Magnetic particle imaging reconstruction method based on calibrated magnetic field generation system matrix

Abstract

The present invention belongs to the field of magnetic particle imaging technology and relates to a magnetic particle imaging reconstruction method based on generating a system matrix using a calibrated magnetic field. The method aims to address the poor applicability of traditional magnetic particle imaging reconstruction methods in asymmetric bilateral magnetic particle imaging devices (A-MPIs) and the inability to image arbitrary regions. The present invention includes the following steps: gridding the imaging area, calibrating the magnetic field intensity, and obtaining the magnetic field gradient distribution at each point; obtaining a theoretical system matrix using mathematical calculations; randomly moving the positions of points without magnetic fields for scanning to obtain the measured magnetic field gradient distribution and system matrix and construct a data set; expanding the data set based on a simulation system; constructing a deep learning model, inputting the expanded data set and measured data into the training deep learning model; inputting the theoretical system matrix at any position, and using the trained model to obtain an accurate system matrix. The present invention can obtain the system matrix at any position using an A-MPI device, thereby enabling imaging at any position.

Description

Technical Field

[0001] The present invention belongs to the technical field of magnetic particle imaging, and in particular relates to a magnetic particle imaging reconstruction method based on a calibration magnetic field generation system matrix. Background Art

[0002] Magnetic particle imaging technology was first proposed in 2005. It uses superparamagnetic iron oxide particles (SPIOs) as contrast agents for imaging, which can produce tomographic images with high sensitivity and high contrast, and has the characteristics of quantification and no depth attenuation. As a new type of molecular imaging technology, it plays an increasingly important role in biomedical research.

[0003] The asymmetric, bilateral magnetic particle imaging device is designed to address the limited imaging range and insufficient depth of imaging at the human torso level of existing magnetic nanoparticle imaging devices. Its main structure consists of a fixed end and a handheld end. The fixed end generates a uniform, variable magnetic field and drives the free magnetic field point to move along the depth direction. The magnetic field-free point is generated by the permanent magnet and magnetic field generating coils in the handheld end. In imaging methods targeting a wide field of view, the position of the magnetic field-free point can be moved by moving the handheld end to achieve scanning of different areas. During the imaging scan, the position of the handheld end allows for targeted local imaging or full-body imaging of the object being measured, and imaging at different depths can be achieved based on the magnitude of the applied current.

[0004] With the development of magnetic particle imaging technology, the commonly used reconstruction methods are mainly system matrix (SM) and X-Space. The X-Space reconstruction method reconstructs the time domain signal of the nonlinear response signal. However, for asymmetric bilateral magnetic particle imaging devices (A-MPI) with small gradients, orthogonal excitation, and orthogonal reception, X-Space reconstruction is prone to produce large image artifacts. The imaging performance of this method depends largely on the quality of the point spread function, which generally performs better when the gradient is large. However, although the A-MPI device has a large imaging field of view, the rapid decline in the selection field gradient will reduce the imaging quality, making the X-Space method ineffective for image reconstruction of A-MPI devices.

[0005] In the reconstruction method based on the system matrix, there are two ways to obtain the system matrix: model-based system matrix calibration and measurement-based system matrix calibration. In the model-based calibration method, due to the complex magnetization behavior of SPIOs in a complex magnetic field environment, it is extremely challenging to accurately establish a physical model. At present, although the measurement-based acquisition method is commonly used and relatively accurate, it is necessary to move the sampling point in the entire imaging field of view to obtain the particle response characteristics at different positions, and the measurement process is extremely time-consuming. For the A-MPI device, its imaging field of view is large and it can be imaged at any position and depth, which theoretically will generate countless system matrices. If the traditional measurement method is used to calibrate the system matrix every time a position is moved during measurement, it is almost impossible to achieve in actual operation, resulting in the traditional reconstruction method based on system matrix calibration being difficult to implement on the A-MPI device, and it is in urgent need of innovation and improvement. Summary of the Invention

