Electrical impedance imaging method based on multilayer full-connection neural network and application

Abstract

The invention discloses an electrical impedance tomography method based on a multi-layer full-connection neural network. The method comprises the following steps: acquiring voltage measurement data of a to-be-measured area of a human thoracic cavity; performing triangle subdivision on the to-be-measured region by using a finite element method to obtain a finite element model; establishing a multi-layer full-connection neural network imaging model according to the finite element model, inputting the voltage measurement data into the neural network imaging model to obtain an initial conductivity distribution sequence of the to-be-measured region, and performing normalization to obtain a training set; training a neural network imaging model by using an early stop method according to the training set to obtain an optimal solution of a network connection weight; therefore, an optimized neural network imaging model is obtained, and a reconstructed image of impedance distribution of the to-be-measured region is obtained according to the optimized neural network imaging model. According to the method, the nonlinear mapping relation between the electrical impedance imaging input quantity and the electrical impedance imaging output quantity can be better represented, and the quality and the robustness of electrical impedance imaging are improved.

Description

Technical Field

[0001] The present application relates to the technical field of electrical impedance tomography, and in particular to an electrical impedance tomography method, device, electronic device and a computer-readable storage medium based on a multi-layer fully connected neural network. Background Art

[0002] At present, there are many technologies for monitoring lung respiratory function, but most of them have the problem of difficulty in achieving continuous imaging of ventilation during the breathing process. The spirometry can achieve dynamic images of lung breathing, but it is difficult to detect lesion information because it cannot display local respiratory conditions. Electrical impedance tomography (EIT), as a non-invasive imaging technology, can reflect the physiological activities of human organs and tissues in real time, such as lung ventilation, and belongs to functional imaging. The morphological structure of lung cancer is not obvious in the early stage, and it is difficult to be detected by traditional medical detection technology. However, the conductivity of the diseased lung tissue has changed at this time, and it can be detected and treated as early as possible through EIT technology. At present, EIT has been applied in clinical research on lung disease monitoring.

[0003] The patent with application number CN107647866A discloses an electrical impedance imaging method based on the partial least squares method, which realizes the image reconstruction of the electrical impedance. The invention adopts the partial least squares method to solve the impedance distribution change matrix of the target area, which improves the calculation accuracy of the impedance distribution change to a certain extent and reduces the dependence on the initial value of the impedance distribution. However, the traditional EIT method is mainly limited by the complexity of the physical model and the nonlinear characteristics of the electrical signal, which is inconsistent with the linearized imaging model set in the partial least squares method, resulting in low accuracy and low resolution of the imaging results in terms of contour. Summary of the invention

[0004] In order to overcome the defects of the above-mentioned prior art, an embodiment of the present invention provides an electrical impedance imaging method and application based on a multi-layer fully connected neural network, which can better characterize the nonlinear mapping relationship between the input and output of electrical impedance imaging and improve the quality and robustness of electrical impedance imaging.

[0005] On the one hand, an embodiment of the present invention proposes an electrical impedance imaging method based on a multi-layer fully connected neural network, the method comprising: obtaining voltage measurement data of a human chest area to be tested through an EIT device; triangulating the area to be tested using a finite element method to obtain a finite element model; establishing a multi-layer fully connected neural network imaging model based on the finite element model, inputting the voltage measurement data into the neural network imaging model to obtain an initial distribution sequence of the conductivity of the area to be tested, and normalizing the initial distribution sequence to obtain a training set; training the neural network imaging model using an early stopping method based on the training set to obtain an optimal solution for the network connection weights in the neural network imaging model; obtaining an optimized neural network imaging model based on the optimal solution for the network connection weights, and obtaining a reconstructed image of the impedance distribution of the area to be tested based on the optimized neural network imaging model.

[0006] In one embodiment of the present invention, the neural network imaging model is a six-layer fully connected neural network architecture, including an input layer, five hidden layers and an output layer, and the neurons in each layer are fully interconnected; wherein the first to third layers are the encoding process of input feature extraction and learning, and the third to sixth layers are the decoding process of mapping input feature values ​​to high-dimensional feature values.

[0007] In one embodiment of the present invention, each hidden layer introduces an activation function ReLU to express the nonlinear relationship between the boundary measurement voltage value and the lung respiratory impedance value; the last hidden layer also introduces an activation function tanh to perform regression tasks to ensure that the neural network imaging model outputs an impedance distribution that conforms to actual physical properties.

