MPI image reconstruction method based on Transformer

Abstract

The present invention belongs to the field of biomedical molecular imaging, and specifically relates to a Transformer-based MPI image reconstruction method, system, and device, which aims to solve the problems of low reconstruction accuracy and poor robustness of the existing traditional MPI mathematical model. The method comprises: constructing a biological lesion model of an organism whose MPI image is to be reconstructed, and injecting a targeting molecule encapsulating SPIOs into the biological lesion model; after the injection, using an MPI device to collect signals from the biological lesion model to obtain a response voltage signal corresponding to the targeting molecule of SPIOs; based on the response voltage signal, reconstructing the distribution of SPIOs probes in the organism through a trained Transformer neural network, and then obtaining a reconstructed MPI image. The present invention improves the accuracy and robustness of MPI reconstruction.

Description

Technical Field

[0001] The present invention belongs to the field of biomedical molecular imaging, and specifically relates to a Transformer-based MPI image reconstruction method, system, and device. Background Art

[0002] In recent years, as a new type of molecular imaging, magnetic particle imaging (MPI) has developed rapidly and quickly become an international frontier research hotspot. The basic principle of MPI imaging is to use the nonlinear response remagnetization of superparamagnetic iron oxide particles (SPIOs) in the non-magnetic field area in a high gradient magnetic field to quantitatively obtain the three-dimensional concentration distribution of SPIOs in the body. The intensity of the response voltage signal collected by MPI is proportional to the concentration of SPIOs. Through modeling analysis, the concentration distribution of SPIOs in the body can be inferred from the collected voltage signal. Since MPI only performs specific imaging of SPIOs, it is less affected by background noise and has the characteristics of high sensitivity and high spatiotemporal resolution. It has been successfully applied in cell tracing, cardiovascular and cerebrovascular imaging, and neural imaging.

[0003] At present, the MPI image reconstruction methods mainly include two categories: solving the inverse problem based on the system matrix and X-space mapping. The former mainly uses different regularization strategies to alleviate the model pathology. Among them, Tikhonov regularization is simple and fast, and it is the most widely used. However, the reconstruction method based on the system matrix requires a lot of prior information and a complex solution process to obtain accurate reconstruction results. The reconstruction parameters involved are numerous and the image reconstruction is poor in robustness. Goodwill and Conolly of the University of California, Berkeley proposed the X-Space image reconstruction algorithm in 2010. Different from the system matrix reconstruction method, the X-space-based reconstruction algorithm directly obtains the reconstructed image by performing velocity compensation, spatial coordinate mapping and gridding on the time domain response signal of SPIOs. However, the X-space-based reconstruction method is greatly affected by the particle relaxation effect, which easily leads to artifacts in the reconstructed image. Therefore, in order to further expand the application of MPI imaging in biomedical research, it is necessary to study an accurate and robust MPI imaging technology. Summary of the invention

[0004] In order to solve the above problems in the prior art, that is, to solve the problems of low reconstruction accuracy and poor robustness of the traditional MPI mathematical model, the first aspect of the present invention proposes a Transformer-based MPI image reconstruction method, which includes the following steps:

[0005] S100, constructing a biological lesion model of a living body whose MPI image is to be reconstructed, and injecting the targeting molecules encapsulating SPIOs into the biological lesion model;

[0006] S200, after injection, using an MPI device to collect signals from the biological lesion model to obtain a response voltage signal corresponding to the targeting molecule of SPIOs;

[0007] S300, based on the response voltage signal, reconstructing the distribution of SPIOs probes in the organism through a trained Transformer neural network, thereby obtaining a reconstructed MPI image;

[0008] The Transformer neural network is constructed based on a time domain encoder and a double-layer unidirectional Transformer encoder.

[0009] In some preferred embodiments, the time domain encoder is used to divide the response voltage signal into n sequence segments of length L, and then encode it into a D-dimensional vector representation through one-dimensional convolution.

[0010] In some preferred embodiments, the training method of the Transformer neural network is:

[0011] A100, constructing a training set; the training samples in the training set include a response voltage signal and its corresponding true value MPI image; the response voltage signal is a time domain voltage signal;

[0012] A200 divides the time domain voltage signal into n sequence segments of length L, and then encodes it into a D-dimensional vector representation through one-dimensional convolution as the first vector;

[0013] A300, add a flag bit CLS before the first vector;

[0014] A400, the first vector with the added flag is subjected to feature extraction through a two-layer unidirectional Transformer encoder, and the output corresponding to the flag CLS is regarded as the feature vector obtained after the entire signal is encoded;

[0015] A500, the feature vector extracted by the two-layer unidirectional Transformer encoder is input into the fully connected layer to obtain the reconstructed MPI image;

[0016] A600, based on the reconstructed MPI image and the true MPI image, calculates the loss value through the pre-built loss function and updates the network parameters of the Transformer neural network;

[0017] A700, loop A100-A600 until the trained Transformer neural network is obtained.

