Magnetic resonance imaging method and apparatus based on quantum convolutional neural network

Abstract

A magnetic resonance imaging method and apparatus based on a quantum convolutional neural network, relating to the field of magnetic resonance imaging. The method comprises: acquiring a quantum convolutional layer (100); replacing a convolutional layer in a convolutional neural network with the quantum convolutional layer to obtain a corresponding quantum convolutional neural network (110); and acquiring an undersampled K-space image, and inputting the undersampled K-space image into the quantum convolutional neural network to obtain a corresponding fully sampled reconstructed image (120).

Description

Magnetic resonance imaging method and device based on quantum convolutional neural network Technical Field

[0001] The present application relates to the field of magnetic resonance imaging. Specifically, the present application relates to a magnetic resonance imaging method, device, electronic device and storage medium based on quantum convolutional neural network. Background Art

[0002] Magnetic resonance imaging (MRI) utilizes static and radiofrequency magnetic fields to image human tissue. It provides rich tissue contrast and is harmless to the human body, making it a powerful tool for clinical diagnosis. However, slow imaging speed has been a major bottleneck restricting its rapid development. Improving scanning speed and thus reducing scanning time while maintaining clinically acceptable imaging quality is crucial.

[0003] Since the emergence of deep learning, using deep neural networks to accelerate MRI image reconstruction has become a mainstream approach in the study of fast MRI. Deep learning algorithms use neural networks to learn the optimal reconstruction parameters from large amounts of training data, or directly learn the mapping relationship between undersampled data and fully sampled images. This allows for better imaging quality and higher acceleration than traditional parallel imaging or compressed sensing methods.

[0004] Traditional parallel imaging uses the sensitivity information of the coil for acceleration, but the acceleration factor is limited, and as the acceleration factor increases, the image will show noise amplification; compressed sensing technology often requires a long reconstruction time due to the use of nonlinear reconstruction, and the reconstruction parameters are difficult to select; although deep learning methods have made up for the shortcomings of traditional fast imaging methods, they also have some problems. For example, data-driven deep learning lacks theoretical guidance and often requires a large amount of training data to achieve good results. Although the model-driven deep learning algorithm that only learns reconstruction parameters requires less training data, the imaging quality will not be particularly high.

[0005] From this we can see that traditional deep neural networks require a large amount of training data to ensure imaging quality, which leads to low imaging efficiency. This situation needs further improvement.

[0006] Summary of the Invention

[0007] This application provides a magnetic resonance imaging method, device, electronic device, and storage medium based on quantum convolutional neural networks, which can solve the problem of low imaging efficiency in related technologies. The technical solution is as follows:

[0008] According to one aspect of the present application, a magnetic resonance imaging method based on a quantum convolutional neural network includes: obtaining a quantum convolutional layer; replacing the convolutional layer in a convolutional neural network with the quantum convolutional layer to obtain a corresponding quantum convolutional neural network, wherein the convolutional neural network includes a fully connected layer, a convolutional layer and a final deconvolution layer; obtaining an undersampled K-space image, and inputting the undersampled K-space image into the quantum convolutional neural network to obtain a corresponding fully sampled reconstructed image.

[0009] According to one aspect of the present application, a magnetic resonance imaging device based on a quantum convolutional neural network includes:

[0010] Quantum convolution layer acquisition module, used to obtain quantum convolution layer;

[0011] A quantum convolutional neural network acquisition module replaces the convolutional layer in the convolutional neural network with the quantum convolution layer to obtain a corresponding quantum convolutional neural network, wherein the convolutional neural network includes a fully connected layer, a convolutional layer and a final deconvolution layer;

[0012] The reconstructed image acquisition module acquires an undersampled K-space image and inputs the undersampled K-space image into a quantum convolutional neural network to obtain a corresponding fully sampled reconstructed image.

[0013] In an exemplary embodiment, the apparatus further includes, but is not limited to:

[0014] Function acquisition module, used to obtain hyperbolic tangent function;

[0015] An activation function determination module, configured to determine the hyperbolic tangent function as an activation function;

[0016] The replacement module obtains the mean square error during the quantum convolutional neural network training process and is used to replace the loss function with the mean square error.

[0017] In an exemplary embodiment, the apparatus further includes, but is not limited to:

[0018] The quantum convolutional neural network includes three layers of fully connected layers, two layers of quantum convolutional layers and one layer of deconvolutional layer.

