Magnetic particle imaging device system matrix calibration method, apparatus, and device

Abstract

The application provides a magnetic particle imaging device system matrix calibration method, device and equipment. The method comprises: performing sampling processing on a preset two-dimensional slice layer based on a cross-down sampling method to obtain original LR SM data; the preset two-dimensional slice layer is each two-dimensional slice layer in a three-dimensional FOV of an MPI device; performing frequency screening on the original LR SM data to obtain high signal-to-noise ratio LR SM data; converting each row of the high signal-to-noise ratio LR SM data into a three-dimensional structure to obtain a plurality of three-dimensional LR SM data; taking L layers and L+1 layers of each three-dimensional LR SM data as first input data and inputting the first input data into a pre-trained system matrix calibration network to perform system matrix calibration processing on the L layers of the three-dimensional LR SM data to obtain calibrated HR SM data; and more prior information of each slice layer can be obtained by staggered sampling on adjacent layers of the preset two-dimensional slice layer, and the system matrix calibration precision is improved.

Description

Technical Field

[0001] This invention relates to the field of biomedical imaging technology, specifically to a matrix calibration method, apparatus, and device for a magnetic particle imaging equipment system. Background Technology

[0002] Magnetic Particle Imaging (MPI), as a novel molecular imaging technique, has shown broad application prospects in the biomedical field due to its unique advantages of high sensitivity, high temporal resolution, and lack of spatial depth limitations. MPI utilizes the nonlinear magnetization response of magnetic particles. Under the combined action of a uniform driving field and a static gradient field, it can generate magnetic particle signals and encode their spatial positions. By reconstructing the received magnetic particle signals within the entire field of view (FOV), the visualized concentration distribution of magnetic particles in the field of view can be obtained. The system matrix (SM)-based MPI reconstruction method can achieve higher reconstructed image quality. In MPI, the measurement-based system matrix is ​​obtained by traversing all grid positions of the sample within the entire FOV using a mechanical translation stage and recording the magnetic particle response signals at the corresponding positions. This is very time-consuming for current more advanced 3D imaging devices suitable for human body size. Therefore, how to accelerate the system matrix calibration process of MPI devices has become a research hotspot in the MPI field.

[0003] Existing system matrix calibration methods mainly include recovering the fully sampled system matrix from the undersampled system matrix using compressed sensing. However, the quality of the fully sampled system matrix recovered by this method decreases as the number of measurement positions decreases, while increasing the number of measurement positions significantly increases the calibration time cost. Some studies have applied deep learning-based super-resolution methods to MPI system matrix calibration, that is, using equally spaced sampling to obtain the LR SM (Low Resolution System Matrix) as prior information for the HR SM (High Resolution System Matrix), and then feeding the obtained pairs of LR SM and HR SM into a CNN-based super-resolution network for training. However, this method has limited available prior information, resulting in low recovery accuracy.

[0004] Therefore, existing system matrix calibration methods for magnetic particle imaging equipment suffer from problems such as long calibration time and low calibration accuracy. Summary of the Invention

[0005] To address the aforementioned problems in the prior art, the present invention provides a method, apparatus, and device for calibrating a matrix of a magnetic particle imaging equipment system.

[0006] The technical problem to be solved by this invention is achieved through the following technical solution:

[0007] In a first aspect, the present invention provides a matrix calibration method for a magnetic particle imaging device system, comprising:

[0008] Sampling processing is performed on the preset two-dimensional slice layer based on the cross-sampling method to obtain the original LR SM data; the preset two-dimensional slice layer is each two-dimensional slice layer within the three-dimensional FOV of the MPI device;

[0009] Frequency filtering was performed on the raw LR SM data to obtain high signal-to-noise ratio LR SM data;

[0010] Each row of the high signal-to-noise ratio LR SM data is converted into a three-dimensional structure to obtain multiple three-dimensional LR SM data.

[0011] Layers L and L+1 of each 3D LR SM data are used as the first input data and fed into a pre-trained system matrix calibration network to perform system matrix calibration on layer L of the 3D LR SM data, resulting in calibrated HR SM data. Here, L∈{1,2,...,Z1}, and Z1 represents the total number of preset 2D slice layers. The cross-sampling method is a staggered sampling of adjacent layers of the preset 2D slice layers. The pre-trained system matrix calibration network is trained using HR SM sample data and 3D LR SM training data. The 3D LR SM training data is sampled from the HR SM sample data based on the cross-sampling method.