[0006] In order to solve the above-mentioned problems in the prior art, namely, to solve the problem that the prior art magnetic particle imaging reconstruction method has poor applicability in A-MPI equipment and cannot image arbitrary areas, the first aspect of the present invention proposes a magnetic particle imaging reconstruction method based on calibrating the magnetic field to generate a system matrix. The method is applied to a magnetic particle imaging device based on an asymmetric bilateral structure, and the traditional measurement-based system matrix calibration is replaced by magnetic field calibration. The method includes the following steps:

[0007] S100, dividing the imaging space into grids, calibrating the magnetic field strength at each point in the grid, and determining the overall magnetic field gradient distribution and the excitation magnetic field strength;

[0008] S200, using the second kind Chebyshev polynomial to determine the mathematical expression of the ideal system matrix of each point, and obtain the theoretical system matrix;

[0009] S300, randomly selecting a position, calibrating the actual system matrix within the imaging field of view of the position, and combining the overall magnetic field gradient distribution to form a data pair of magnetic field gradient distribution-system matrix to construct a data set;

[0010] S400, add real noise through simulation system to expand the data set;

[0011] S500, constructing a U-net network model, and using the expanded data set to pre-train the U-net network model;

[0012] S600, fine-tuning the pre-trained U-net network model using data of the magnetic field gradient distribution-system matrix to obtain a fine-tuned U-net network model;

[0013] S700 , inputting the magnetic field gradient value of any position to be measured and the theoretical system matrix into the fine-tuned U-net network model to obtain the system matrix of the position to be measured, and generating a reconstructed image based on the system matrix.

[0014] In some preferred embodiments, the magnetic field strength calibration method is as follows:

[0015] S110, calibrating the divergent magnetic field of the permanent magnet at the handheld end, gridding the uniform magnetic field of the permanent magnet at the fixed end, and calibrating the uniform magnetic field intensity at each grid point;

[0016] S120, obtaining a magnetic field gradient value at each grid by fitting calculation based on the uniform magnetic field strength;

[0017] S130, interpolating the magnetic field gradient distribution, and defining the gradient value of each position according to the position of each grid point;

[0018] S140, calibrating the excitation magnetic field strength of the handheld terminal excitation coil at the grid;

[0019] S150 , determining the overall magnetic field gradient distribution by combining the gradient value and the excitation magnetic field strength.

[0020] In some preferred embodiments, the theoretical system matrix is ​​obtained by:

[0021] The tensor product of the second kind of Chebyshev polynomials is used to calculate the matrix of the two-dimensional system. To express:

[0022] ;

[0023] in, k 、 l represents the order of the Chebyshev polynomial, x 、 y express FOV The two-dimensional coordinates under different orders correspond to the frequency components in the actual system matrix; the coordinates correspond to the positions of the actual system matrix, that is, FOV The number of system matrix calibration points;

[0024] Let the excitation frequency in the x and y directions satisfy ,but:

[0025] ;

[0026] ;

[0027] in, Indicates the frequency System Matrix At coordinates The value of Indicates frequency i The corresponding matrix diagram of the entire system; f x 、f y They are x 、 y The excitation frequency of the direction, is the density parameter of the Lissajous trajectory during scanning.

[0028] In some preferred embodiments, the U-net network model includes an encoder and a decoder, the encoder adopts ResNet18, and the decoder has the same number of layers as the encoder; the U-net network model performs feature splicing when outputting, and the features output by different layers in the encoder are fused with the corresponding layers of the decoder.

[0029] In some preferred embodiments, the data set is constructed by:

[0030] S310, selecting magnetic field parameters of a grid corresponding to a desired imaging area according to a position of a single imaging, and obtaining a system matrix based on the magnetic field distribution at the position through actual measurement;

[0031] S320, randomly select the positions of multiple points without magnetic field, measure the system matrix under different magnetic field parameters, and obtain the actual system matrix ; wherein, mid-frequency i The corresponding system matrix diagram is ;

[0032] S330 , combining the actual system matrix with the magnetic field gradient value at the corresponding position in the overall magnetic field gradient distribution to form a data pair of magnetic field gradient distribution-system matrix, and then constructing a data set.