[0019] In an exemplary embodiment, the device also includes but is not limited to: after the quantum convolution kernel receives the feature data extracted from the undersampled K-space image, the quantum convolution layer traverses the feature data according to the receptive field and the stride, and then uses the quantum convolution layer encoding module to encode the digitized feature data into the quantum state of the quantum bit, and the entangled quantum state is measured to obtain a feature map.

[0020] In an exemplary embodiment, the apparatus further includes but is not limited to: the stride size is 1, and the quantum convolution kernel adopts a size of 2x2.

[0021] According to one aspect of the present application, an electronic device includes at least one processor and at least one memory, wherein the memory stores computer-readable instructions; the computer-readable instructions are executed by one or more of the processors, so that the electronic device implements the magnetic resonance imaging method based on quantum convolutional neural network as described above.

[0022] According to one aspect of the present application, a storage medium stores computer-readable instructions thereon, and the computer-readable instructions are executed by one or more processors to implement the magnetic resonance imaging method based on quantum convolutional neural network as described above.

[0023] According to one aspect of the present application, a computer program product includes computer-readable instructions, which are stored in a storage medium. One or more processors of an electronic device read the computer-readable instructions from the storage medium, load and execute the computer-readable instructions, so that the electronic device implements the magnetic resonance imaging method based on quantum convolutional neural network as described above.

[0024] The beneficial effects of the technical solution provided by this application are: compared with traditional deep networks, the quantum convolution operation in the quantum convolutional neural network is a reversible operation, which maps the input state to the output state, and can achieve different transformations by adjusting the weights. The introduced quantum convolution layer utilizes the concepts of quantum mechanics, such as entanglement, superposition and interference, to achieve faster and more powerful processing capabilities, thereby speeding up the computing power during magnetic resonance imaging and improving the efficiency of magnetic resonance imaging.

[0025] In the above technical solution, a quantum convolution layer is obtained, and then the quantum convolution layer replaces the convolution layer in the convolutional neural network to obtain a corresponding quantum convolutional neural network. After the quantum convolutional neural network is determined, the obtained undersampled K-space image can be input into the quantum convolutional neural network to obtain the corresponding fully sampled reconstructed image. Through the entanglement, superposition and interference of the quantum convolution kernel, faster and more powerful processing capabilities are achieved, which can improve the imaging efficiency of magnetic resonance imaging, thereby effectively solving the problem of low imaging efficiency in related technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for describing the embodiments of the present application. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without inventive efforts.

[0027] FIG1 is a schematic diagram of an implementation environment according to the present application;

[0028] FIG2 is a flow chart showing a magnetic resonance imaging method based on a quantum convolutional neural network according to an exemplary embodiment;

[0029] FIG3 is a flowchart of steps S111 to S113 in a magnetic resonance imaging method based on a quantum convolutional neural network according to an exemplary embodiment;

[0030] FIG4 is a schematic diagram of a quantum convolution kernel in a magnetic resonance imaging method based on a quantum convolutional neural network;

[0031] FIG5 is a schematic diagram of data processing of a quantum convolution kernel in a magnetic resonance imaging method based on a quantum convolutional neural network;

[0032] FIG6 is a structural diagram of a quantum convolutional neural network in another magnetic resonance imaging method based on a quantum convolutional neural network according to an exemplary embodiment;

[0033] FIG7 is a block diagram of a magnetic resonance imaging device based on a quantum convolutional neural network according to an exemplary embodiment;

[0034] Fig. 8 is a structural block diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0035] The following describes in detail embodiments of the present application, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and are not to be construed as limiting the present application.

[0036] It will be understood by those skilled in the art that, unless expressly stated otherwise, the singular forms "a", "an", "said" and "the" used herein may also include the plural forms. It should be further understood that the term "comprising" used in the specification of the present disclosure refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. In addition, "connected" or "coupled" as used herein may include wireless connections or wireless couplings. The term "and / or" used herein includes all or any units and all combinations of one or more associated listed items.

[0037] As mentioned earlier, traditional deep neural networks require a large amount of training data to ensure imaging quality, and often require a large amount of training data to achieve good results. Although model-driven deep learning algorithms that learn reconstruction parameters require less training data, the imaging quality will not be particularly high. Related technologies still have the defect of low imaging efficiency.

[0038] To this end, the magnetic resonance imaging method based on quantum convolutional neural network provided by the present application can effectively improve the accuracy of magnetic resonance imaging based on quantum convolutional neural network. Accordingly, the magnetic resonance imaging method based on quantum convolutional neural network is applicable to a magnetic resonance imaging device based on quantum convolutional neural network, and the magnetic resonance imaging device based on quantum convolutional neural network can be deployed in electronic equipment.

