Feature extraction method and device, computer equipment and storage medium
Feature extraction and registration of functional imaging videos through deep learning feature extraction model, solving the problem of noise and motion displacement interference, improving the accuracy of registration and anti-interference ability.
Patent Information
- Application Number
- CN202510259054.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-06
- Publication Date
- 2025-07-04
AI Technical Summary
Functional imaging videos are susceptible to noise and motion displacement interference during the registration process, resulting in poor accuracy of data analysis results.
The feature extraction model of deep learning is used to extract feature of functional imaging videos. The feature extraction model includes sequentially connected feature extraction networks, feature compression filtering networks and feature output networks, through which target features are obtained and registered.
The anti-interference ability of functional imaging video registration is improved, and the extracted features are more conducive to frame registration and reduce the impact of noise interference.
Smart Images
Figure CN120259936A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of feature extraction, and particularly to a feature extraction method, apparatus, computer device, and storage medium. Background Art
[0002] Neurons are the basic functional units in the brain, and the electrical activities and chemical signal transmissions between them are the basis of brain information processing. Functional imaging can record large-scale neuronal activities and help scientists explore the complex functions of the neuronal brain.
[0003] However, such imaging technologies are easily interfered by noise and motion displacement, which will affect the accuracy of data analysis results and lead to poor registration effects for functional imaging videos. Summary of the Invention
[0004] Based on this, in view of the technical problem of poor registration effect of the existing functional imaging videos, a feature extraction method, apparatus, computer device, and storage medium are proposed.
[0005] In a first aspect, a feature extraction method is provided, and the method includes:
[0006] Obtain a functional imaging video;
[0007] Based on the functional imaging video and a trained feature extraction model, obtain target features, where the feature extraction model includes a feature extraction network, a feature compression and filtering network, and a feature output network connected in sequence;
[0008] Register the functional imaging video based on the target features.
[0009] In a second aspect, a feature extraction apparatus is provided, and the apparatus includes:
[0010] An obtaining module, configured to obtain a functional imaging video;
[0011] A feature extraction module, configured to obtain target features based on the functional imaging video and a trained feature extraction model, where the feature extraction model includes a feature extraction network, a feature compression and filtering network, and a feature output network connected in sequence;
[0012] A registration module, configured to register the functional imaging video based on the target features.
[0013] In a third aspect, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above feature extraction method are implemented.
[0014] Fourthly, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above feature extraction method are implemented.
[0015] For the feature extraction method proposed by the present invention, by obtaining a functional imaging video, and then based on the functional imaging video and a trained feature extraction model, target features are obtained. Among them, the feature extraction model includes a feature extraction network, a feature compression and filtering network, and a feature output network connected in sequence. Finally, the functional imaging video is registered based on the target features. The present invention uses a deep learning-based feature extraction model to extract features from the functional imaging video, and then uses the extracted features to perform frame-to-frame registration on the functional imaging video. The extracted features are more conducive to registration and are less affected by noise interference, improving the anti-interference ability of the registration method. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.
[0017] Among them:
[0018] Figure 1 It is an application environment diagram of the feature extraction method in an embodiment;
[0019] Figure 2 It is a flowchart of the feature extraction method in an embodiment;
[0020] Figure 3 It is a schematic diagram of the network structure of the feature extraction model of the feature extraction method in an embodiment;
[0021] Figure 4 It is an output result diagram of the feature extraction model of the feature extraction method in an embodiment;
[0022] Figure 5 It is a structural block diagram of a feature extraction device in an embodiment;
[0023] Figure 6 It is a structural block diagram of a computer device in an embodiment;
[0024] Figure 7 It is a structural block diagram of a computer device in another embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the technical field to which this application belongs; the terms used in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above drawings are intended to cover non-exclusive inclusion. The terms "first", "second", etc. in the specification and claims of this application or the above drawings are used to distinguish different objects and not to describe a specific order.
[0026] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in connection with the embodiments can be included in at least one embodiment of this application. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.
[0027] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the protection scope of the present invention.
[0028] The feature extraction method provided by the embodiments of the present invention can be applied to an application environment such as Figure 1 where the client 110 communicates with the server 120 through a network. The server 120 can receive and obtain a functional imaging video through the client 110, and then the server 120 obtains target features based on the functional imaging video and a trained feature extraction model, where the feature extraction model includes a feature extraction network, a feature compression and filtering network, and a feature output network connected in sequence, and finally registers the functional imaging video based on the target features. By using a deep learning-based feature extraction model to extract features from the functional imaging video and then using the extracted features to perform frame-to-frame registration on the functional imaging video, the extracted features are more conducive to registration and are less affected by noise, improving the anti-interference ability of the registration method. Among them, the client 110 can be, but is not limited to, various personal computers, laptop computers, smart phones, tablet computers, and portable wearable devices. The server 120 can be implemented by an independent server or a server cluster composed of multiple servers. The present invention will be described in detail below through specific embodiments.
[0029] Please refer to Figure 2 as shown Figure 2A flowchart of a feature extraction method provided by an embodiment of the present invention, including the following steps:
[0030] Step S101: Obtain a functional imaging video;
[0031] Wherein, the functional imaging video refers to a video that records large-scale neuronal activities.