[0033] In some preferred embodiments, the data set is expanded by:

[0034] S410, simulate the magnetic nanoparticle response signal based on the Langevin model to generate a simulation system matrix corresponding to the actual system matrix ;

[0035] S420, the simulation system matrix and the actual system matrix Perform interpolation to obtain the actual noise distribution ;

[0036] S430, setting the gradient distribution of the simulated magnetic field to the randomly sampled magnetic field gradient distribution in the overall magnetic field gradient distribution, simulating again to obtain the system matrix and adding , get the simulated overall magnetic field system matrix ; mid-frequency i The corresponding system matrix diagram is ;

[0037] S440. Combine the overall magnetic field gradient distribution and the simulated overall magnetic field system matrix to obtain a magnetic field gradient distribution-simulation system matrix data pair, thereby completing data set expansion.

[0038] In some preferred embodiments, the magnetic field distribution setting of the simulation system is consistent with the magnetic field distribution setting actually measured.

[0039] In some preferred embodiments, the U-net network model includes an encoder and a decoder, and the U-net network model performs feature splicing at output, fusing the features output by different layers in the encoder with the corresponding layers of the decoder.

[0040] Beneficial effects of the present invention:

[0041] The present invention uses magnetic field calibration to generate a system matrix at any position from the magnetic field distribution through an established deep learning model. This makes it possible to obtain the system matrix at any position under the A-MPI device, thereby enabling imaging at any position. This greatly improves the flexibility of magnetic particle imaging and meets diverse imaging needs, such as comprehensive imaging of complex human structures or precise imaging of specific micro-regions.

[0042] By optimizing the preliminary theoretical system matrix through a deep learning model, an accurate system matrix is ​​obtained, which helps to improve the accuracy of image reconstruction and thus enhance the quality of magnetic particle imaging, providing clearer and more accurate image information for medical diagnosis and biological research.

[0043] It avoids the tedious operation of moving sampling points across the entire imaging field of view when traditionally acquiring a system matrix based on measurement, reduces measurement costs, saves a lot of time, and improves imaging efficiency, making magnetic particle imaging technology more practical and clinically valuable, and is conducive to the widespread promotion of this technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Other features, objects and advantages of the present application will become more apparent upon reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0045] Figure 1 is a flow chart of a magnetic particle imaging reconstruction method based on a calibration magnetic field generation system matrix of the present invention;

[0046] Figure 2 is a flow chart of expanding a data set in an embodiment of the present invention;

[0047] Figure 3 This is a flow chart of obtaining a system matrix at any position based on the U-net network model in an embodiment of the present invention. DETAILED DESCRIPTION

[0048] The present application will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are intended only to illustrate the relevant invention and are not intended to limit the invention. It should also be noted that, for ease of description, only portions relevant to the relevant invention are shown in the accompanying drawings.

[0049] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0050] A-MPI features low gradients, orthogonal excitation, and orthogonal reception. To address the problem that traditional reconstruction methods based on system matrix calibration cannot be directly applied to A-MPI, the present invention provides a magnetic particle imaging reconstruction method based on generating a system matrix using a calibrated magnetic field. This method replaces traditional measurement-based system matrix calibration with magnetic field calibration. By establishing a deep learning model, the system matrix at any position is generated from the magnetic field distribution. A precise system matrix is ​​generated based on the overall magnetic field gradient distribution and the theoretical system matrix, enabling imaging at any position and making it possible to obtain the system matrix at any position using an A-MPI device.

[0051] A magnetic particle imaging reconstruction method based on a calibration magnetic field generation system matrix of the present invention comprises the following steps:

[0052] S100, dividing the imaging space into grids, calibrating the magnetic field strength at each point in the grid, and determining the overall magnetic field gradient distribution and the excitation magnetic field strength;

[0053] S200, using the second kind Chebyshev polynomial to determine the mathematical expression of the ideal system matrix of each point, and obtain the theoretical system matrix;

[0054] S300, randomly selecting a position, calibrating the actual system matrix within the imaging field of view of the position, and combining the overall magnetic field gradient distribution to form a data pair of magnetic field gradient distribution-system matrix to construct a data set;

[0055] S400, add real noise through simulation system to expand the data set;

[0056] S500, constructing a U-net network model, and using the expanded data set to pre-train the U-net network model;

[0057] S600, fine-tuning the pre-trained U-net network model using data of the magnetic field gradient distribution-system matrix to obtain a fine-tuned U-net network model;

[0058] S700 , inputting the magnetic field gradient value of any position to be measured and the theoretical system matrix into the fine-tuned U-net network model to obtain the system matrix of the position to be measured, and generating a reconstructed image based on the system matrix.