[0018] In some preferred embodiments, the response voltage signal in the training data is obtained by:

[0019] The digital mouse was used as a standard imaging space, in which the distribution of SPIOs of different positions, types, and concentrations was set;

[0020] The Debye simulation algorithm is used to accurately simulate the SPIOs distribution and obtain its corresponding response voltage signal.

[0021] In some preferred embodiments, the loss function of the Transformer neural network during training is:

[0022]

[0023] Among them, y and Represent the true MPI image and the reconstructed MPI image, μ y and Represents y and The average value, δ y and Represents y and The standard deviation of Represents y and The covariance of , c1 and c2 are constants.

[0024] The second aspect of the present invention proposes a Transformer-based MPI image reconstruction system, which includes: a lesion model building module, a signal acquisition module, and an image reconstruction module;

[0025] The lesion model construction module is configured to construct a biological lesion model of the organism whose MPI image is to be reconstructed, and inject the targeting molecules encapsulating SPIOs into the biological lesion model;

[0026] The signal acquisition module is configured to collect signals from the biological lesion model using an MPI device after injection to obtain a response voltage signal corresponding to the targeting molecule of SPIOs;

[0027] The image reconstruction module is configured to reconstruct the distribution of SPIOs probes in the organism through a trained Transformer neural network based on the response voltage signal, thereby obtaining a reconstructed MPI image;

[0028] The Transformer neural network is constructed based on a time domain encoder and a double-layer unidirectional Transformer encoder.

[0029] The third aspect of the present invention proposes a device, at least one processor; and a memory communicatively connected to at least one of the processors; wherein the memory stores instructions executable by the processor, and the instructions are used to be executed by the processor to implement the above-mentioned Transformer-based MPI image reconstruction method.

[0030] In a fourth aspect of the present invention, a computer-readable storage medium is proposed, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to be executed by the computer to implement the above-mentioned Transformer-based MPI image reconstruction method.

[0031] Beneficial effects of the present invention:

[0032] The present invention improves the accuracy and robustness of MPI reconstruction.

[0033] 1) Based on data-driven machine learning theory, the present invention uses a neural network to directly learn the remagnetization response process of SPIOs in a magnetic field-free region, avoiding the model error caused by constructing a magnetization response model and solving a mathematical model, and improving the image reconstruction accuracy and robustness of MPI;

[0034] 2) The present invention constructs a standard imaging space based on the digital mouse and a simulation algorithm based on the Debye model, thus overcoming the limitations of the number and simulation accuracy of the training data set used in the MPI deep learning algorithm;

[0035] 3) The present invention overcomes the difficulty of phase estimation in processing frequency domain signals by the traditional MPI reconstruction algorithm and directly processes time domain sequence signals;

[0036] 4) The attention mechanism in the Transformer encoder used in the present invention has the characteristic of ignoring the internal distance of the sequence, can learn the internal structure of the sequence signal and model its long-term dependency, and the multi-head attention mechanism can extract multi-dimensional features, further enhancing the accuracy of the reconstruction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings.

[0038] Figure 1 It is a flowchart of a Transformer-based MPI image reconstruction method according to an embodiment of the present invention;

[0039] Figure 2 It is a schematic diagram of the framework of a Transformer-based MPI image reconstruction system according to an embodiment of the present invention;

[0040] Figure 3 It is a flowchart of the training process and application process of the Transformer neural network of an embodiment of the present invention;

[0041] Figure 4 is a detailed schematic diagram of a Transformer-based MPI image reconstruction method according to an embodiment of the present invention;

[0042] Figure 5 is a schematic diagram of the structure of a Transformer encoder according to an embodiment of the present invention;

[0043] Figure 6 It is a structural diagram of a computer system of an embodiment of the present invention that is suitable for implementing an electronic device of an embodiment of the present application. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution in the embodiment of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] The present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It is also necessary to explain that, for ease of description, only the parts related to the relevant invention are shown in the accompanying drawings.

[0046] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application may be combined with each other.