[0012] Optionally, sampling is performed on a preset two-dimensional slice layer using a cross-sampling method to obtain the original LR SM data, including:

[0013] Based on the cross-sampling method, corresponding sampling points are obtained in each preset two-dimensional slice layer;

[0014] The preset two-dimensional slice layer is sampled according to the sampling points to obtain sub-LR SM data; the sub-LR SM data is the LR SM sampling data of each two-dimensional slice layer.

[0015] The sub-LR SM data are combined into a matrix to obtain the original LR SM data.

[0016] Alternatively, the cross-sampling method can be expressed as:

[0017] When the number of preset two-dimensional slice layers L is odd:

[0018]

[0019] When the number of preset two-dimensional slice layers L is even:

[0020]

[0021] Where, p L (x,y) represents the sampling location point, where x represents the horizontal coordinate index of the preset two-dimensional slice layer, and y represents the vertical coordinate index of the preset two-dimensional slice layer.

[0022] Optionally, frequency filtering is performed on the original LR SM data to obtain high signal-to-noise ratio LR SM data, including:

[0023] Perform a Fourier transform on the original LR SM data to obtain frequency domain LR SM data;

[0024] Data exceeding a preset frequency threshold in the frequency domain LR SM data are used to construct high signal-to-noise ratio LR SM data.

[0025] Optionally, the pre-trained system matrix calibration network includes: a data fusion module, a feature encoding module, and an upsampling reconstruction module; the upsampling reconstruction module includes: a first upsampling reconstruction module and a second upsampling reconstruction module;

[0026] The data fusion module, the feature encoding module, and the second upsampling reconstruction module are connected in series to form the first structure;

[0027] The first upsampling reconstruction module is connected in parallel with the first structure.

[0028] Optionally, the training process of the pre-trained system matrix calibration network includes:

[0029] Obtain HR SM sample data;

[0030] Frequency filtering was performed on the HR SM sample data to obtain high signal-to-noise ratio HR SM sample data;

[0031] For each row of the high signal-to-noise ratio HR SM sample data, a 3D transformation process is performed to obtain multiple 3D HR SM initial data;

[0032] Multiple 3D HRSM initial data are sampled and processed using the cross-sampling method to obtain multiple 3D LRSM training data;

[0033] The L-th layer, L+1-th layer, and high signal-to-noise ratio (SNR) HRSM sample data corresponding to the L-th layer of each 3D LRSM training data are input into the initial system matrix calibration network, and the initial system matrix calibration network is trained to obtain the pre-trained system matrix calibration network; where L∈{1,2,...,Z2}, and Z2 represents the total number of layers of the 3D LRSM training data; the L-th and L+1-th layers of the 3D LRSM sample data are the second input data, and the high SNR HRSM sample data corresponding to the L-th layer of the 3D LRSM training data are the label data; the initial system matrix calibration network and the pre-trained system matrix calibration network have the same structure.

[0034] Optionally, the L-th layer of each 3D LRSM training data, the L+1-th layer of the 3D LRSM training data, and the high signal-to-noise ratio (SNR) HRSM sample data corresponding to the L-th layer of the 3D LRSM training data are input into the initial system matrix calibration network, and the initial system matrix calibration network is trained to obtain a pre-trained system matrix calibration network, including:

[0035] Input the L-th layer of each 3D LRSM training data, the L+1-th layer of the 3D LRSM training data, and the high signal-to-noise ratio HRSM sample data corresponding to the L-th layer of the 3D LRSM training data into the initial system matrix calibration network;

[0036] The initial system matrix calibration network is continuously trained in the direction of decreasing loss function;

[0037] When the loss function meets a preset loss threshold or the number of iterations meets a preset iteration threshold, the corresponding initial system matrix calibration network is used as the pre-trained system matrix calibration network.

[0038] Optionally, the feature encoding module includes: a convolutional module and a Transformer module connected in sequence;

[0039] The Transformer module includes N Transformer encoders connected in series.

[0040] In a second aspect, the present invention provides a matrix calibration device for a magnetic particle imaging equipment system, which includes: a sampling unit, a screening unit, a conversion unit, and an input / output unit.

[0041] The sampling unit is used to: perform sampling processing based on the cross-sampling method in a preset two-dimensional slice layer to obtain the original LR SM data; the preset two-dimensional slice layer is each two-dimensional slice layer within the three-dimensional FOV of the MPI device;

[0042] The filtering unit is used to: perform frequency filtering on the raw LR SM data to obtain high signal-to-noise ratio LR SM data;

[0043] The conversion unit is used to convert each row of high signal-to-noise ratio LRSM data into a three-dimensional structure to obtain multiple three-dimensional LRSM data.