[0059] In order to more clearly illustrate the magnetic particle imaging reconstruction method based on the calibration magnetic field generation system matrix of the present invention, the following is combined with Figure 1 Each step in the embodiment of the present invention is described in detail.

[0060] The magnetic particle imaging reconstruction method based on the calibration magnetic field generation system matrix of the first embodiment of the present invention includes the following steps S100 to S700, and is applied to an asymmetric bilateral magnetic particle imaging device. Each step is described in detail as follows:

[0061] S100, dividing the imaging space into grids, calibrating the magnetic field strength at each point in the grid, and determining the overall magnetic field gradient distribution and the excitation magnetic field strength, the method is as follows:

[0062] S110, calibrate the divergent magnetic field of the permanent magnet on the handheld , grid the uniform magnetic field of the fixed end permanent magnet and calibrate the uniform magnetic field intensity at each grid point;

[0063] S120, based on the uniform magnetic field strength, obtain the magnetic field gradient value at each grid by fitting calculation ;

[0064] S130, interpolating the magnetic field gradient distribution, and defining the gradient value of each position according to the position of each grid point;

[0065] S140, calibrate the excitation magnetic field strength of the handheld excitation coil at the grid, defined as ; The total magnetic field at each position is ;

[0066] S150 , determining the overall magnetic field gradient distribution by combining the gradient value and the excitation magnetic field strength.

[0067] In a two-dimensional plane, the grid size is divided according to the actual imaging field of view (FOV). In this embodiment, assuming that the FOV of A-MPI is 18*18cm, the grid is divided according to a step size of 5mm, and the number of grid points is 37*37=1369. The uniform magnetic field intensity at each grid point is calibrated; then the magnetic field gradient value at each grid point is obtained by fitting calculation. Next, the magnetic field gradient distribution is interpolated to increase the resolution to 1mm, and the number of grid points is 181*181=32761. The gradient value at each position is defined according to the position of each grid point. , where i=1,2,……32761; then the total magnetic field is ,because It has two components in the x and y directions, so the final magnetic field gradient distribution is a matrix of 181*181*2.

[0068] S200, using the second kind of Chebyshev polynomials to determine the mathematical expression of the system matrix, and obtain the theoretical system matrix, the method is:

[0069] The order of the Chebyshev polynomial is selected according to the number of system matrix frequency points selected in the actual system matrix. Secondly, the number of grid points is generated according to the size of the actual calibration FOV. According to the formula, the theoretical system matrix expressed by the Chebyshev polynomial at different orders at each calibration point is obtained.

[0070] Preferably, the theoretical system matrix is ​​obtained by:

[0071] The second kind of Chebyshev polynomials expresses the system matrix under uniform excitation, uniform reception, and ideal particle model:

[0072] ;

[0073] Assuming the excitation magnetic field is uniform, the differences in the system matrix between different positions come from the differences in the gradient values. The above formula shows a one-dimensional Chebyshev expression. Under the scanning trajectory of Lissajou, the two-dimensional system matrix is ​​expressed using the tensor product of the second-kind Chebyshev polynomials:

[0074] ;

[0075] in, k 、 l represents the order of the Chebyshev polynomial, x 、 y express FOV The two-dimensional coordinates under different orders correspond to the row vectors in the actual system matrix, that is, the frequency components; the coordinates correspond to the columns of the actual system matrix, that is, FOV The number of system matrix calibration points;

[0076] makex 、 y Directional excitation frequency satisfies ,but:

[0077] ;

[0078] ;

[0079] in, Indicates the frequency System Matrix At coordinates The value of Indicates frequency i The corresponding matrix diagram of the entire system; f x 、f y They are x 、 y The excitation frequency of the direction, is the density parameter of the Lissajous trajectory during scanning.