[0047] The present invention provides a Transformer-based MPI image reconstruction method, such as Figure 1 As shown, the method comprises the following steps:

[0048] S100, constructing a biological lesion model of a living body whose MPI image is to be reconstructed, and injecting the targeting molecules encapsulating SPIOs into the biological lesion model;

[0049] S200, after injection, using an MPI device to collect signals from the biological lesion model to obtain a response voltage signal corresponding to the targeting molecule of SPIOs;

[0050] S300, based on the response voltage signal, reconstructing the distribution of SPIOs probes in the organism through a trained Transformer neural network, thereby obtaining a reconstructed MPI image;

[0051] The Transformer neural network is constructed based on a time domain encoder and a double-layer unidirectional Transformer encoder.

[0052] In order to more clearly illustrate the Transformer-based MPI image reconstruction method of the present invention, each step in an embodiment of the method of the present invention is described in detail below with reference to the accompanying drawings.

[0053] In the following embodiments, the training process of the Transformer neural network is first described, and then the process of obtaining the reconstructed MPI image by the Transformer-based MPI image reconstruction method is described in detail. Figure 3 As shown, the details are as follows:

[0054] 1. Training process of Transformer neural network

[0055] A100, constructing a training set; the training samples in the training set include a response voltage signal and its corresponding true value MPI image; the response voltage signal is a time domain voltage signal;

[0056] In this embodiment, a digital mouse is used as a standard imaging space, in which SPIOs of different positions, types and concentrations are set; the SPIOs distribution is simulated using the Debye simulation algorithm to obtain its corresponding response voltage signal.

[0057] The actual SPIOs distribution inside the digital mouse (i.e., the true MPI image) and the corresponding response voltage information are collected to construct a large number of training samples and a training set to train the Transformer neural network.

[0058] In the present invention, the Transformer neural network includes two parts: a time domain encoder and a Transformer encoder. The input of the Transformer neural network is a time domain voltage signal, and the output of the network is an MPI reconstructed image. The specific structure and function of the time domain encoder and the Transformer encoder are described in detail in the following content.

[0059] A200 divides the time domain voltage signal into n sequence segments of length L, and then encodes it into a D-dimensional vector representation through one-dimensional convolution as the first vector;

[0060] In this embodiment, the mixed time domain signal is converted into a corresponding representation in the feature space by a time domain encoder. That is, the time domain voltage signal is divided into n sequence segments of length L, and then encoded into a D-dimensional vector representation by one-dimensional convolution. like Figure 4 As shown, the D-dimensional vector represents x i as the first vector.

[0061] A300, add a flag bit CLS before the first vector;

[0062] A400, the first vector with the added flag is subjected to feature extraction through a two-layer unidirectional Transformer encoder, and the output corresponding to the flag CLS is regarded as the feature vector obtained after the entire signal is encoded;

[0063] A500, the feature vector extracted by the two-layer unidirectional Transformer encoder is input into the fully connected layer to obtain the reconstructed MPI image;

[0064] In this embodiment, the Transformer encoder is responsible for feature extraction and analysis of time series signals, converting the input sequence into a dense vector of fixed dimension, and subsequently generating task targets through the dense vector, such as Figure 5 As shown, specifically:

[0065] The output (i.e., the first vector) obtained by the time domain encoder is preceded by a flag bit CLS as a representation of the entire vector representation, and then the vector representation of the signal is concatenated with its position encoding to obtain the input of the Transformer encoder; secondly, the input is subjected to multi-dimensional feature extraction through a multi-head attention layer. The multi-head attention can be regarded as randomly initializing the input at each position and assigning h self-attention layers to calculate the implicit information in the signal sequence from multiple dimensions, that is, for a specific x i For example, we need to use multiple sets of randomly initialized matrices W Q , W K , W V Multiply it with the signal, and finally concatenate the obtained results as the output of the attention layer. In addition, the residual connection module is used to increase the network depth, and layer normalization is used to ensure the stability of training when the input dimension scale is different. The feedforward layer consists of two linear layers and a ReLu activation function. After two layers of Transformer encoders, the output corresponding to the flag CLS is regarded as the feature vector obtained after the entire signal is encoded, and then the feature vector is used as the input of the fully connected layer to reconstruct the MPI image.

[0066] in addition, Figure 4 The T in here also refers to the Transformer encoder module, and Z iRefers to the feature vector after Transformer encoding.

[0067] A600, based on the reconstructed MPI image and the true MPI image, calculates the loss value through the pre-built loss function and updates the network parameters of the Transformer neural network;

[0068] In this embodiment, based on the reconstructed MPI image obtained in the above steps and combined with the true MPI image, the SSIM (structural similarity) standard is used as the loss function to calculate the loss value and update the network parameters of the Transformer neural network, as shown in formula (1):

[0069]

[0070] Among them, y and Represent the true MPI image and the reconstructed MPI image, μ y and Represents y and The average value, δ y and Represents y and The standard deviation of Represents y and The covariance of , c1, c2 are constants to avoid systematic errors with denominators of 0.

