Three-dimensional microwave brain imaging method and apparatus based on image compression

Abstract

The application discloses a three-dimensional microwave brain imaging method and device based on image compression, which comprises the following steps: based on the microwave brain imaging, a three-dimensional antenna array is used to emit microwaves to detect the brain, a deep neural network is first trained to realize the encoding and decoding of the two-dimensional section image of the human brain dielectric constant, and then the three-dimensional brain model is represented by the hidden space parameters of the continuous section. In the three-dimensional brain imaging, given the measured microwave data, the objective function with the multilayer section hidden space parameters as the independent variables is minimized, the optimal hidden space parameters are solved, and finally the solved hidden space parameters are decoded into the three-dimensional human brain dielectric constant model. The application replaces the three-dimensional human brain represented by the point array with the hidden space parameters of the deep neural network, greatly reduces the imaging unknowns, and has higher calculation efficiency. The common structural features of the human brain can be included in the inversion by training the neural network, and the ill-posedness of the inverse problem solving is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of biomedical computing imaging, in particular to a three-dimensional microwave brain imaging method and device based on image compression. BACKGROUND

[0002] Microwave imaging has the advantages of non-invasiveness, no radiation damage, and large detection depth. Its measurement equipment is convenient and has low requirements for the measurement environment, making it an ideal imaging method suitable for early screening and bedside dynamic monitoring of stroke. The electrical parameters (dielectric constant and conductivity) of brain tissue and blood are different, and intracerebral hemorrhage or ischemic injury can change the spatial distribution of electrical parameters in the brain, thereby affecting the spatial electromagnetic field distribution and electromagnetic wave propagation. By using microwave detection imaging technology, the electrical parameter distribution in the brain can be reconstructed from the electromagnetic signals measured by the sensor, and the diagnosis of abnormal lesions in the brain tissue can be realized.

[0003] Common imaging methods decompose the inversion domain into pixels, and then reconstruct the discrete dielectric constant by minimizing the residual between the simulated data and the measured data. The number of pixels is usually much larger than the number of data, so it is a nonlinear ill-posed problem that needs to rely on prior knowledge for reasonable reconstruction. Since the number of unknowns is usually tens of thousands or even millions, the calculation amount of the minimization process of the objective function is huge. In addition, the way of using prior information in inversion is not flexible enough. The common structural features of the human brain are difficult to describe in mathematical form, so they cannot constrain the inversion process, bringing challenges to model reconstruction and interpretation. SUMMARY

[0004] The present application aims to at least partially solve one of the problems in the related art.

[0005] To this end, a three-dimensional human brain electromagnetic model big data set is constructed based on the electromagnetic property data of organs and tissues and the CT / MRI three-dimensional model of the human brain, and the common features of the human brain are autonomously extracted using deep learning technology and integrated into the three-dimensional reconstruction algorithm, which can reduce the inversion multiplicity and improve the structural features of microwave imaging. The present application proposes a three-dimensional microwave brain imaging method based on image compression. The structural information and dielectric constant information of the human brain can be comprehensively utilized to realize high compression rate of the three-dimensional model and improve the speed and accuracy of three-dimensional microwave brain imaging.

[0006] The second aspect of the present application is to propose a three-dimensional microwave brain imaging device based on image compression.

[0007] To achieve the above purpose, the present application proposes a three-dimensional microwave brain imaging method based on image compression, which comprises:

[0008] constructing an image coding deep neural network using a two-dimensional image training set based on two-dimensional cross-sectional images;

[0009] optimizing an objective function of the image coding deep neural network to train the image coding deep neural network to obtain network parameters;

[0010] constructing a three-dimensional brain coding vector based on the network parameters and the two-dimensional sectional image, and constructing an inversion objective function according to the three-dimensional brain coding vector and microwave brain measurement data;

[0011] decoding the three-dimensional brain coding vector into a three-dimensional brain dielectric constant model represented by a dot matrix based on the optimized inversion objective function, to perform three-dimensional microwave brain imaging according to the three-dimensional brain dielectric constant model.

[0012] The image compression-based three-dimensional microwave brain imaging device implemented by the application can further have the following additional technical features:

[0013] Further, the two-dimensional sectional image includes a two-dimensional sectional image describing the spatial distribution of brain dielectric constant based on a brain CT or MRI image.

[0014] Further, the objective function is trained by minimizing the objective function through an optimization algorithm until the process converges, to obtain the trained image coding deep neural network.

[0015] Further, the three-dimensional brain coding vector includes a latent space parameter of the two-dimensional sectional image.