[0044] The input / output unit is used to: take the L and L+1 layers of each 3D LR SM data as the first input data and input them into the pre-trained system matrix calibration network to perform system matrix calibration processing on the L layer of the 3D LR SM data to obtain the calibrated HR SM data; where L∈{1,2,...,Z1}, Z1 represents the total number of preset 2D slice layers; the cross-sampling method is to perform staggered sampling on adjacent layers of the preset 2D slice layer; the pre-trained system matrix calibration network is trained together using HR SM sample data and 3D LR SM training data; the 3D LR SM training data is sampled from the HR SM sample data based on the cross-sampling method.

[0045] Thirdly, the present invention provides a magnetic particle imaging device system matrix calibration device, comprising: a processor, a storage medium and a bus, wherein the storage medium stores machine-readable instructions executable by the processor, and when the magnetic particle imaging device system matrix calibration is ready to run, the processor communicates with the storage medium via the bus, and the processor executes the machine-readable instructions to perform the steps of the magnetic particle imaging device system matrix calibration method of the first aspect described above.

[0046] This invention provides a method, apparatus, and device for calibrating a matrix of a magnetic particle imaging equipment system. The magnetic particle imaging device system matrix calibration method includes: sampling processing based on a cross-sampling method in a preset two-dimensional slice layer to obtain original LR SM data; the preset two-dimensional slice layer is each two-dimensional slice layer within the three-dimensional FOV of the MPI device; frequency filtering is performed on the original LRSM data to obtain high signal-to-noise ratio LR SM data; each row of the high signal-to-noise ratio LR SM data is converted into a three-dimensional structure to obtain multiple three-dimensional LR SM data; the L layer and L+1 layer of each three-dimensional LR SM data are used as the first input data and input into a pre-trained system matrix calibration network to perform system matrix calibration processing on the L layer of the three-dimensional LR SM data to obtain calibrated HR SM data; where L∈{1,2,...,Z1}, Z1 represents the total number of layers in the preset two-dimensional slice layer; the cross-sampling method is a staggered sampling of adjacent layers of the preset two-dimensional slice layer; the pre-trained system matrix calibration network is trained using HR SM sample data and three-dimensional LR SM training data; the three-dimensional LR SM training data is sampled from the HR SM sample data based on the cross-sampling method. In this invention, by performing staggered sampling on adjacent layers of a preset two-dimensional slice layer, and performing system matrix calibration processing on the original LR SM data obtained by staggered sampling based on a pre-trained system matrix calibration network, the calibrated HR SM data is finally obtained. Since this method does not require the measurement of magnetic particle response signals at all measurement positions, it improves the system matrix calibration time of the MPI device. In addition, since this invention uses staggered sampling on adjacent layers in the preset two-dimensional slice layer, it avoids the problem of fixed data information and little prior information caused by existing equal-interval sampling, and can obtain more prior information at the same sampling rate, thereby improving the system matrix calibration accuracy.

[0047] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0048] Figure 1 A schematic flowchart of a matrix calibration method for a magnetic particle imaging device system provided in an embodiment of the present invention;

[0049] Figure 2 This is a schematic diagram of 2x cross-sampling provided in an embodiment of the present invention;

[0050] Figure 3 A schematic diagram illustrating the calibration process of the pre-trained system matrix calibration network provided in an embodiment of the present invention;

[0051] Figure 4 This is a schematic diagram of the calibrated system matrix image generated based on the method of the present invention, provided in an embodiment of the present invention.

[0052] Figure 5 This is a schematic diagram of the structure of a matrix calibration device for a magnetic particle imaging equipment system provided in an embodiment of the present invention;

[0053] Figure 6 This is a schematic diagram of the structure of a magnetic particle imaging equipment system matrix calibration device provided in an embodiment of the present invention. Detailed Implementation

[0054] The present invention will be further described in detail below with reference to specific embodiments, but the implementation of the present invention is not limited thereto.

[0055] To improve the system matrix calibration time and accuracy of MPI devices, this invention provides a system matrix calibration method for magnetic particle imaging devices. Figure 1 This is a schematic flowchart illustrating a matrix calibration method for a magnetic particle imaging device system provided in an embodiment of the present invention. Figure 1 As shown, it includes:

[0056] S101. Sampling is performed on the preset two-dimensional slice layer based on the cross-sampling method to obtain the original LR SM data. The preset two-dimensional slice layer is each two-dimensional slice layer within the three-dimensional FOV of the MPI device.