[0008] In one embodiment of the present invention, the voltage measurement data includes boundary voltage values ​​of the area to be measured, the number of the boundary voltage values ​​is used as the number of input layer nodes of the neural network, the number of finite element divisions corresponding to the number of lung respiratory impedance values ​​is used as the number of output layer nodes of the neural network, and the number of hidden layer nodes of the neural network is determined based on the simulation results, thereby establishing a mapping relationship between the boundary voltage value and the lung respiratory impedance value.

[0009] In one embodiment of the present invention, the connection between the input and output of the nth neuron in each layer of the neural network imaging model is: Among them, f n is the output of the nth neuron, m represents the total number of neurons in the corresponding layer, and w jn is the jth input z of the nth neuron j The connection weight between the neuron and b n is the bias of the nth neuron.

[0010] In one embodiment of the present invention, the loss function used in the neural network imaging model adopts a cross entropy function plus an L2 regularization constraint term, which is expressed as: Where N represents the total number of pixels in the reconstructed image, x i Represents the set i-th true conductivity pixel value, represents the ith conductivity pixel value predicted by the grid, λ represents the regularization parameter, and L2 represents the regularization constraint term.

[0011] In one embodiment of the present invention, the training of the neural network imaging model includes: dividing the conductivity change and the corresponding voltage value data into a training set and a validation set, obtaining test results on the validation set after each training of the neural network imaging model, and stopping the training if the test error is found to increase on the validation set, otherwise continuing the training, and using the weights obtained after the training is stopped as the final parameters of the neural network imaging model.

[0012] On the other hand, an embodiment of the present invention also proposes an electrical impedance imaging device based on a multi-layer fully connected neural network, including: a voltage measurement data acquisition module, used to obtain voltage measurement data of a human chest area to be tested through an EIT device; a finite element model construction module, used to triangulate the area to be tested using the finite element method to obtain a finite element model; a neural network imaging model establishment module, used to establish a multi-layer fully connected neural network imaging model according to the finite element model, input the voltage measurement data into the neural network imaging model to obtain an initial distribution sequence of the conductivity of the area to be tested, and normalize the initial distribution sequence to obtain a training set; a neural network imaging model training module, used to train the neural network imaging model using the early stopping method according to the training set to obtain an optimal solution for the network connection weights in the neural network imaging model; an impedance distribution image reconstruction module, used to obtain an optimized neural network imaging model according to the optimal solution of the network connection weights, and obtain a reconstructed image of the impedance distribution of the area to be tested according to the optimized neural network imaging model.

[0013] On the other hand, an embodiment of the present invention further proposes an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method described in any one of the above embodiments when executing the program.

[0014] On the other hand, an embodiment of the present invention further proposes a computer-readable storage medium, which stores computer-executable instructions, and the computer-executable instructions are used to execute the electrical impedance imaging method based on a multi-layer fully connected neural network as described in any one of the above embodiments.

[0015] It can be seen from the above that, compared with the prior art, the above embodiments of the present invention can have at least one or more of the following beneficial effects:

[0016] 1) By using the finite element method to triangulate the area to be tested, the problem to be solved is first discretized for unit analysis, and then the results of each unit analysis are combined for a comprehensive analysis of the whole, so that the finite difference equation is used to approximate the partial differential equation in the EIT problem to obtain the value of the field function at each discrete point, accurately and quickly approximate the true solution of the nonlinear problem, and establish a true and effective finite element model of the lung cross section;

[0017] 2) Using real lung electrical impedance distribution samples as supervisory signals, the parameters of the neural network model are optimized through the back-propagation algorithm to achieve accurate mapping from surface voltage measurement data to in vivo electrical impedance distribution;

[0018] 3) The multi-level abstract feature learning mechanism of the multi-layer fully connected neural network is used to effectively improve the quality and resolution of image reconstruction and reduce the impact of noise interference, thereby solving the problem of insufficient image accuracy faced by traditional EIT technology in clinical applications;

[0019] 4) The activation function ReLU is used in the neural network model to enhance the model's ability to express nonlinear relationships, so that it can capture complex patterns in the impedance distribution, enabling the model to deeply explore and express subtle differences and complex changes in the impedance distribution, and select the activation function tanh suitable for regression tasks to ensure that the model output conforms to the actual physical characteristics of the impedance distribution. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0021] Figure 1 A flowchart of an electrical impedance imaging method based on a multi-layer fully connected neural network provided by an embodiment of the present invention;