[0016] Further, the brain training set includes input data and label data; and the structure of the image coding deep neural network includes multiple forms.

[0017] To achieve the above object, the application further provides an image compression-based three-dimensional microwave brain imaging device, which comprises:

[0018] a network construction module configured to construct an image coding deep neural network by using a two-dimensional image training set based on a two-dimensional sectional image;

[0019] a network training module configured to optimize an objective function of the image coding deep neural network to train the image coding deep neural network to obtain network parameters;

[0020] an objective function construction module configured to construct a three-dimensional brain coding vector based on the network parameters and the two-dimensional sectional image, and construct an inversion objective function according to the three-dimensional brain coding vector and microwave brain measurement data;

[0021] a microwave imaging module configured to decode the three-dimensional brain coding vector into a three-dimensional brain dielectric constant model represented by a dot matrix based on the optimized inversion objective function, to perform three-dimensional microwave brain imaging according to the three-dimensional brain dielectric constant model.

[0022] The three-dimensional microwave brain imaging method and device based on image compression according to the embodiment of the present application use the hidden space parameters of the deep neural network to replace the three-dimensional human brain represented by the dot matrix, greatly reduce the imaging unknowns, and have higher calculation efficiency. The common structural features of the human brain can be included in the inversion by training the neural network, and the ill-conditioned nature of the inverse problem solving is reduced.

[0023] Additional aspects and advantages of the present application will be in part apparent and in part pointed out hereinafter. BRIEF DESCRIPTION OF DRAWINGS

[0024] The above and / or additional aspects and advantages of the present application will become apparent and be readily appreciated from the following description, taken in conjunction with the accompanying drawings, in which:

[0025] Figure 1 A flowchart of the three-dimensional microwave brain imaging method based on image compression according to the embodiment of the present application is shown in FIG. 1.

[0026] Figure 2 An architecture diagram of the three-dimensional microwave brain imaging method based on image compression according to the embodiment of the present application is shown in FIG. 2.

[0027] Figure 3 A partial training data diagram used in the simulation example according to the embodiment of the present application is shown in FIG. 3.

[0028] FIGS. 4(a), 4(b) and 4(c) are respectively encoding test diagrams according to the embodiment of the present application.

[0029] FIGS. 5(a), 5(b) and 5(c) are respectively comparison diagrams of the microwave imaging results and the conventional pixel-based microwave imaging results according to the embodiment of the present application.

[0030] Figure 6 An image compression neural network diagram according to the embodiment of the present application is shown in FIG. 6.

[0031] Figure 7 A structure diagram of the three-dimensional microwave brain imaging device based on image compression according to the embodiment of the present application is shown in FIG. 7. DETAILED DESCRIPTION

[0032] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0033] In order to make the person skilled in the art better understand the technical scheme of the present application, the technical scheme in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by the person skilled in the art without creative labor should belong to the protection scope of the present application.

[0034] The image compression-based three-dimensional microwave brain imaging method and device according to the embodiments of the present application are described below with reference to the drawings.

[0035] Figure 1 is a flowchart of the image compression-based three-dimensional microwave brain imaging method according to the embodiments of the present application.

[0036] As Figure 1 shown, the method includes but is not limited to the following steps:

[0037] S1, constructing an image coding deep neural network by using a two-dimensional image training set based on two-dimensional cross-sectional images;

[0038] S2, optimizing an objective function of the image coding deep neural network to train the image coding deep neural network to obtain network parameters;

[0039] S3, constructing a three-dimensional human brain coding vector based on the network parameters and the two-dimensional cross-sectional images, and constructing an inversion objective function according to the three-dimensional human brain coding vector and the microwave brain measurement data;

[0040] S4, decoding the three-dimensional human brain coding vector into a three-dimensional human brain dielectric constant model represented by a dot matrix based on the optimized inversion objective function, to perform three-dimensional microwave brain imaging according to the three-dimensional human brain dielectric constant model.