[0071] The present invention adopts the SSIM standard as the loss function because it takes brightness, contrast and structure indicators into consideration and is close to human visual perception.

[0072] A700, loop A100-A600 until the trained Transformer neural network is obtained.

[0073] In this embodiment, the model is trained in a loop until a trained Transformer neural network is obtained.

[0074] 2. Transformer-based MPI image reconstruction method

[0075] S100, constructing a biological lesion model of a living body whose MPI image is to be reconstructed, and injecting the targeting molecules encapsulating SPIOs into the biological lesion model;

[0076] S200, after injection, using an MPI device to collect signals from the biological lesion model to obtain a response voltage signal corresponding to the targeting molecule of SPIOs;

[0077] S300, based on the response voltage signal, reconstructing the distribution of SPIOs probes in the organism through a trained Transformer neural network, thereby obtaining a reconstructed MPI image.

[0078] In this embodiment, a biological lesion model of an organism is first constructed, and then the targeting molecules wrapped in SPIOs are injected. After the injection, signal acquisition is performed based on the MPI device to obtain a response voltage signal; the MPI device in the present invention preferably uses a small animal three-dimensional MPI imaging device based on magnetic field-free lines or a small animal three-dimensional MPI imaging device based on magnetic field-free points. In other embodiments, it can be selected according to actual needs.

[0079] The response voltage signal is input into the trained Transformer neural network to reconstruct the distribution of SPIOs probes in the organism and obtain the reconstructed MPI image.

[0080] The present invention is based on a data-driven neural network method to directly learn the remagnetization response process of SPIOs in a magnetic field-free region, thereby achieving accurate and robust MPI imaging.

[0081] A Transformer-based MPI image reconstruction system according to a second embodiment of the present invention is as follows: Figure 2 As shown, it specifically includes the following modules: a lesion model building module 100, a signal acquisition module 200, and an image reconstruction module 300;

[0082] The lesion model construction module 100 is configured to construct a biological lesion model of a living body whose MPI image is to be reconstructed, and inject the targeting molecules encapsulating SPIOs into the biological lesion model;

[0083] The signal acquisition module 200 is configured to collect signals from the biological lesion model using an MPI device after injection to obtain a response voltage signal corresponding to the targeting molecule of SPIOs;

[0084] The image reconstruction module 300 is configured to reconstruct the distribution of SPIOs probes in the organism through a trained Transformer neural network based on the response voltage signal, thereby obtaining a reconstructed MPI image;

[0085] The Transformer neural network is constructed based on a time domain encoder and a double-layer unidirectional Transformer encoder.

[0086] Those skilled in the art can 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.

[0087] It should be noted that the Transformer-based MPI image reconstruction system provided in the above embodiment is only illustrated by the division of the above functional modules. In practical 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 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 only for distinguishing the modules or steps, and are not regarded as improper limitations of the present invention.

[0088] A device according to the third embodiment of the present invention comprises at least one processor; and a memory communicatively connected to at least one of the processors; wherein the memory stores instructions executable by the processor, and the instructions are used to be executed by the processor to implement the above-mentioned Transformer-based MPI image reconstruction method.

[0089] A fourth embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to be executed by the computer to implement the above-mentioned Transformer-based MPI image reconstruction method.

[0090] Those skilled in the art can clearly understand that for the convenience and simplicity not described, the specific working process and related instructions of the above-described device and computer-readable storage medium can refer to the corresponding process in the aforementioned method example and will not be repeated here.

[0091] Reference below Figure 6 , which shows a schematic diagram of the structure of a computer system of a server suitable for implementing the system, method, and device embodiments of the present application. Figure 6 The server shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0092] like Figure 6As shown, the computer system includes a central processing unit (CPU) 601, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 602 or the program loaded from the storage part 608 to the random access memory (RAM) 603. In the RAM 603, various programs and data required for system operation are also stored. The CPU 601, ROM 602 and RAM 603 are connected to each other through a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0093] The following components are connected to the I / O interface 605: an input section 606 including a keyboard, a mouse, etc.; an output section 607 including a cathode ray tube, a liquid crystal display, etc., and a speaker, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a local area network card, a modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the I / O interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 610 as needed, so that a computer program read therefrom is installed into the storage section 608 as needed.

[0094] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 609, and / or installed from the removable medium 611. When the computer program is executed by the CPU 601, the above-mentioned functions defined in the method of the present application are executed. It should be noted that the above-mentioned computer-readable medium of the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can be, for example, but not limited to: a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection with one or more conductors, a portable computer disk, a hard disk, a RAM, a ROM, an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, an apparatus, or a device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in combination with an instruction execution system, an apparatus, or a device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, etc., or any suitable combination of the above.