[0080] S300 , randomly select a position, calibrate the actual system matrix within the imaging field of view of the position, and combine the overall magnetic field gradient distribution obtained in step S100 to form a data pair of magnetic field gradient distribution-system matrix to construct a data set.

[0081] Preferably, the data set is constructed by:

[0082] S310, selecting magnetic field parameters of a grid corresponding to a desired imaging area according to a position of a single imaging, and obtaining a system matrix based on the magnetic field distribution at the position through actual measurement;

[0083] S320, randomly select the positions of multiple points without magnetic field, measure the system matrix under different magnetic field parameters, and obtain the actual system matrix ;

[0084] in, Represents the system matrix diagram corresponding to frequency i in the actual system matrix.

[0085] S330 , combining the actual system matrix with the magnetic field gradient value at the corresponding position in the overall magnetic field gradient distribution to form a data pair of magnetic field gradient distribution-system matrix, and then constructing a data set.

[0086] Preferably, in this embodiment, assuming that the field of view of a single imaging is 2*2cm, the corresponding magnetic field gradient distribution map is a matrix I of 21*21*2; assuming that the resolution of the system matrix during calibration is 1mm, and a system matrix takes k modulation frequencies, then the system matrix is a 21*21*k matrix, where Indicates frequency i The corresponding system matrix diagram.

[0087] S400, adding actual noise to the data set through the simulation system, the method is as follows:

[0088] S410, based on Langevin The model simulates the response signal of magnetic nanoparticles and generates a simulation system matrix corresponding to the actual system matrix ;

[0089] The magnetic field distribution setting of the simulation system is consistent with the magnetic field distribution setting of the actual measurement;

[0090] S420, the simulation system matrix and the actual system matrix Perform interpolation to obtain the actual noise distribution ;

[0091] S430, setting the gradient distribution of the simulated magnetic field to the randomly sampled magnetic field gradient distribution in the overall magnetic field gradient distribution, simulating again to obtain the system matrix and adding , get the simulated overall magnetic field system matrix ;in, Represents the frequency in the matrix of the simulated overall magnetic field system i The corresponding system matrix diagram;

[0092] S440. Combine the overall magnetic field gradient distribution and the simulated overall magnetic field system matrix to obtain a magnetic field gradient distribution-simulation system matrix data pair, thereby completing data set expansion.

[0093] based on Langevin The model simulates the response signal of magnetic nanoparticles, which can be implemented by those skilled in the art based on conventional knowledge and experience, and therefore will not be described in detail in this specification.

[0094] Preferably, in this embodiment, the Langevin function is used to fit the particle response, and the simulated signal is used to generate the corresponding simulated measured system matrix , the parameter settings of the simulation are consistent with those of the actual measurement.

[0095] S500: Construct a U-net network model and pre-train it using the expanded dataset. The U-net network model includes an encoder and a decoder. The encoder uses ResNet18, and the decoder has the same number of layers as the encoder. The U-net network model performs feature concatenation on its output, fusing features output from different encoder layers with those from the corresponding decoder layers.

[0096] Preferably, in this embodiment, the ResNet18 includes four residual modules. The first residual module contains two convolutional layers and has 64 output channels; the second residual module has 128 output channels; the third residual module has 256 output channels; and the fourth residual module has 512 output channels. Each decoding module gradually restores the resolution through upsampling operations and further extracts features through specific convolution operations. This is achieved sequentially by four decoding layers, ensuring effective fusion of features in terms of spatial dimensions and semantic information. The final decoder output passes through the convolutional layers to generate a result consistent with the input image size.

[0097] Through feature splicing, the features output by different layers of the encoder are fused with the corresponding layers of the decoder, thereby retaining more detailed information in the upsampling stage.