[0039] In order to make the objectives, technical solutions and advantages of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0040] Figure 1 is a schematic diagram of an implementation environment for a magnetic resonance imaging method based on a quantum convolutional neural network. It should be noted that this implementation environment is merely an example adapted for the present invention and should not be construed as limiting the scope of application of the present invention. The implementation environment includes an acquisition terminal 110 and a server terminal 130.

[0041] Specifically, the acquisition end 110 can also be considered as an image acquisition device, including but not limited to electronic devices with shooting functions such as cameras, cameras, and video recorders.

[0042] Server 130 can be an electronic device such as a desktop computer, laptop computer, or server. It can also be a computer cluster consisting of multiple servers, or even a cloud computing center consisting of multiple servers. Server 130 is used to provide background services, such as, but not limited to, magnetic resonance imaging services.

[0043] A network communication connection is pre-established between the server 130 and the acquisition terminal 110 via a wired or wireless method, and data transmission between the server 130 and the acquisition terminal 110 is achieved via the network communication connection. The transmitted data includes but is not limited to: undersampled K-space images, etc.

[0044] In one application scenario, through interaction between the acquisition terminal 110 and the server 130 , the acquisition terminal 110 obtains an undersampled K-space image and uploads the undersampled K-space image to the server 130 to request the server 130 to provide magnetic resonance imaging services.

[0045] Please refer to Figure 2. An embodiment of the present application provides a magnetic resonance imaging method based on a quantum convolutional neural network. The method is applicable to an electronic device, which can be the server 130 in the implementation environment shown in Figure 1.

[0046] In the following method embodiments, for ease of description, the execution subject of each step of the method is taken as an electronic device as an example for illustration, but this does not constitute a specific limitation.

[0047] As shown in FIG2 , the method may include the following steps:

[0048] S100, obtain the quantum convolution layer;

[0049] Among them, the quantum convolution layer has some unique properties compared with the classical convolution layer. In the quantum convolution layer, quantum mechanical phenomena such as quantum superposition and quantum entanglement can be used to process input data.

[0050] S110, replacing the convolutional layer in the convolutional neural network with the quantum convolutional layer to obtain a corresponding quantum convolutional neural network;

[0051] Among them, the convolutional neural network includes a fully connected layer, a convolutional layer and a final deconvolution layer, and during the acquisition of the quantum convolutional neural network, when the convolutional layer of the convolutional neural network is replaced with a quantum convolutional layer, it should be pointed out here that the quantum convolutional neural network includes three fully connected layers, two quantum convolutional layers and one deconvolution layer.

[0052] In the process of obtaining the quantum convolutional neural network, it is also necessary to determine the parameters of the quantum convolutional neural network at the same time, as shown in Figure 3, so the method also includes:

[0053] S111, obtaining a hyperbolic tangent function;

[0054] S112, determining the hyperbolic tangent function as the activation function of the quantum convolutional neural network;

[0055] S113, during the quantum convolutional neural network training process, obtain the mean square error and replace the loss function with the mean square error.

[0056] In the above process, after the quantum convolutional neural network can be determined, the activation function is determined, and the loss function is replaced by the mean square error, the parameters of the quantum convolutional neural network can be determined.

[0057] S120, obtaining an undersampled K-space image, and inputting the undersampled K-space image into a quantum convolutional neural network to obtain a corresponding fully sampled reconstructed image.

[0058] After the quantum convolution kernel receives the feature data extracted from the undersampled K-space image, the quantum convolution layer traverses the feature data according to the receptive field and stride. It should be noted that in the embodiment of the present application, the stride size is 1 and the quantum convolution kernel size is 2×2.

[0059] The quantum convolution layer encoding module is then used to encode the digitized feature data into the quantum state of the quantum bit. The gate operation of the entanglement module then contains trainable weights. The entangled quantum state is measured to obtain a feature map, and the output of each quantum bit is individually formed into a feature map of a channel. In addition, a schematic diagram of the quantum convolution kernel is shown in FIG4 ; with reference to FIG5 , where F(x) is parameterized by the input feature, F(x) typically represents a quantum circuit or quantum operation that depends on the input data x, and G(w) is parameterized by the model weights. G(w) can be regarded as part of the quantum convolution layer, responsible for performing specific quantum transformations.