[0022] Figure 2 A schematic diagram of a finite element mesh division of a human chest cavity provided in an embodiment of the present invention;

[0023] Figure 3 A schematic diagram of the structure of a fully connected neural network regression model provided by an embodiment of the present invention;

[0024] Figure 4 A schematic structural diagram of an electrical impedance imaging device based on a multi-layer fully connected neural network provided by an embodiment of the present invention;

[0025] Figure 5 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention;

[0026] Figure 6 A schematic diagram of the structure of a computer-readable storage medium provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0027] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other. The present invention will be described below with reference to the accompanying drawings and in combination with the embodiments.

[0028] In order to enable ordinary technicians in the field to better understand the technical solution of the present invention, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments, and all should belong to the protection scope of the present invention.

[0029] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are applicable to distinguishing similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition. The terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or are proprietary to these processes, methods, products or devices.

[0030] It should also be noted that the division of multiple embodiments in the present invention is only for the convenience of description and should not constitute a special limitation. The features in various embodiments can be combined and referenced to each other without contradiction.

[0031] like Figure 1As shown, the first embodiment of the present invention proposes an electrical impedance imaging method based on a multi-layer fully connected neural network, for example, including: step S1, obtaining voltage measurement data of a human chest area to be measured through an EIT device; step S2, triangulating the area to be measured using a finite element method to obtain a finite element model; step S3, establishing a multi-layer fully connected neural network imaging model based on the finite element model, inputting the voltage measurement data into the neural network imaging model to obtain an initial distribution sequence of the conductivity of the area to be measured, and normalizing the initial distribution sequence to obtain a training set; step S4, training the neural network imaging model using an early stopping method based on the training set to obtain an optimal solution for the network connection weights in the neural network imaging model; step S5, obtaining an optimized neural network imaging model based on the optimal solution for the network connection weights, and obtaining a reconstructed image of the impedance distribution of the area to be measured based on the optimized neural network imaging model.

[0032] In step S1 , for example, the electrode belt of the EIT hardware device is directly or indirectly connected to the chest area to be tested of the tester, and voltage measurement data during the contraction and relaxation of the tester's lungs is collected.

[0033] In step S2, the area to be tested is triangulated using triangles as the subdivision unit and the finite element method is used to perform triangulation. Simulation examples of different lung conditions are generated in the area to be tested, and the impedance distribution data value of the target area is set. Figure 2 As shown, the finite element division details of the area to be measured.

[0034] Specifically, the shape, density, electrode arrangement and other details of the finite element partitioning model are selected and set through the mk_common_model function in the Eidors toolbox to obtain a finite element partitioning that conforms to the attributes of the human body. Preferably, the determined finite element model indicators are: 1312 finite element partitions, 715 grid nodes, and 16 electrodes (arranged clockwise), which can accurately distinguish the lung contour and ensure the timeliness of the calculation. Among them, each finite element partition has a definite and accurate coordinate label, which can keep the conductivity change information consistent with the position correspondence.

[0035] Furthermore, the 6-layer neural network in step S3 is in a fully connected form. The 6-layer fully connected neural network realizes the image reconstruction process by forward propagating data and back propagating gradients. Figure 3 The first to third layers in the network are the encoding process of input feature extraction and learning, and the third to sixth layers are the decoding process of mapping feature values ​​to high-dimensional feature values, and finally reconstructing the predicted conductivity pixel distribution image. The whole process is called forward propagation, in which the connection between the input and output of the nth neuron in each layer of the network is:

[0036]

[0037] In the formula, f n is the output of the nth neuron, m represents the total number of neurons in the corresponding layer, and w jn is the jth input z of the nth neuron j The connection weight between the neuron and b n is the bias of the nth neuron.

[0038] Furthermore, the loss function used by the 6-layer fully connected neural network in step S3 adopts the cross entropy function plus the regularization constraint term, that is:

[0039]

[0040] Where N represents the total number of pixels in the reconstructed image, x i Represents the set i-th true conductivity pixel value, represents the ith conductivity pixel value predicted by the grid, λ represents the regularization parameter, and L2 represents the regularization constraint term.