[0057] Optionally, sampling is performed on a preset two-dimensional slice layer using a cross-sampling method to obtain the original LR SM data, including:

[0058] Based on the cross-sampling method, corresponding sampling points are obtained in each preset two-dimensional slice layer;

[0059] The preset two-dimensional slice layer is sampled according to the sampling points to obtain sub-LR SM data; the sub-LR SM data is the LR SM sampling data of each two-dimensional slice layer.

[0060] The sub-LR SM data are combined into a matrix to obtain the original LR SM data.

[0061] Alternatively, the cross-sampling method can be expressed as:

[0062] When the number of preset two-dimensional slice layers L is odd:

[0063]

[0064] When the number of preset two-dimensional slice layers L is even:

[0065]

[0066] Where, p L(x,y) represents the sampling location point, where x represents the horizontal coordinate index of the preset two-dimensional slice layer, and y represents the vertical coordinate index of the preset two-dimensional slice layer.

[0067] Additionally, in this embodiment of the invention, it can be set that when p L The position is selected when (x,y) is 1, and when p L The position is not selected when (x,y) takes a value of 0.

[0068] S102. Frequency filtering is performed on the original LR SM data to obtain high signal-to-noise ratio LR SM data.

[0069] Optionally, S102 may specifically include:

[0070] Perform a Fourier transform on the original LR SM data to obtain frequency domain LR SM data;

[0071] Data exceeding a preset frequency threshold in the frequency domain LR SM data are used to construct high signal-to-noise ratio LR SM data.

[0072] S103. Convert each row of the high signal-to-noise ratio LR SM data into a three-dimensional structure to obtain multiple three-dimensional LR SM data.

[0073] S104. Take the L layer and L+1 layer of each 3D LR SM data as the first input data and input them into the pre-trained system matrix calibration network to perform system matrix calibration processing on the L layer of the 3D LR SM data to obtain the calibrated HR SM data.

[0074] Where L∈{1,2,...,Z1}, Z1 represents the total number of preset two-dimensional slice layers; the cross-sampling method is to perform staggered sampling on adjacent layers of the preset two-dimensional slice layer; the pre-trained system matrix calibration network is trained together using HR SM sample data and three-dimensional LR SM training data; the three-dimensional LR SM training data is sampled from the HR SM sample data based on the cross-sampling method.

[0075] It should be noted that when restoring to layer Z1, since layer Z1+1 does not exist, layers Z1 and Z1-1 can be used as the first input data.

[0076] This invention provides a system matrix calibration method for a magnetic particle imaging device, comprising: performing staggered sampling on adjacent layers of a preset two-dimensional slice layer, and performing system matrix calibration processing on the original LR SM data obtained by staggered sampling based on a pre-trained system matrix calibration network, ultimately obtaining calibrated HR SM data. Since this method does not require traversal measurements at multiple measurement locations, it improves the system matrix calibration time of the MPI device. Furthermore, since this invention uses staggered sampling on adjacent layers in a preset two-dimensional slice layer, it avoids the problem of fixed data information and insufficient prior information caused by existing equal-interval sampling, and can obtain more prior information at the same sampling rate, thereby improving the system matrix calibration accuracy.

[0077] Optionally, the pre-trained system matrix calibration network includes: a data fusion module, a feature encoding module, and an upsampling reconstruction module; the upsampling reconstruction module includes: a first upsampling reconstruction module and a second upsampling reconstruction module;

[0078] The data fusion module, the feature encoding module, and the second upsampling reconstruction module are connected in series to form the first structure;

[0079] The first upsampling reconstruction module is connected in parallel with the first structure.

[0080] It should be noted that, in this embodiment of the invention, the data processing procedure of the feature encoding module is as follows:

[0081] (1) Perform convolution operations on the baseline image (L layer of 3D LR SM data) and the reference image (L+1 layer of 3D LR SM data) using a convolutional layer respectively to obtain the baseline image feature F1 and the reference image feature F2;

[0082] (2) Perform dot product operations between the reference layer image feature F1 and the reference layer image feature F2 and the reference layer image feature F1 respectively to obtain the affinity mappings A between F1, F2 and F1 respectively. 1,1 and A 1,2 The calculation formula is:

[0083] A 11 =F1·F1;

[0084] A 12 =F1·F2;

[0085] (3) Calculate the self-affinity feature M1 of the reference layer image:

[0086]

[0087] (4) Calculate the unique feature M2 of the reference layer image, which is independent of the baseline layer image:

[0088]

[0089] in, This indicates that corresponding elements are multiplied.