[0060] Therefore, the structure of the manifold approximation automatic transformation with the addition of a quantum convolutional neural network can be represented as Figure 6. This quantum convolutional neural network is consistent with the convolutional neural network and takes undersampled k-space data as input. The network structure consists of three fully connected layers, two quantum convolutional layers, and a final deconvolutional layer. The output of the network is a fully sampled reconstructed image. The hyperbolic tangent function is used as the activation function. Due to the introduction of the quantum convolutional layer, the loss function in the training process can be changed to the mean square error for training. The relevant expressions are as follows:

[0061] Where N represents the number of elements, Represents the output of the quantum convolutional layer.

[0062] The following are device embodiments of the present application, which can be used to perform the magnetic resonance imaging method based on quantum convolutional neural networks involved in this application. For details not disclosed in the device embodiments of the present application, please refer to the method embodiments of the magnetic resonance imaging method based on quantum convolutional neural networks involved in this application.

[0063] Referring to FIG7 , an embodiment of the present application provides a magnetic resonance imaging device based on a quantum convolutional neural network, including but not limited to:

[0064] A quantum convolution layer acquisition module 200 is used to acquire a quantum convolution layer;

[0065] A quantum convolutional neural network acquisition module 210 replaces the convolutional layer in the convolutional neural network with the quantum convolutional layer to obtain a corresponding quantum convolutional neural network, wherein the convolutional neural network includes a fully connected layer, a convolutional layer, and a final deconvolutional layer;

[0066] The reconstructed image acquisition module 220 acquires an undersampled K-space image and inputs the undersampled K-space image into a quantum convolutional neural network to acquire a corresponding fully sampled reconstructed image.

[0067] In an exemplary embodiment, the apparatus further includes, but is not limited to:

[0068] Function acquisition module 300, used to obtain a hyperbolic tangent function;

[0069] An activation function determination module 310, configured to determine the hyperbolic tangent function as an activation function;

[0070] The replacement module 320 obtains the mean square error during the quantum convolutional neural network training process and uses it to replace the loss function with the mean square error.

[0071] In an exemplary embodiment, the apparatus further includes, but is not limited to:

[0072] The quantum convolutional neural network includes three layers of fully connected layers, two layers of quantum convolutional layers and one layer of deconvolutional layer.

[0073] In an exemplary embodiment, the device also includes but is not limited to: after the quantum convolution kernel receives the feature data extracted from the undersampled K-space image, the quantum convolution layer traverses the feature data according to the receptive field and the stride, and then uses the quantum convolution layer encoding module to encode the digitized feature data into the quantum state of the quantum bit, and the entangled quantum state is measured to obtain a feature map.

[0074] In an exemplary embodiment, the apparatus further includes but is not limited to: the stride size is 1, and the quantum convolution kernel adopts a size of 2x2.

[0075] It should be noted that the magnetic resonance imaging method based on quantum convolutional neural network provided in the above embodiment

[0076] When the imaging device performs magnetic resonance imaging based on quantum convolutional neural network, only the division of the above-mentioned functional modules is used as an example to illustrate. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the magnetic resonance imaging device based on quantum convolutional neural network will be divided into different functional modules to complete all or part of the functions described above.

[0077] In addition, the magnetic resonance imaging device based on quantum convolutional neural network and the magnetic resonance imaging method based on quantum convolutional neural network provided in the above embodiments belong to the same concept, and the specific manner in which each module performs operations has been described in detail in the method embodiment and will not be repeated here.

[0078] Please refer to FIG8 . An embodiment of the present application provides an electronic device 4000 . The electronic device 400 may include a desktop computer, a laptop computer, a server, etc.

[0079] In FIG. 8 , the electronic device 4000 includes at least one processor 4001 and at least one memory 4003 .

[0080] Data exchange between the processor 4001 and the memory 4003 can be achieved through at least one communication bus 4002. The communication bus 4002 may include a path for transmitting data between the processor 4001 and the memory 4003. The communication bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The communication bus 4002 can be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is used in FIG8 , but this does not mean that there is only one bus or one type of bus.

[0081] Optionally, the electronic device 4000 may further include a transceiver 4004, which may be used for data exchange between the electronic device and other electronic devices, such as data transmission and / or data reception. It should be noted that in actual applications, the number of transceivers 4004 is not limited to one, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present application.

[0082] Processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It may implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 4001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like.

[0083] The memory 4003 can be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program instructions or codes in the form of instructions or data structures and can be accessed by the electronic device 400, but is not limited to these.

[0084] Computer-readable instructions are stored in the memory 4003 , and the processor 4001 can read the computer-readable instructions stored in the memory 4003 through the communication bus 4002 .

[0085] The computer-readable instructions are executed by one or more processors 4001 to implement the magnetic resonance imaging method based on quantum convolutional neural network in the above embodiments.