[0041] Specifically, the present application pre-trains a deep neural network to realize the coding and decoding of the two-dimensional cross-section of the human brain dielectric constant, and the inversion of the three-dimensional brain measurement microwave data, as Figure 2 shown, wherein the pre-trained deep neural network includes the following steps: constructing two-dimensional cross-sectional images describing the spatial distribution of brain dielectric constant by brain CT or MRI images, to construct a human brain two-dimensional electromagnetic model training set; designing a deep neural network (including an encoder and a decoder) for image compression according to the training set; constructing an objective function for training the image compression deep neural network; minimizing the training of the objective function by an optimization algorithm until the process converges, to obtain the trained image compression deep neural network;

[0042] Furthermore, the three-dimensional brain measurement microwave data inversion includes the following steps: acquiring microwave brain measurement data; representing the three-dimensional human brain using two-dimensional continuous image cross-sections, and constructing a three-dimensional human brain encoding vector composed of latent space parameters of the two-dimensional continuous cross-sections; constructing a three-dimensional brain measurement microwave data inversion objective function with the three-dimensional human brain encoding vector as the independent variable; minimizing the objective function using an optimization algorithm; and decoding the inverted three-dimensional human brain encoding vector into a three-dimensional human brain dielectric constant model represented by a dot matrix.

[0043] Furthermore, the human brain training set includes input data and labeled data.

[0044] Furthermore, the structure of deep neural networks includes various forms.

[0045] Furthermore, the training objective function of deep neural networks can take various forms.

[0046] Furthermore, algorithms for optimizing the objective function of deep neural networks come in various forms.

[0047] Furthermore, algorithms for optimizing the objective function of microwave data inversion come in various forms.

[0048] Furthermore, some of the training data used in the simulation examples of this invention are as follows: Figure 3 The image shown is a two-dimensional dielectric constant (real part) image of the human brain constructed by MRI cross-sectional scanning; the image size is 0.3*0.3m2, and the color represents the magnitude of the relative dielectric constant.

[0049] Furthermore, Figures 4(a), 4(b), and 4(c) are schematic diagrams of the coding test, namely, cross-sections of the three-dimensional human brain electromagnetic model and the codes generated by the neural network. It can be understood that this three-dimensional human brain electromagnetic model is described by a 100*100*100 dot matrix. After image compression coding, each transverse cross-section generates a 128*1 latent space code, resulting in 100 128*1 latent space variables.

[0050] Furthermore, Figures 5(a), 5(b), and 5(c) show a comparison between microwave imaging results according to an embodiment of the present invention and conventional pixel-based microwave imaging results. Figure 5(a) represents the actual model, Figure 5(b) represents the microwave imaging results according to an embodiment of the present invention, and Figure 5(c) represents the conventional pixel-based microwave imaging results.

[0051] Furthermore, such as Figure 6 As shown, the input and output sizes of the image compression neural network are 100*100. When the input is a 100*100 cross-sectional image m of a human brain, the encoder compresses it into a 128*1 vector v. The vector v is then decoded by the decoder to output the image. The training objective is to minimize the difference between m and... The difference between the original image and the reconstructed image is as small as possible.

[0052] In three-dimensional microwave brain imaging, the microwave data is simulated by the following equation:

[0053] d = F(M) (1)

[0054] where M is the permittivity discretized by a lattice with size 100*100*100 in this example, d is the simulated scattering field data according to Maxwell equations, and F is the forward modeling function.

[0055] The encoding vector V representing the three-dimensional human brain is composed of all the hidden space variables of the two-dimensional sections of the human brain, as shown in Fig. 4(c), i.e.

[0056] V = [v1, v2,..., v K ], v i = E(m(:,:,i)), i = 1, 2,..., K (2)

[0057] Meanwhile, the decoding of V can recover the lattice description of the three-dimensional human brain model M = D(V). Therefore, the process of calculating electromagnetic data from hidden space variables can be represented by S:

[0058] d = F(D(V)) = S(V) (3)

[0059] Microwave brain imaging can be considered as finding the optimal parameter update ΔV that minimizes the following objective function:

[0060]

[0061] where || || denotes the L2 norm, V k is the initial value, d obs is the microwave measurement data, is the gradient operator in the vertical direction, and γ and λ are the regularization coefficients of the objective function, which are used to stabilize the function optimization process.

[0062] In the Gauss-Newton method, the minimization of equation (5) is performed by iteration. In the kth iteration, ΔV can be solved by the following matrix equation:

[0063] H k ·ΔV = -g k , (5)

[0064] In the equation:

[0065]

[0066]

[0067]

[0068] where J is the Jacobian matrix, which is obtained by the chain rule: first, the derivative of the microwave data with respect to the model parameters is calculated, then the derivative of the model parameters with respect to the latent space variables is calculated, and finally the two are added by vector dot multiplication. More specifically, J p,q represents the element of the pth row and the pth column of the Jacobian matrix, and means the derivative of the pth microwave data d p with respect to the qth latent space variable V q is obtained by the derivative of the pth microwave data d p with respect to the lth discrete permittivity M l is obtained by the derivative of the lth discrete permittivity M l with respect to the qth latent space variable V q is obtained by the chain rule.