[0090] (5) Segment M1 and M2 to obtain complementary fusion feature F.

[0091] Corresponding to the above process, schematically, Figure 3 This is a schematic diagram illustrating the calibration process of the pre-trained system matrix calibration network provided in an embodiment of the present invention. Figure 3 The baseline layer system matrix image in the image corresponds to layer L of the 3D LR SM data, and the adjacent layer system matrix image corresponds to layer L+1 of the 3D LR SM data.

[0092] In this embodiment of the invention, the upsampling reconstruction module consists of two parts: a pixel shuffle upsampling module (the second upsampling reconstruction module) and an interpolation upsampling module (the first upsampling reconstruction module). The pixel shuffle upsampling module comprises a pixel shuffle layer and two 3x3 convolutional blocks, while the interpolation upsampling module comprises a linear interpolation layer and two 3x3 convolutional blocks. The calculation process of the upsampling reconstruction module is as follows:

[0093] (1) The global feature representation output by the feature encoding module is fed into the pixelshuffle upsampling module for operation to obtain a high-resolution feature map I1;

[0094] (2) The input baseline image is fed into the interpolation upsampling module to obtain a high-resolution feature map I2;

[0095] (3) Add the high-resolution feature map I1 and the high-resolution feature map I2 together to output the calibrated HR SM data.

[0096] Optionally, the training process of the pre-trained system matrix calibration network includes:

[0097] Obtain HR SM sample data;

[0098] Frequency filtering was performed on the HR SM sample data to obtain high signal-to-noise ratio HR SM sample data;

[0099] For each row of the high signal-to-noise ratio HR SM sample data, a 3D transformation process is performed to obtain multiple 3D HR SM initial data;

[0100] Multiple 3D HRSM initial data are sampled and processed using the cross-sampling method to obtain multiple 3D LRSM training data;

[0101] The L-th layer, L+1-th layer, and high signal-to-noise ratio (SNR) HRSM sample data corresponding to the L-th layer of each 3D LRSM training data are input into the initial system matrix calibration network, and the initial system matrix calibration network is trained to obtain the pre-trained system matrix calibration network; where L∈{1,2,...,Z2}, and Z2 represents the total number of layers of the 3D LRSM training data; the L-th and L+1-th layers of the 3D LRSM sample data are the second input data, and the high SNR HRSM sample data corresponding to the L-th layer of the 3D LRSM training data are the label data; the initial system matrix calibration network and the pre-trained system matrix calibration network have the same structure.

[0102] In this embodiment of the invention, the HR SM sample data obtained from full sampling measurements is first subjected to frequency filtering to obtain high signal-to-noise ratio (SNR) HR SM sample data; each row of the filtered high SNR HR SM sample data is converted into a three-dimensional structure and sliced ​​along the z-axis direction, according to the cross-sampling method (see...). Figure 2 The sampling location points of each slice layer are calculated, and the data at the corresponding points are extracted to obtain the 3D LRSM training data. Specifically, Figure 2 This is a schematic diagram of 2x cross-sampling provided in an embodiment of the present invention, as shown below. Figure 2 As shown, black grids indicate that the location point is selected, and blank grids indicate that the location point is not selected. Furthermore, it should be noted that the 2x cross-sampling method provided in this embodiment is merely exemplary; it can be adjusted according to device performance and actual needs. For example, it can also be set to 2... T times, where T is a positive integer.

[0103] Optionally, the L-th layer of each 3D LRSM training data, the L+1-th layer of the 3D LRSM training data, and the high signal-to-noise ratio (SNR) HRSM sample data corresponding to the L-th layer of the 3D LRSM training data are input into the initial system matrix calibration network, and the initial system matrix calibration network is trained to obtain a pre-trained system matrix calibration network, including:

[0104] Input the L-th layer of each 3D LRSM training data, the L+1-th layer of the 3D LRSM training data, and the high signal-to-noise ratio HRSM sample data corresponding to the L-th layer of the 3D LRSM training data into the initial system matrix calibration network;

[0105] The initial system matrix calibration network is continuously trained in the direction of decreasing loss function;

[0106] When the loss function meets a preset loss threshold or the number of iterations meets a preset iteration threshold, the corresponding initial system matrix calibration network is used as the pre-trained system matrix calibration network.