[0098] Preferably, when pre-training the U-net network model using the expanded data set, the input data is first pre-processed by:

[0099] Regarding the Each layer Normalize and convert it into a single-channel image, and scale the image to the specified size; add a dimension to the overall magnetic field gradient distribution, which is the Chebyshev polynomial theoretical representation of the system matrix corresponding to frequency i , normalize it, convert it into a three-channel RGB image, and scale the image to the specified size.

[0100] Further preferably, the output of the U-net network model is a single-channel certain layer system matrix diagram or a multi-channel multi-layer system matrix diagram.

[0101] By inputting expanded data, measured data, etc. into the training deep learning model, the relationship between the magnetic field gradient distribution under each magnetic field grid and the measured system matrix is ​​fitted.

[0102] In this embodiment, the specified size is 64*64, and each layer of the system matrix Normalize to the [0,255] interval, convert to a single-channel image, and finally scale the image to 64*64 size; add a dimension to the magnetic field gradient distribution corresponding to the frequency i The Chebyshev polynomial theory representation of the system matrix , and then normalized to the [0,255] interval to become a three-channel RGB image, and finally scaled to 64*64. The input and output of the model can be done in two ways:

[0103] Method 1: The model input is a 64*64*3 image of the magnetic field gradient and the theoretical system matrix in the channel dimension. The model output is a 64*64*1 image of the system matrix at a single channel layer. This method requires training k models.

[0104] Method 2: The model input is a 64*64*3 magnetic field gradient and theoretical system matrix splicing image, and the model output is a 64*64*k multi-channel multi-layer system matrix diagram.

[0105] The measured data set is used for data augmentation to extract the noise distribution of the measured data. Combined with the simulation results, the expanded data close to the measured data is constructed.

[0106] S600 , fine-tune the pre-trained U-net network model using the real data (data pairs of magnetic field gradient distribution and system matrix) obtained in step S300 to obtain a fine-tuned U-net network model and determine the model weights.

[0107] Pre-training on expanded data and fine-tuning on measured data is beneficial to model convergence.

[0108] S700. Input the magnetic field gradient value of any position to be measured and the theoretical system matrix into the fine-tuned U-net network model. The network can generate a system matrix close to the actual measured current position, obtain the system matrix of the position to be measured, and obtain a reconstructed image based on the system matrix and the detected signal based on the system matrix reconstruction method.

[0109] Although the various steps in the above embodiment are described in the above-mentioned order, those skilled in the art will understand that in order to achieve the effect of this embodiment, different steps do not have to be executed in such an order. They can be executed simultaneously (in parallel) or in a reverse order. These simple changes are within the scope of protection of the present invention.

[0110] A magnetic particle imaging and reconstruction system based on a calibration magnetic field generation system matrix according to a second embodiment of the present invention includes a signal acquisition device and a central processing device;

[0111] The signal acquisition device includes an asymmetric bilateral structure magnetic particle imaging device, which includes a fixed end magnetic field generating coil, a handheld end permanent magnet, a handheld end excitation coil, and a handheld end receiving coil; the signal acquisition device is configured to grid the imaging space, calibrate the magnetic field strength at each point in the grid, and obtain a measurement signal;

[0112] The central processing device includes a CPU and a GPU; the central processing device includes a matrix generation module, a model construction module, a model fine-tuning module, a data set construction module, an image reconstruction module and a simulation module;

[0113] The matrix generation module is configured to determine the overall magnetic field gradient distribution and the excitation magnetic field strength based on the calibration data, determine the mathematical expression of the ideal system matrix at each point using the second-kind Chebyshev polynomial, and obtain the theoretical system matrix; determine the actual system matrix based on the measurement signal; and input the magnetic field gradient value at any position to be measured and the theoretical system matrix into the fine-tuned U-net network model to obtain the system matrix at the position to be measured;

[0114] The model building module is configured to build a U-net network model and pre-train the U-net network model using the expanded data set;

[0115] The model fine-tuning module is configured to fine-tune the pre-trained U-net network model using the data of the magnetic field gradient distribution-system matrix to obtain a fine-tuned U-net network model;

[0116] The data set construction module is configured to construct a data set by combining the overall magnetic field gradient distribution to form a data pair of magnetic field gradient distribution-system matrix;

[0117] The simulation module is configured to add actual noise to the expanded data set through the simulation system;

[0118] The image reconstruction module is configured to generate a reconstructed image based on the system matrix of the position to be measured.