[0069] After △V is obtained, the starting value V0 is updated by the following formula:

[0070] V0 = V0 + △V (9)

[0071] After V0 is obtained, it is decoded by the decoder D, and the permittivity model M is obtained:

[0072] M = D(V0) (10)

[0073] The above process is iterated until the simulation data and the observed data are fitted, the inversion is ended, and the final decoded model M is output.

[0074] According to the three-dimensional microwave brain imaging method based on image compression provided in the embodiments of the present application, the latent space parameters of the deep neural network are used to replace the three-dimensional human brain represented by the point array, so that the unknowns of imaging are greatly reduced, and the calculation efficiency is higher. The common structural features of the human brain can be included in the inversion by training the neural network, so as to reduce the ill-conditioned nature of the inverse problem.

[0075] In order to realize the above-mentioned embodiments, as Figure 7 shown in the embodiments, the three-dimensional microwave brain imaging device 10 based on image compression is also provided, which comprises a network construction module 100, a network training module 200, a target function construction module 300 and a microwave imaging module 400.

[0076] The network construction module 100 is used to construct an image coding deep neural network by using a two-dimensional image training set based on a two-dimensional sectional image;

[0077] The network training module 200 is used to optimize the target function of the image coding deep neural network, so as to train the image coding deep neural network to obtain network parameters;

[0078] The target function construction module 300 is configured to construct a three-dimensional brain encoding vector based on the network parameters and the two-dimensional cross-sectional image, and construct an inversion target function according to the three-dimensional brain encoding vector and the microwave brain measurement data;

[0079] The microwave imaging module 400 is configured to decode the three-dimensional brain encoding vector into a three-dimensional brain dielectric constant model in a dot array representation based on the optimized inversion target function, and perform three-dimensional microwave brain imaging according to the three-dimensional brain dielectric constant model.

[0080] Further, the two-dimensional cross-sectional image comprises a two-dimensional cross-sectional image describing the spatial distribution of brain dielectric constant constructed based on a brain CT or MRI image.

[0081] Further, the trained image encoding deep neural network is obtained by minimizing the training target function through an optimization algorithm until the process converges.

[0082] Further, the three-dimensional brain encoding vector comprises a latent space parameter of the two-dimensional cross-sectional image.

[0083] Further, the brain training set comprises input data and label data; and the structure of the image encoding deep neural network comprises various forms.

[0084] According to the three-dimensional microwave brain imaging device based on image compression, the latent space parameter of the deep neural network is used to replace the three-dimensional brain in a dot array representation, so that the unknowns of imaging are greatly reduced, and the calculation efficiency is higher. The common structural features of the brain can be included in the inversion by training the neural network, so as to reduce the ill-conditioned nature of the inverse problem.

[0085] It should be noted that the computer-readable medium in the above disclosure can be a computer-readable signal medium or a computer-readable storage medium or any combination of the two. The computer-readable storage medium may, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or apparatus, or any combination of the above. More specific examples of computer-readable storage media can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (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 disclosure, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device or apparatus. In the present disclosure, the computer-readable signal medium can include a data signal carried in a baseband or as a part of a carrier wave, which carries computer-readable program code. Such a propagated data signal can take many forms, including but not limited to an electromagnetic signal, an optical signal or any suitable combination of the above. The computer-readable signal medium can also be any computer-readable medium other than the computer-readable storage medium, which can send, propagate or transmit a program for use by or in conjunction with an instruction execution system, device or apparatus. The program code contained in the computer-readable medium can be transmitted by any suitable medium, including but not limited to a wire, a cable, an RF (radio frequency) or the like, or any suitable combination of the above.

[0086] The computer-readable medium described above can be contained in the electronic device described above; or can exist separately and not be assembled into the electronic device. The computer-readable medium described above carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the determination method of the sediment content in a flowing water body described above.

[0087] Computer program code for carrying out operations of the present disclosure can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0088] In the description of the specification, the description of the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Also, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or more embodiments or examples. In addition, the person skilled in the art can combine and combine the different embodiments or examples described in the specification and the features of the different embodiments or examples, without contradiction.

[0089] In addition, the terms "first", "second", etc. are used only for the purpose of description and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features defined with "first", "second" can explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "a plurality of" is at least two, for example, two, three, etc., unless otherwise specifically limited.