[0107] Optionally, the feature encoding module includes: a convolutional module and a Transformer module connected in sequence;

[0108] The Transformer module includes N Transformer encoders connected in series.

[0109] It should be noted that, in this embodiment of the invention, the Transformer module is used to extract the global feature representation of the input data. The Transformer module can be exemplarily composed of six Transformer encoders connected in series. Each Transformer encoder consists of a Multi-head Self-Attention (MHSA) module and a Feedforward Network (FFN) module. The MHSA module uses multiple self-attention layers to compute the relevant information between every two embedding block sequences. Each MHSA module is followed by an FFN module to map the data from low dimension to high dimension and then back to low dimension to improve the feature representation. The FFN module consists of two fully connected layers and a GeLU activation function.

[0110] To verify the effectiveness of the matrix calibration method for a magnetic particle imaging device system provided in this embodiment of the invention, simulation verification was also performed. Figure 4 This is a schematic diagram of the calibrated system matrix image generated based on the method of the present invention, provided as an embodiment of the invention. Figure 4 As shown, the first row of data is the low-resolution system matrix image obtained by sampling processing based on the cross-sampling method; the second row of data is the corresponding original system matrix image (used for comparison with the generated calibrated system matrix image schematic); and the third row of data is the calibrated system matrix image. Figure 4 It can be seen that the calibrated system matrix image obtained by the method of the present invention is consistent with the original system matrix image in appearance, which proves the effectiveness of the method of the embodiment of the present invention.

[0111] In summary, this invention provides a system matrix calibration method for magnetic particle imaging devices. It employs cross-downsampling of adjacent layers in a 3DFOV to acquire information from two adjacent layers as prior information, thus solving the problem of low system matrix recovery quality caused by insufficient input information. Furthermore, a pre-trained system matrix calibration network is provided, comprising: an information fusion module, a CNN-Transformer concatenated feature encoding module, and an upsampling reconstruction module. The information fusion module effectively fuses complementary features from multiple LR SMs; the CNN-Transformer concatenated feature encoding module extracts local and global feature representations of the fused features; and the upsampling reconstruction module reconstructs the HR SM from the acquired feature representations. Combining these modules facilitates the fusion of complementary features from multiple LR SMs, improving the system matrix recovery quality.

[0112] The method provided in this embodiment of the invention can be applied to electronic devices. Specifically, the electronic device can be a desktop computer, a portable computer, a smart mobile terminal, a server, etc., and this embodiment of the invention does not limit the application to such devices.

[0113] Based on the same inventive concept, embodiments of the present invention also provide a matrix calibration device for a magnetic particle imaging equipment system. Figure 5 This is a schematic diagram of a matrix calibration device for a magnetic particle imaging equipment system provided in an embodiment of the present invention. Figure 5 As shown, it includes: a sampling unit 501, a filtering unit 502, a conversion unit 503, and an input / output unit 504;

[0114] The sampling unit 501 is used to: perform sampling processing on a preset two-dimensional slice layer based on the cross-sampling method to obtain the original LR SM data; the preset two-dimensional slice layer is each two-dimensional slice layer within the three-dimensional FOV of the MPI device;

[0115] The filtering unit 502 is used to: perform frequency filtering on the raw LR SM data to obtain high signal-to-noise ratio LR SM data;

[0116] The conversion unit 503 is used to convert each row of the high signal-to-noise ratio LR SM data into a three-dimensional structure to obtain multiple three-dimensional LR SM data.

[0117] The input / output unit 504 is used to: take the L layer and L+1 layer of each three-dimensional LR SM data as the first input data and input them into the pre-trained system matrix calibration network to perform system matrix calibration processing on the L layer of the three-dimensional LR SM data to obtain the calibrated HR SM data; where L∈{1,2,...,Z1}, Z1 represents the total number of preset two-dimensional slice layers; the cross-sampling method is to perform staggered sampling on the adjacent layers of the preset two-dimensional slice layer.

[0118] Figure 6 This is a schematic diagram of a magnetic particle imaging device system matrix calibration device provided in an embodiment of the present invention. It includes a processor 710, a storage medium 720, and a bus 730. The storage medium 720 stores machine-readable instructions executable by the processor 710. When the magnetic particle imaging device system matrix calibration device is running, the processor 710 communicates with the storage medium 720 via the bus 730. The processor 710 executes the machine-readable instructions to perform the steps of the above-described method embodiment. Specific implementation methods and technical effects are similar and will not be described in detail here.

[0119] The storage medium may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the storage medium may also be at least one storage device located remotely from the aforementioned processor.