[0095] The computer program code for performing the operation of the present application can 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 language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network or a wide area network, or can be connected to an external computer (e.g., using an Internet service provider to connect through the Internet).

[0096] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the box can also occur in a sequence different from 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 with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0097] The terms "first", "second", etc. are used to distinguish similar objects rather than to describe or indicate a particular order or sequence.

[0098] The term "comprise" or any other similar term is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus / device that includes a list of elements includes not only those elements but also other elements not expressly listed, or also includes elements inherent to such process, method, article, or apparatus / device.

[0099] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims

1. A Transformer-based MPI image reconstruction method, characterized in that: The method comprises the following steps: S100, constructing a biological lesion model of a living body whose MPI image is to be reconstructed, and injecting the targeting molecules encapsulating SPIOs into the biological lesion model; S200, after injection, using an MPI device to collect signals from the biological lesion model to obtain a response voltage signal corresponding to the targeting molecule of SPIOs; S300, based on the response voltage signal, reconstructing the distribution of SPIOs probes in the organism through a trained Transformer neural network, thereby obtaining a reconstructed MPI image; The Transformer neural network is constructed based on a time domain encoder and a double-layer unidirectional Transformer encoder; the time domain encoder is used to divide the response voltage signal into m sequence segments of length L, and then encode it into a D-dimensional vector representation through one-dimensional convolution i=1, 2, ...n.

2. The Transformer-based MPI image reconstruction method according to claim 1, characterized in that: The training method of the Transformer neural network is as follows: A100, constructing a training set; the training samples in the training set include a response voltage signal and its corresponding true value MPI image; the response voltage signal is a time domain voltage signal; A200 divides the time domain voltage signal into m sequence segments of length L, and then encodes it into a D-dimensional vector representation through one-dimensional convolution i=1, 2, ...n, as the first vector; A300, add a flag bit CLS before the first vector; A400, the first vector with the added flag is subjected to feature extraction through a two-layer unidirectional Transformer encoder, and the output corresponding to the flag CLS is regarded as the feature vector obtained after the entire signal is encoded; A500, the feature vector extracted by the two-layer unidirectional Transformer encoder is input into the fully connected layer to obtain the reconstructed MPI image; A600, based on the reconstructed MPI image and the true MPI image, calculates the loss value through the pre-built loss function and updates the network parameters of the Transformer neural network; A700, loop A100-A600 until the trained Transformer neural network is obtained.

3. The Transformer-based MPI image reconstruction method according to claim 2, characterized in that: The response voltage signal in the training data is obtained by: The digital mouse was used as a standard imaging space, in which the distribution of SPIOs of different positions, types, and concentrations was set; The Debye simulation algorithm is used to simulate the SPIOs distribution and obtain its corresponding response voltage signal.

4. The Transformer-based MPI image reconstruction method according to claim 2, characterized in that: The loss function of the Transformer neural network during training is: Among them, y and Represent the true MPI image and the reconstructed MPI image, μ y and Represents y and The average value, δ y and Represents y and The standard deviation of Represents y and The covariance of , c1 and c2 are constants.

5. A Transformer-based MPI image reconstruction system, characterized in that: The system includes: a lesion model building module, a signal acquisition module, and an image reconstruction module; The lesion model construction module is configured to construct a biological lesion model of the organism whose MPI image is to be reconstructed, and inject the targeting molecules encapsulating SPIOs into the biological lesion model; The signal acquisition module is configured to collect signals from the biological lesion model using an MPI device after injection to obtain a response voltage signal corresponding to the targeting molecule of SPIOs; The image reconstruction module is configured to reconstruct the distribution of SPIOs probes in the organism through a trained Transformer neural network based on the response voltage signal, thereby obtaining a reconstructed MPI image; The Transformer neural network is constructed based on a time domain encoder and a double-layer unidirectional Transformer encoder; the time domain encoder is used to divide the response voltage signal into n sequence segments of length L, and then encode it into a D-dimensional vector representation through one-dimensional convolution i=1, 2, ...n.

6. A Transformer-based MPI image reconstruction device, characterized in that: include: at least one processor; as well as a memory communicatively connected to at least one of the processors; wherein, The memory stores instructions executable by the processor, and the instructions are used to be executed by the processor to implement the Transformer-based MPI image reconstruction method according to any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to be executed by the computer to implement the Transformer-based MPI image reconstruction method according to any one of claims 1 to 4.

Images