[0090] Any process or method descriptions or descriptions of the flow diagrams in the specification or elsewhere in this document, can be understood as representing the steps of the code of the modules, segments or portions of the code for implementing specific logic functions or steps in the process, and the scope of the preferred embodiments of the present application includes additional implementation in which the steps are performed in different order, including an essentially simultaneous performance of the functions according to the involved functions, or in reverse order, which should be understood by those skilled in the art of the embodiments of the present application.

[0091] The logic and / or steps represented in the flowcharts and / or described herein, for example, can be considered as a sequence of instructions to implement logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, apparatus, or device, such as a computer-based system, processor- based system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For purposes of this specification, a "computer-readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can be a computer- readable storage medium or a computer-readable signal medium. The computer- readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer-readable storage medium include the following: an electrical connection having one or more wires (electrical connections), a portable computer diskette (a magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium can even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, for example, via optical scanning of the paper or other medium, then compiled, interpreted, or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.

[0092] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the above embodiments, the various steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, and in another embodiment, any of the following technologies, known in the art, or their combinations can be used: discrete logic circuitry having logic gates for implementing logic functions on data signals, application specific integrated circuits having appropriate combinational logic gates, programmable gate arrays (PGA), field programmable gate arrays (FPGA), and the like.

[0093] Those skilled in the art can understand that all or part of the steps carried out by the above-mentioned embodiment methods can be completed by programs instructing related hardware, and the programs can be stored in a computer-readable storage medium. When the programs are executed, they include one or a combination of the steps of the method embodiments.

[0094] In addition, each of the functional units in the various embodiments of the present application can be integrated in one processing module, or each of the units can be physically present separately, or two or more units can be integrated in one module. The integrated module can be realized in the form of hardware or in the form of a software functional module. When the integrated module is realized in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer readable storage medium.

[0095] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A method of three-dimensional microwave brain imaging based on image compression, characterized in that, The method comprises the following steps: constructing an image coding deep neural network by using a two-dimensional image training set based on two-dimensional section images; optimizing an objective function of the image coding deep neural network to train the image coding deep neural network and obtain network parameters; constructing a three-dimensional brain coding vector based on the network parameters and the two-dimensional section images, and constructing an inversion objective function according to the three-dimensional brain coding vector and microwave brain measurement data; decoding the three-dimensional brain coding vector into a three-dimensional brain dielectric constant model represented by a dot matrix based on the optimized inversion objective function, and performing three-dimensional microwave brain imaging according to the three-dimensional brain dielectric constant model.

2. The method of claim 1, wherein, The two-dimensional section images comprise two-dimensional section images describing spatial distribution of brain dielectric constants, which are constructed based on brain CT or MRI images.

3. The method of claim 1, wherein, The target function is minimized by an optimization algorithm until the process converges, and a trained image coding deep neural network is obtained.

4. The method of claim 1, wherein, The three-dimensional brain coding vector comprises hidden space parameters of the two-dimensional section images.

5. The method of claim 1, wherein, The two-dimensional image training set comprises input data and label data, and the structure of the image coding deep neural network comprises multiple forms.

6. An apparatus for three-dimensional microwave brain imaging based on image compression, characterized in that, The method comprises the following steps: a network construction module configured to construct an image coding deep neural network by using a two-dimensional image training set based on two-dimensional section images; a network training module configured to optimize an objective function of the image coding deep neural network to train the image coding deep neural network and obtain network parameters; an objective function construction module configured to construct a three-dimensional brain coding vector based on the network parameters and the two-dimensional section images, and construct an inversion objective function according to the three-dimensional brain coding vector and microwave brain measurement data; a microwave imaging module configured to decode the three-dimensional brain coding vector into a three-dimensional brain dielectric constant model represented by a dot matrix based on the optimized inversion objective function, and perform three-dimensional microwave brain imaging according to the three-dimensional brain dielectric constant model.

7. The apparatus of claim 6, wherein, The two-dimensional section images comprise two-dimensional section images describing spatial distribution of brain dielectric constants, which are constructed based on brain CT or MRI images.

8. The apparatus of claim 6, wherein, The target function is minimized by an optimization algorithm until the process converges, and a trained image coding deep neural network is obtained.

9. The apparatus of claim 6, wherein, The three-dimensional brain coding vector comprises hidden space parameters of the two-dimensional section images.

10. The apparatus of claim 6, wherein, The two-dimensional image training set comprises input data and label data, and the structure of the image coding deep neural network comprises multiple forms.

Images