[0120] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0121] It should be noted that the terms "first," "second," etc., are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention.

[0122] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, those skilled in the art can combine and integrate the different embodiments or examples described in this specification.

[0123] Although the invention has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings and the disclosure, will understand and implement other variations of the disclosed embodiments in carrying out the claimed invention. In the description of the invention, the word "comprising" does not exclude other components or steps, "a" or "an" does not exclude a plurality, and "a plurality" means two or more, unless otherwise explicitly specified. Furthermore, while different embodiments may describe certain measures, this does not mean that these measures cannot be combined to produce good results.

[0124] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the inventive concept, and all such modifications and substitutions should be considered within the scope of protection of the present invention.

Claims

1. A matrix calibration method for a magnetic particle imaging device system, characterized in that, include: Sampling is performed on a pre-defined two-dimensional slice layer using a cross-sampling method to obtain the original LRSM data; The preset two-dimensional slice layer is each two-dimensional slice layer within the three-dimensional FOV of the MPI device; Frequency filtering is performed on the original LR SM data to obtain high signal-to-noise ratio LR SM data; Each row of the high signal-to-noise ratio LR SM data is converted into a three-dimensional structure to obtain multiple three-dimensional LR SM data. The L-layer and L+1-layer of each of the three-dimensional LR SM data are used as the first input data and input into the pre-trained system matrix calibration network to perform system matrix calibration processing on the L-layer of the three-dimensional LR SM data, thereby obtaining the calibrated HR SM data; wherein, , This represents the total number of preset two-dimensional slice layers; the cross-sampling method is a staggered sampling of adjacent layers of the preset two-dimensional slice layer; the pre-trained system matrix calibration network is trained using HR SM sample data and three-dimensional LR SM training data; the three-dimensional LR SM training data is sampled from the HR SM sample data based on the cross-sampling method; The cross-sampling method is expressed as follows: When the number of preset two-dimensional slice layers L is odd: ； When the number of preset two-dimensional slice layers L is even: ； in, Indicates the sampling location point. This indicates the x-coordinate index of the preset two-dimensional slice layer. This indicates the ordinate position index of the preset two-dimensional slice layer; The pre-trained system matrix calibration network includes: a data fusion module, a feature encoding module, and an upsampling reconstruction module; the upsampling reconstruction module includes: a first upsampling reconstruction module and a second upsampling reconstruction module. The data fusion module, the feature encoding module, and the second upsampling reconstruction module are connected in series to form the first structure; The first upsampling reconstruction module is connected in parallel with the first structure; The data fusion module is used to fuse features from multiple 3D LRSM data to obtain fused features; The feature encoding module is used to extract the local feature representation and global feature representation of the fused features; The upsampling reconstruction module is used to reconstruct the HR SM based on the local feature representation and the global feature representation.

2. The matrix calibration method for a magnetic particle imaging device system according to claim 1, characterized in that, The sampling process performed on the preset two-dimensional slice layer based on the cross-downsampling method to obtain the original LR SM data includes: Based on the cross-sampling method, corresponding sampling points are obtained in each of the preset two-dimensional slice layers; The preset two-dimensional slice layer is sampled according to the sampling points to obtain sub-LR SM data; the sub-LR SM data is the LR SM sampling data of each two-dimensional slice layer; The sub-LR SM data is combined into a matrix to obtain the original LR SM data.

3. The matrix calibration method for a magnetic particle imaging device system according to claim 1, characterized in that, The step of frequency filtering the original LR SM data to obtain high signal-to-noise ratio LR SM data includes: Perform a Fourier transform on the original LR SM data to obtain frequency domain LR SM data; The high signal-to-noise ratio LR SM data is formed by taking the data in the frequency domain LR SM data that are greater than a preset frequency threshold.

4. The matrix calibration method for a magnetic particle imaging device system according to claim 1, characterized in that, The training process of the pre-trained system matrix calibration network includes: Obtain HR SM sample data; Frequency filtering was performed on the HR SM sample data to obtain high signal-to-noise ratio HR SM sample data; For each row of the high signal-to-noise ratio HR SM sample data, a three-dimensional transformation process is performed to obtain multiple three-dimensional HR SM initial data; The initial data of the multiple three-dimensional HRSMs are sampled and processed using the cross-sampling method to obtain multiple three-dimensional LRSM training data. The L-th layer of each of the three-dimensional LR SM training data, the L+1-th layer of the three-dimensional LR SM training data, and the high signal-to-noise ratio (SNR) HR SM sample data corresponding to the L-th layer of the three-dimensional LR SM training data are input into the initial system matrix calibration network, and the initial system matrix calibration network is trained to obtain the pre-trained system matrix calibration network; wherein, , The total number of layers in the 3D LRSM training data is represented; wherein, the L layer and the L+1 layer of the 3D LRSM sample data are the second input data, and the high signal-to-noise ratio HRSM sample data corresponding to the Lth layer of the 3D LRSM training data are the label data; the initial system matrix calibration network has the same structure as the pre-trained system matrix calibration network.