[0086] In addition, an embodiment of the present application provides a storage medium on which computer-readable instructions are stored. The computer-readable instructions are executed by one or more processors to implement the magnetic resonance imaging method based on quantum convolutional neural network as described above.

[0087] In an embodiment of the present application, a computer program product is provided. The computer program product includes computer-readable instructions, which are stored in a storage medium. One or more processors of an electronic device read the computer-readable instructions from the storage medium, load and execute the computer-readable instructions, so that the electronic device implements the magnetic resonance imaging method based on quantum convolutional neural network as described above.

[0088] Compared with traditional deep networks, the quantum convolution operation in quantum convolutional neural networks is a reversible operation that maps the input state to the output state, and can achieve different transformations by adjusting the weights. The introduced quantum convolution layer uses concepts of quantum mechanics such as entanglement, superposition, and interference to achieve faster and more powerful processing capabilities, thereby speeding up the computing power during magnetic resonance imaging and improving the efficiency of magnetic resonance imaging.

[0089] It should be understood that although the steps in the flowcharts of the accompanying drawings are shown in sequence as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some of the steps in the flowcharts of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

[0090] The above description is only part of the implementation methods of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A magnetic resonance imaging method based on quantum convolutional neural network, characterized in that: include: Get the quantum convolution layer; Replacing the convolutional layer in a convolutional neural network with the quantum convolutional layer to obtain a corresponding quantum convolutional neural network, wherein the convolutional neural network includes a fully connected layer, a convolutional layer, and a final deconvolutional layer; An undersampled K-space image is obtained, and the undersampled K-space image is input into a quantum convolutional neural network to obtain a corresponding fully sampled reconstructed image.

2. The method according to claim 1, wherein In the method for obtaining the corresponding quantum convolutional neural network, the method further includes: Get the hyperbolic tangent function; Using the hyperbolic tangent function as an activation function; During the quantum convolutional neural network training process, the mean square error is obtained and the loss function is replaced by the mean square error.

3. The method according to claim 1, wherein In the method for obtaining the corresponding quantum convolutional neural network, the quantum convolutional neural network includes three layers of the fully connected layers, two layers of the quantum convolutional layers and one layer of the deconvolutional layer.

4. The method according to claim 1, wherein In the method of inputting the undersampled K-space image into a quantum convolutional neural network to obtain a corresponding fully sampled reconstructed image, the method further includes: After the quantum convolution kernel receives the feature data extracted from the undersampled K-space image, the quantum convolution layer traverses the feature data according to the receptive field and the stride, and then uses the quantum convolution layer encoding module to encode the digitized feature data into the quantum state of the quantum bit. The entangled quantum state is measured to obtain a feature map.

5. The method according to claim 4, wherein In the method in which the quantum convolution layer traverses feature data according to the field size and the stride size, the method further includes: the stride size is 1, and the quantum convolution kernel adopts a size of 2x2.

6. A magnetic resonance imaging device based on quantum convolutional neural network, characterized in that: include: Quantum convolution layer acquisition module, used to obtain quantum convolution layer; A quantum convolutional neural network acquisition module replaces the convolutional layer in the convolutional neural network with the quantum convolution layer to obtain a corresponding quantum convolutional neural network, wherein the convolutional neural network includes a fully connected layer, a convolutional layer and a final deconvolution layer; The reconstructed image acquisition module acquires an undersampled K-space image and inputs the undersampled K-space image into a quantum convolutional neural network to obtain a corresponding fully sampled reconstructed image.

7. The device according to claim 6, characterized in that The device further comprises: Function acquisition module, used to obtain hyperbolic tangent function; An activation function determination module, configured to determine the hyperbolic tangent function as an activation function; The replacement module obtains the mean square error during the quantum convolutional neural network training process and is used to replace the loss function with the mean square error.

8. The device according to claim 7, wherein The device also includes: the quantum convolutional neural network includes three layers of fully connected layers, two layers of quantum convolutional layers and one layer of deconvolutional layer.

9. An electronic device, characterized in that: include: at least one processor and at least one memory, wherein: The memory has computer-readable instructions stored thereon; The computer-readable instructions are executed by one or more of the processors, so that the electronic device implements the magnetic resonance imaging method based on quantum convolutional neural network according to any one of claims 1 to 5.

10. A storage medium having computer-readable instructions stored thereon, characterized in that: The computer-readable instructions are executed by one or more processors to implement the magnetic resonance imaging method based on quantum convolutional neural network according to any one of claims 1 to 5.

Images