[0041] Furthermore, in the neural network model construction part of step S3, for example, a deep structure design is adopted, that is, a fully connected neural network architecture including an input layer, five hidden layers, and an output layer. Each layer of neurons is fully interconnected, and this dense connection method allows information to fully flow and integrate throughout the network. Each hidden layer is embedded with an activation function ReLU to enhance the model's ability to express nonlinear relationships, so that it can capture complex patterns in the electrical impedance distribution, allowing the model to deeply explore and express subtle differences and complex changes in the electrical impedance distribution. In particular, after the last hidden layer, an activation function tanh suitable for regression tasks is selected to ensure that the model output conforms to the actual physical properties of the electrical impedance distribution.

[0042] Specifically, for example, the number of boundary voltage values ​​is used as the number of input layer nodes of the neural network, the impedance values ​​of different lung breathing conditions are used as the number of output layer nodes of the neural network, and the number of hidden layer nodes is determined according to the simulation experiment results, so as to establish a mapping relationship between "boundary measurement voltage value" and "impedance value of lung breathing condition". Preferably, the input layer is a 208*1 column vector, which represents the input voltage, that is, the voltage signal obtained by the EIT hardware acquisition device. The number of neurons in the five hidden layers is: 820*1, 620*1, 480*1, 620*1, 820*1, which can provide a sufficient neural network structure for the mapping model. The final output layer is a 1312*1 column vector, which strictly corresponds to the number of finite element divisions and is the calculation result of the impedance value change during the imaging process.

[0043] Furthermore, in step S4, in order to obtain the best generalization performance of the 6-layer fully connected neural network and prevent the model from overfitting, a simple and effective early stopping method is used to find the optimal solution for the network connection weights in the fully connected neural network imaging model to ensure high accuracy in the model mapping relationship. In addition, this operation reduces the effective scale of each parameter dimension, optimizes and improves the parameter scale and computational timeliness, and finally establishes a high-precision mathematical mapping model from the measured voltage data to the real electrical impedance distribution.

[0044] Specifically, the conductivity change and corresponding voltage value data obtained from the positive problem are divided into a training set and a validation set. After each epoch, the test results are obtained on the validation set. As the epoch increases, if the test error is found to increase on the validation set, the training is stopped; otherwise, it is continued. Finally, the weights after stopping are used as the final parameters of the network. To avoid misjudgment, in the neural network training, when the best validation set accuracy so far is recorded, 12 epochs are performed continuously. If the accuracy does not improve any more, it is considered to have reached the highest accuracy.

[0045] In step S5, the boundary voltage value of the object to be measured is input as input data into the optimized neural network imaging model to obtain the impedance distribution data of the object to be measured, and the color value of the corresponding rule is assigned according to the impedance distribution data to obtain the impedance imaging diagram.

[0046] In summary, the electrical impedance imaging method based on a multi-layer fully connected neural network proposed in the first embodiment of the present invention uses the finite element method to triangulate the measured area, first discretizes the problem to be solved for unit analysis, and then combines the results of each unit analysis to conduct a comprehensive analysis of the whole, so as to use finite difference equations to approximate the partial differential equations in the EIT problem, so as to obtain the value of the field function at each discrete point, accurately and quickly approximate the true solution of the nonlinear problem, and establish a true and effective finite element model of the lung cross section; using a real lung electrical impedance distribution sample as a supervisory signal, the parameters of the neural network model are optimized through the back propagation algorithm to achieve the transformation from body surface voltage to body surface voltage. Accurate mapping of measured data to the electrical impedance distribution in the body; using the multi-level abstract feature learning mechanism of a multi-layer fully connected neural network to effectively improve the quality and resolution of image reconstruction and reduce the impact of noise interference, thereby solving the problem of insufficient image accuracy faced by traditional EIT technology in clinical applications; the activation function ReLU used in the neural network model to enhance the model's ability to express nonlinear relationships, so that it can capture complex patterns in the electrical impedance distribution, allowing the model to deeply explore and express subtle differences and complex changes in the electrical impedance distribution, and select the activation function tanh suitable for regression tasks to ensure that the model output conforms to the actual physical properties of the electrical impedance distribution.