5. The matrix calibration method for a magnetic particle imaging device system according to claim 1, characterized in that, The step of inputting the L-th layer of each of the three-dimensional LRSM training data, the L+1-th layer of the three-dimensional LRSM training data, and the high signal-to-noise ratio (SNR) HRSM sample data corresponding to the L-th layer of the three-dimensional LRSM training data into the initial system matrix calibration network, and training the initial system matrix calibration network to obtain the pre-trained system matrix calibration network includes: The L-th layer of each of the three-dimensional LR SM training data, the L+1-th layer of the three-dimensional LR SM training data, and the high signal-to-noise ratio HR SM sample data corresponding to the L-th layer of the three-dimensional LR SM training data are input into the initial system matrix calibration network; The initial system matrix calibration network is continuously trained in the direction of decreasing loss function; When the loss function satisfies a preset loss threshold or the number of iterations satisfies a preset iteration threshold, the corresponding initial system matrix calibration network is used as the pre-trained system matrix calibration network.

6. The matrix calibration method for a magnetic particle imaging device system according to claim 1, characterized in that, The feature encoding module includes: a convolutional module and a Transformer module connected in sequence; The Transformer module includes N Transformer encoders connected in series.

7. A matrix calibration device for a magnetic particle imaging equipment system, characterized in that, The magnetic particle imaging equipment system matrix calibration device includes: a sampling unit, a screening unit, a conversion unit, and an input / output unit; The sampling unit is used to: perform sampling processing based on the cross-sampling method in a preset two-dimensional slice layer to obtain the original LR SM data; the preset two-dimensional slice layer is each two-dimensional slice layer within the three-dimensional FOV of the MPI device; The filtering unit is used to: perform frequency filtering on the original LR SM data to obtain high signal-to-noise ratio LR SM data; The conversion unit is used to: convert each row of the high signal-to-noise ratio LR SM data into a three-dimensional structure to obtain multiple three-dimensional LR SM data; The input / output unit is used to: input the L layer and L+1 layer of each of the three-dimensional LR SM data as first input data into the pre-trained system matrix calibration network, so as to perform system matrix calibration processing on the L layer of the three-dimensional LR SM data to obtain calibrated HR SM data; wherein, , This represents the total number of preset two-dimensional slice layers; the cross-sampling method involves misaligned sampling of adjacent layers of the preset two-dimensional slice layer; the pre-trained system matrix calibration network is trained using HR SM sample data and three-dimensional LR SM training data; the three-dimensional LR SM training data is obtained by sampling from the HR SM sample data based on the cross-sampling method. The cross-sampling method is expressed as follows: When the number of preset two-dimensional slice layers L is odd: ； When the number of preset two-dimensional slice layers L is even: ； in, Indicates the sampling location point. This indicates the x-coordinate index of the preset two-dimensional slice layer. This indicates the ordinate position index of the preset two-dimensional slice layer; The pre-trained system matrix calibration network includes: a data fusion module, a feature encoding module, and an upsampling reconstruction module; the upsampling reconstruction module includes: a first upsampling reconstruction module and a second upsampling reconstruction module. The data fusion module, the feature encoding module, and the second upsampling reconstruction module are connected in series to form the first structure; The first upsampling reconstruction module is connected in parallel with the first structure; The data fusion module is used to fuse features from multiple 3D LRSM data to obtain fused features; The feature encoding module is used to extract the local feature representation and global feature representation of the fused features; The upsampling reconstruction module is used to reconstruct the HR SM based on the local feature representation and the global feature representation.

8. A matrix calibration device for a magnetic particle imaging equipment system, characterized in that, include: The system includes a processor, a storage medium, and a bus. The storage medium stores machine-readable instructions executable by the processor. When the magnetic particle imaging device system matrix calibration is ready to run, the processor communicates with the storage medium via the bus, and the processor executes the machine-readable instructions to perform the steps of the magnetic particle imaging device system matrix calibration method as described in any one of claims 1-6.

Images