[0119] It should be noted that the magnetic particle imaging and reconstruction system based on the calibration magnetic field generation system matrix provided in the above embodiment is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the modules or steps in the embodiments of the present invention can be further decomposed or combined. For example, the modules in the above embodiment can be combined into one module or further divided into multiple sub-modules to complete all or part of the functions described above. The names of the modules and steps involved in the embodiments of the present invention are merely for the purpose of distinguishing the modules or steps and are not to be considered as improper limitations of the present invention.

[0120] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working process and related instructions of the system described above can refer to the corresponding process in the aforementioned method embodiment and will not be repeated here.

[0121] An electronic device according to a third embodiment of the present invention includes:

[0122] at least one processor; and

[0123] a memory communicatively connected to at least one of the processors; wherein,

[0124] The memory stores instructions that can be executed by the processor, and the instructions are used to be executed by the processor to implement the above-mentioned magnetic particle imaging reconstruction method based on the calibration magnetic field generation system matrix.

[0125] A fourth embodiment of the present invention provides a computer-readable storage medium storing computer instructions, wherein the computer instructions are used to be executed by a computer to implement the above-mentioned magnetic particle imaging reconstruction method based on the calibration magnetic field generation system matrix.

[0126] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes and related instructions of the electronic device and computer-readable storage medium described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0127] Those skilled in the art should be able to appreciate that the modules and method steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two, and the programs corresponding to the software modules and method steps can be placed in random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in the art. In order to clearly illustrate the interchangeability of electronic hardware and software, the above description has generally described the components and steps of each example according to their functions. Whether these functions are performed in electronic hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art may 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 the present invention.

[0128] Computer program code for performing the operations of the present application may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0129] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, program segment or a part of code, and the module, program segment or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented by a dedicated hardware-based system that performs the specified function or operation, or can be implemented by a combination of dedicated hardware and computer instructions.

[0130] The terms "first", "second", etc. are used to distinguish similar objects, rather than to describe or indicate a particular order or sequence.

[0131] The term "comprise" or any other similar term is intended to cover non-exclusive inclusion such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0132] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

Claims

1. A magnetic particle imaging reconstruction method based on a calibration magnetic field generation system matrix, applied to a magnetic particle imaging device based on an asymmetric bilateral structure, characterized in that: The method comprises the following steps: S100, dividing the imaging space into grids, calibrating the magnetic field strength at each point in the grid, and determining the overall magnetic field gradient distribution and the excitation magnetic field strength; S200, using the second kind Chebyshev polynomial to determine the mathematical expression of the ideal system matrix of each point, and obtain the theoretical system matrix; S300, randomly selecting a position, calibrating the actual system matrix within the imaging field of view of the position, and combining the overall magnetic field gradient distribution to form a data pair of magnetic field gradient distribution-system matrix to construct a data set; S400, adding actual noise through a simulation system to expand the data set; S500, constructing a U-net network model, and using the expanded data set to pre-train the U-net network model; S600, fine-tuning the pre-trained U-net network model using data of the magnetic field gradient distribution-system matrix to obtain a fine-tuned U-net network model; S700 , inputting the magnetic field gradient value of any position to be measured and the theoretical system matrix into the fine-tuned U-net network model to obtain the system matrix of the position to be measured, and generating a reconstructed image based on the system matrix.

2. The magnetic particle imaging reconstruction method based on the calibration magnetic field generation system matrix according to claim 1, characterized in that: The magnetic field strength calibration method is as follows: S110, calibrating the divergent magnetic field of the permanent magnet at the handheld end, gridding the uniform magnetic field of the permanent magnet at the fixed end, and calibrating the uniform magnetic field intensity at each grid point; S120, obtaining a magnetic field gradient value at each grid by fitting calculation based on the uniform magnetic field strength; S130, interpolating the magnetic field gradient distribution, and defining the gradient value of each position according to the position of each grid point; S140, calibrating the excitation magnetic field strength of the handheld terminal excitation coil at the grid; S150 , determining the overall magnetic field gradient distribution by combining the gradient value and the excitation magnetic field strength.