[0047] In addition, if Figure 4 As shown, the second embodiment of the present invention also proposes an electrical impedance imaging device 20 based on a multi-layer fully connected neural network, including: a voltage measurement data acquisition module 201, a finite element model construction module 202, a neural network imaging model establishment module 203, a neural network imaging model training module 204 and an impedance distribution image reconstruction module 205.

[0048] Among them, the voltage measurement data acquisition module 201 is used to obtain the voltage measurement data of the human chest area to be tested through the EIT device. The finite element model construction module 202 is used to triangulate the area to be tested using the finite element method to obtain a finite element model. The neural network imaging model establishment module 203 is used to establish a multi-layer fully connected neural network imaging model based on the finite element model, input the voltage measurement data into the neural network imaging model to obtain the initial distribution sequence of the conductivity of the area to be tested, and normalize the initial distribution sequence to obtain a training set. The neural network imaging model training module 204 is used to train the neural network imaging model using the early stopping method according to the training set to obtain the optimal solution of the network connection weights in the neural network imaging model. The impedance distribution image reconstruction module 205 is used to obtain an optimized neural network imaging model based on the optimal solution of the network connection weights, and obtain a reconstructed image of the impedance distribution of the area to be tested based on the optimized neural network imaging model.

[0049] The electrical impedance imaging method based on a multi-layer fully connected neural network implemented by the electrical impedance imaging device based on a multi-layer fully connected neural network disclosed in the second embodiment of the present invention is as described in the first embodiment above, so it will not be described in detail here. Optionally, the various modules and the above-mentioned other operations or functions in the second embodiment are respectively for implementing the method described in the first embodiment, and the beneficial effects of the electrical impedance imaging device based on a multi-layer fully connected neural network provided by this embodiment are the same as the beneficial effects of the electrical impedance imaging method based on a multi-layer fully connected neural network provided by the first embodiment above, and for the sake of brevity, they are not repeated here.

[0050] like Figure 5 As shown, the third embodiment of the present invention further provides an electronic device, 30, for example, comprising: at least one processing unit 31, and at least one storage unit 32, wherein the storage unit 32 stores a computer program, and when the computer program is executed by the processing unit 31, the processing unit 31 executes the method as described in the first embodiment, and the beneficial effects of the electronic device 30 provided by this embodiment are the same as the beneficial effects of the electrical impedance imaging method based on a multi-layer fully connected neural network provided in the first embodiment.

[0051] The fourth embodiment of the present invention also provides a computer-readable storage medium 40 on which a computer program is stored. When the program is executed by a processor, the steps of the above method are implemented. The beneficial effects of the computer-readable storage medium 40 provided by this embodiment are the same as the beneficial effects of the electrical impedance imaging method based on a multi-layer fully connected neural network provided by the first embodiment.

[0052] In addition, it can be understood that the aforementioned embodiments are merely illustrative descriptions of the present application. Under the premise that the technical features do not conflict, the structures do not contradict, and the purpose of the invention of the present application is not violated, the technical solutions of the various embodiments can be arbitrarily combined and used in combination.

[0053] Among them, computer-readable storage media may include, but are not limited to, any type of disk, including floppy disks, optical disks, DVDs, CD-ROMs, microdrives and magneto-optical disks, ROMs, RAMs, EPROMs, EEPROMs, DRAMs, VRAMs, flash memory devices, magnetic or optical cards, nanosystems (including molecular memory ICs), or any type of medium or device suitable for storing instructions and / or data.

[0054] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0055] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0056] In the several embodiments provided in the present application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of the units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0057] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0058] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0059] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a memory, including a number of instructions to enable a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned memory includes: U disk, read-only memory (ROM), random access memory (RAM), mobile hard disk, disk or optical disk and other media that can store program code.

[0060] A person skilled in the art may understand that all or part of the steps in the various methods of the above embodiments may be completed by entering a program to instruct the relevant hardware. The program may be stored in a computer-readable memory, and the memory may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0061] The above is only an exemplary embodiment of the present disclosure, and the scope of the present disclosure cannot be limited thereto. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the specification and practicing the disclosure here, those skilled in the art will easily think of the implementation scheme of the present disclosure. This application is intended to cover any modification, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the technical field not recorded in the present disclosure. The description and examples are regarded as exemplary only, and the scope and spirit of the present disclosure are defined by the claims.

