametric imaging method protected in the above embodiments is used to effectively jointly reconstruct weighted images at different times, and fit and estimate T. 1ρ The parameters achieve better performance compared to other traditional methods.
[0117] Please refer to the simulation data used in the experiment. Figure 5 , Figure 5 The image displayed shows the "gold standard" weighted images at acquisition times 1, 20, 40, 60, and 80. For magnetic resonance parametric imaging images implemented using a traditional image-domain local low-rank regularized model based on these "gold standard" weighted images, please refer to [link to relevant documentation]. Figure 6 For fast magnetic resonance parametric imaging images implemented using the gradient domain elastic mesh convolutional underfimum regularization model provided in this application based on the above "gold standard" weighted images, please refer to [link to relevant documentation]. Figure 7 .go through Figure 6 and Figure 7 The comparison shows that the joint reconstruction of traditional methods returns a weighted image containing artifacts, while the matrix elastic mesh convolution underfimum regularization accelerated magnetic resonance parametric imaging method provided in this application achieves a good balance between artifacts and detail preservation.
[0118] Please continue reading. Figure 8 , Figure 8 This diagram illustrates a comparison of parameter estimation between the traditional method provided in this application and the matrix elastic mesh convolution underfimum regularized accelerated magnetic resonance imaging method of this application. From... Figure 8 The comparison clearly shows that the final estimated parameter map of the matrix elastic mesh convolution underfimum regularization accelerated magnetic resonance parametric imaging method of this application presents structural details, especially fine details, that are closer to the original image.
[0119] Those skilled in the art will understand that, in the above-described method of the specific implementation, the order in which each step is written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of each step should be determined by its function and possible internal logic.
[0120] To implement the matrix elastic mesh convolution underfimum regularized accelerated magnetic resonance parametric imaging method of the above embodiments, this application also proposes a terminal device, which can be found in the following details. Figure 9 , Figure 9 This is a schematic diagram of the structure of an embodiment of the terminal device provided in this application.
[0121] The terminal device 500 of this application embodiment includes a memory 51 and a processor 52, wherein the memory 51 and the processor 52 are coupled together.
[0122] The memory 51 is used to store program data, and the processor 52 is used to execute the program data to implement the matrix elastic mesh convolution underfimum regularization accelerated magnetic resonance parametric imaging method described in the above embodiments.
[0123] In this embodiment, processor 52 can also be referred to as a CPU (Central Processing Unit). Processor 52 may be an integrated circuit chip with signal processing capabilities. Processor 52 can also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The general-purpose processor can be a microprocessor, or processor 52 can be any conventional processor.
[0124] This application also provides a computer storage medium, such as Figure 10 As shown, the computer storage medium 600 is used to store program data 61. When the program data 61 is executed by the processor, it is used to implement the matrix elastic mesh convolution underfimum regularization accelerated magnetic resonance parametric imaging method as described in the above embodiments.
[0125] This application also provides a computer program product, wherein the computer program product includes a computer program operable to cause a computer to perform the matrix elastic mesh convolutional underfimum regularized accelerated magnetic resonance parametric imaging method as described in the embodiments of this application. The computer program product can be a software installation package.
[0126] The matrix elastic mesh convolution underfimum regularized accelerated magnetic resonance parametric imaging method described in the above embodiments of this application, when implemented as a software functional unit and sold or used as an independent product, can be stored in a device, such as a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0127] The above description is merely an embodiment of this application and does not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.
Claims
1. A matrix elastic net convolution underfimum regularized accelerated magnetic resonance parametric imaging method, characterized in that, The matrix elastic mesh convolution lower bound regularization accelerated magnetic resonance parametric imaging method includes: Read the K-space data of a set of undersampled weighted maps; The K-space data of the undersampled weighted image is input into a pre-designed gradient domain elastic mesh convolutional infimum regularization model to solve for a set of weighted images; Parameter values are fitted based on a set of weighted images obtained from the reconstruction. The gradient domain elastic network convolutional underfimum regularization model introduces time domain regularization and spatial domain regularization terms into the local elastic network model, and uses gradient domain operators as feature operators. The gradient-domain elastic mesh convolutional lower bound regularization model is defined as follows: ； in, For characteristic operators, as well as All are regularization parameters, where b represents the weighted image. The indices of the divided image local blocks, where A is the undersampled Fourier coding matrix. Let y be the weighted image matrix, and y be the corresponding K-space undersampled data.
2. The matrix elastic mesh convolution underfimum regularized accelerated magnetic resonance parametric imaging method according to claim 1, characterized in that, Before inputting the K-space data of the undersampled weighted map into the pre-designed gradient domain elastic mesh convolutional infimum regularization model, the following steps are also included: Each weighted image in the undersampled weighted image is divided into several local image blocks; Record the index of each local image block in the weighted image matrix; The K-space data of the undersampled weighted image and the index of each local image block are input into the gradient domain elastic mesh convolutional underfimum regularization model.
3. The matrix elastic mesh convolution underfimum regularized accelerated magnetic resonance parametric imaging method according to claim 1, characterized in that, The matrix elastic mesh convolution lower bound regularized accelerated magnetic resonance parametric imaging method further includes: An iterative solution algorithm is constructed based on the alternating direction multiplier method to solve the gradient domain elastic mesh convolutional fimum regularization model. The gradient domain elastic mesh convolutional underfimum regularization model is optimized and solved based on the iterative solution algorithm to obtain a set of weighted images.
4. The matrix elastic mesh convolution underfimum regularized accelerated magnetic resonance parametric imaging method according to claim 3, characterized in that, The optimization solution of the gradient domain elastic mesh convolutional fimum regularization model based on the iterative solution algorithm to obtain a set of weighted images includes: An augmented Lagrangian function is constructed based on a gradient-domain elastic mesh convolutional fimum regularization model, wherein the augmented Lagrangian function includes several operators; The problem function for solving each operator is derived from the augmented Lagrange function; Input the K-space data of the undersampled weighted graph, and iteratively solve the problem function of solving the plurality of operators to obtain the values of the plurality of operators; Image reconstruction is performed based on the values of the aforementioned operators to obtain a set of weighted images.
5. The matrix elastic mesh convolution underfimum regularized accelerated magnetic resonance parametric imaging method according to claim 1, characterized in that, The process of reading a set of K-space data of an undersampled weighted map includes: K-space data of the undersampled weighted map of the weighted image matrix to be reconstructed is obtained based on the K-space undersampling mechanism; The K-space data of the undersampled weighted graph is input into the gradient domain elastic mesh convolutional underfimum regularization model.
6. A terminal device, characterized in that, The terminal device includes a memory and a processor, wherein the memory is coupled to the processor; The memory is used to store program data, and the processor is used to execute the program data to implement the matrix elastic mesh convolution underfimum regularized accelerated magnetic resonance parametric imaging method according to any one of claims 1-5.
7. A computer storage medium, characterized in that, The computer storage medium is used to store program data, which, when executed by a processor, is used to implement the matrix elastic mesh convolution underfimum regularized accelerated magnetic resonance parametric imaging method according to any one of claims 1-5.