3. The magnetic particle imaging reconstruction method based on the calibration magnetic field generation system matrix according to claim 1, characterized in that: The theoretical system matrix is ​​obtained as follows: The tensor product of the second kind of Chebyshev polynomials is used to calculate the matrix of the two-dimensional system. To express: ； in, k 、 l represents the order of the Chebyshev polynomial, x 、 y express FOV The two-dimensional coordinates under different orders correspond to the frequency components in the actual system matrix; the coordinates correspond to the positions of the actual system matrix, that is, FOV The number of system matrix calibration points; make x 、 y Directional excitation frequency satisfies ,but: ； ； in, Indicates the frequency System Matrix At coordinates The value of Indicates frequency i The corresponding matrix diagram of the entire system; f x 、f y They are x 、 y The excitation frequency of the direction, is the density parameter of the Lissajous trajectory during scanning.

4. The magnetic particle imaging reconstruction method based on the calibration magnetic field generation system matrix according to claim 2, characterized in that: The method to construct the dataset is: S310, selecting magnetic field parameters of a grid corresponding to a desired imaging area according to a position of a single imaging, and obtaining a system matrix based on the magnetic field distribution at the position through actual measurement; S320, randomly select the positions of multiple points without magnetic field, measure the system matrix under different magnetic field distributions, and obtain the actual system matrix ; Among them, the mid-frequency i The corresponding system matrix diagram is ; S330 , combining the actual system matrix with the magnetic field gradient value at the corresponding position in the overall magnetic field gradient distribution to form a data pair of magnetic field gradient distribution-system matrix, and then constructing a data set.

5. The magnetic particle imaging reconstruction method based on the calibration magnetic field generation system matrix according to claim 4, characterized in that: The dataset is expanded by: S410, based on Langevin The model simulates the response signal of magnetic nanoparticles and generates a simulation system matrix corresponding to the actual system matrix ; S420, the simulation system matrix and the actual system matrix Perform interpolation to obtain the actual noise distribution ; S430, setting the gradient distribution of the simulated magnetic field to the randomly sampled magnetic field gradient distribution in the overall magnetic field gradient distribution, simulating again to obtain the system matrix and adding , get the simulated overall magnetic field system matrix ; Among them, the The system matrix diagram corresponding to the intermediate frequency i is: ; S440. Combine the overall magnetic field gradient distribution and the simulated overall magnetic field system matrix to obtain a magnetic field gradient distribution-simulation system matrix data pair, thereby completing data set expansion.

6. The magnetic particle imaging reconstruction method based on the calibration magnetic field generation system matrix according to claim 5, characterized in that: The magnetic field distribution setting of the simulation system is consistent with the magnetic field distribution setting of actual measurement.

7. The magnetic particle imaging reconstruction method based on the calibration magnetic field generation system matrix according to claim 1, characterized in that: The U-net network model includes an encoder and a decoder. The U-net network model performs feature splicing at output, and fuses the features output by different layers in the encoder with the corresponding layers of the decoder.

8. The magnetic particle imaging reconstruction method based on a calibration magnetic field generation system matrix according to any one of claims 5 to 6, characterized in that: Use the expanded data set to pre-train the U-net network model. First, pre-process the input data. The method is as follows: Regarding the Each layer Normalize, convert to a single-channel image, and scale the image to the specified size; The overall magnetic field gradient distribution is expressed by the Chebyshev polynomial theory by adding a dimension to the system matrix corresponding to frequency i. , then normalize it, convert it into a three-channel RGB image, and scale the image to the specified size.

9. The magnetic particle imaging reconstruction method based on the calibration magnetic field generation system matrix according to claim 8, characterized in that: described and All are normalized to the range [0, 255].

10. The magnetic particle imaging reconstruction method based on the calibration magnetic field generation system matrix according to claim 9, characterized in that: The output of the U-net network model is a single-channel layer system matrix diagram or a multi-channel multi-layer system matrix diagram.

Images