[0062] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0063] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An electrical impedance tomography method based on a multi-layer fully connected neural network, characterized in that: include: Obtain voltage measurement data of the human chest area to be tested through the EIT device; Using the finite element method to triangulate the area to be tested to obtain a finite element model; Establishing a multi-layer fully connected neural network imaging model according to the finite element model, inputting the voltage measurement data into the neural network imaging model to obtain an initial distribution sequence of the conductivity of the area to be measured, and normalizing the initial distribution sequence to obtain a training set; Training the neural network imaging model using an early stopping method according to the training set to obtain an optimal solution for network connection weights in the neural network imaging model; An optimized neural network imaging model is obtained according to the optimal solution of the network connection weights, and a reconstructed image of the impedance distribution of the area to be measured is obtained according to the optimized neural network imaging model.

2. The electrical impedance tomography method based on a multi-layer fully connected neural network according to claim 1, characterized in that: The neural network imaging model is a six-layer fully connected neural network architecture, including an input layer, five hidden layers and an output layer, and the neurons in each layer are fully interconnected; Among them, the first to third layers are the encoding process of input feature extraction and learning, and the third to sixth layers are the decoding process of mapping input feature values ​​to high-dimensional feature values.

3. The electrical impedance tomography method based on a multi-layer fully connected neural network according to claim 2, characterized in that: Each hidden layer introduces an activation function ReLU to express the nonlinear relationship between the boundary measurement voltage value and the lung respiratory impedance value; the last hidden layer also introduces an activation function tanh to perform regression tasks to ensure that the neural network imaging model outputs an impedance distribution that conforms to actual physical properties.

4. The electrical impedance tomography method based on a multi-layer fully connected neural network according to claim 2, characterized in that: The voltage measurement data includes the boundary voltage values ​​of the area to be measured. The number of the boundary voltage values ​​is used as the number of input layer nodes of the neural network, the number of finite element divisions corresponding to the number of lung respiratory impedance values ​​is used as the number of output layer nodes of the neural network, and the number of hidden layer nodes of the neural network is determined according to the simulation results, thereby establishing a mapping relationship between the boundary voltage value and the lung respiratory impedance value.

5. The electrical impedance tomography method based on a multi-layer fully connected neural network according to claim 2, characterized in that: The connection between the input and output of the nth neuron in each layer of the neural network imaging model is: Among them, f n is the output of the nth neuron, m represents the total number of neurons in the corresponding layer, and w jn is the jth input z of the nth neuron j The connection weight between the neuron and b n is the bias of the nth neuron.

6. The electrical impedance tomography method based on a multi-layer fully connected neural network according to claim 2, characterized in that: The loss function used in the neural network imaging model adopts the cross entropy function plus the L2 regularization constraint term, which is expressed as: Where N represents the total number of pixels in the reconstructed image, x i Represents the set i-th true conductivity pixel value, represents the ith conductivity pixel value predicted by the grid, λ represents the regularization parameter, and L2 represents the regularization constraint term.

7. The electrical impedance tomography method based on a multi-layer fully connected neural network according to claim 2, characterized in that: The training of the neural network imaging model includes: The conductivity change and the corresponding voltage value data are divided into a training set and a validation set. The neural network imaging model obtains test results on the validation set after each training. If the test error is found to increase on the validation set, the training is stopped, otherwise the training is continued, and the weights obtained after the training is stopped are used as the final parameters of the neural network imaging model.

8. An electrical impedance imaging device based on a multi-layer fully connected neural network, characterized in that: include: A voltage measurement data acquisition module is used to acquire voltage measurement data of a human chest area to be tested through an EIT device; A finite element model building module, used for triangulating the area to be tested using the finite element method to obtain a finite element model; A neural network imaging model building module, used to build a multi-layer fully connected neural network imaging model according to the finite element model, input the voltage measurement data into the neural network imaging model to obtain an initial distribution sequence of the conductivity of the area to be measured, and normalize the initial distribution sequence to obtain a training set; A neural network imaging model training module, used to train the neural network imaging model using an early stopping method according to the training set to obtain an optimal solution for network connection weights in the neural network imaging model; The impedance distribution image reconstruction module is used to obtain an optimized neural network imaging model according to the optimal solution of the network connection weights, and obtain a reconstructed image of the impedance distribution of the area to be measured according to the optimized neural network imaging model.

9. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method according to any one of claims 1 to 7 when executing the program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to execute the steps of the method according to any one of claims 1 to 7.

Images