[0032] Step S102: Based on the functional imaging video and the trained feature extraction model, obtain target features, wherein the feature extraction model includes a feature extraction network, a feature compression and filtering network, and a feature output network connected in sequence;
[0033] Step S103: Register the functional imaging video based on the target features.
[0034] In this embodiment, by using a feature extraction model of deep learning to extract features from a functional imaging video, and then using the extracted features to perform frame-to-frame registration on the functional imaging video, the extracted features are more conducive to registration and are less affected by noise interference, improving the anti-interference ability of the registration method.
[0035] In one embodiment, referring to Figure 3 , the feature extraction model includes a feature extraction network, a feature compression and filtering network, and a feature output network. The output end of the feature extraction network is connected to the input end of the feature compression and filtering network, and the output end of the feature compression and filtering network is connected to the input end of the feature output network.
[0036] In one embodiment, referring to Figure 3 , the feature extraction network adopts a Unet network. For example, Figure 3 the feature extraction Unet therein refers to the Unet network. As an example, the Unet network is composed of a 2D encoder module, a 2D decoder module, and three skip connections from the encoder module to the decoder module. In the 2D encoder module, there are three encoder blocks. Each encoding block contains 3 convolutional layers, 3 normalization layers, and 3 activation layers. Each convolutional layer is followed by a normalization layer (BatchNorm2d) and an activation layer (LeakyReLU). The stride of the third convolutional layer is 2 to achieve the effect of downsampling. In the decoder module, there are three decoder blocks. Each encoding block contains 2 convolutional layers and a transposed convolutional layer. The convolutional layer or the transposed convolutional layer is followed by a normalization (BatchNorm2d) and an activation layer (LeakyReLU). The stride of the transposed convolutional layer is 2 to achieve the effect of upsampling. The skip connections link them by connecting the feature maps of low-level features and high-level features.
[0037] In one embodiment, referring to Figure 3, the feature compression and filtering network includes n encoding modules and n decoding modules connected in sequence, where n is an integer greater than or equal to 1. Among them, the encoding module includes a first convolutional layer, a first normalization layer, a first activation layer, a second convolutional layer, a second normalization layer, a second activation layer, a third convolutional layer, a third normalization layer, and a third activation layer connected in sequence. The decoding module includes a fourth convolutional layer, a fourth normalization layer, a fourth activation layer, a fifth convolutional layer, a fifth normalization layer, a fifth activation layer, a transposed convolutional layer, a sixth normalization layer, and a sixth activation layer connected in sequence. As an example, n can be 4. The number of intermediate features in the feature compression and filtering network is also much smaller than that of the previous Unet network, which can form a bottleneck to achieve the purpose of filtering unnecessary information.
[0038] In one embodiment, referring to Figure 3 , the feature output network includes a seventh convolutional layer, a seventh normalization layer, a seventh activation layer, an eighth convolutional layer, an eighth normalization layer, an eighth activation layer, a ninth convolutional layer, a ninth normalization layer, and a ninth activation layer connected in sequence. The function of the feature output network is to output features with a specified number of channels and further screen and output information more suitable for registration.
[0039] In one embodiment, referring to Figure 3 , the feature extraction model further includes a feature reconstruction network. The output channel of the feature reconstruction network is 1, and the model structure of the feature reconstruction network is the same as that of the feature output network. The output of the feature reconstruction network is used to calculate the loss function. For example, the output of the feature reconstruction network and the template are used to calculate MSELoss as warmstart.
[0040] In one embodiment, referring to Figure 4 , through experimental verification, the output result of the feature extraction model using the functional imaging video is less affected by noise, and the network loss drops faster.
[0041] Please refer to Figure 5 As shown, in one embodiment, a feature extraction device is provided. The device includes:
[0042] An acquisition module for acquiring a functional imaging video;
[0043] A feature extraction module for obtaining target features based on the functional imaging video and the trained feature extraction model, where the feature extraction model includes a feature extraction network, a feature compression and filtering network, and a feature output network connected in sequence;
[0044] A registration module for registering the functional imaging video based on the target features.
[0045] In one embodiment, the feature extraction model includes a feature extraction network, a feature compression and filtering network, and a feature output network. The output end of the feature extraction network is connected to the input end of the feature compression and filtering network, and the output end of the feature compression and filtering network is connected to the input end of the feature output network.
[0046] In one embodiment, the feature extraction network adopts a Unet network.
[0047] In one embodiment, the feature compression and filtering network includes n encoding modules and n decoding modules connected in sequence, where n is an integer greater than or equal to 1. Among them, the encoding module includes a first convolutional layer, a first normalization layer, a first activation layer, a second convolutional layer, a second normalization layer, a second activation layer, a third convolutional layer, a third normalization layer, and a third activation layer connected in sequence. The decoding module includes a fourth convolutional layer, a fourth normalization layer, a fourth activation layer, a fifth convolutional layer, a fifth normalization layer, a fifth activation layer, a deconvolutional layer, a sixth normalization layer, and a sixth activation layer connected in sequence. As an example, the feature compression and filtering network includes encoding module 1, encoding module 2, encoding module 3, encoding module 4, decoding module 1, decoding module 2, decoding module 3, and decoding module 4 connected in sequence.
[0048] In one embodiment, n is 4.
[0049] In one embodiment, the feature output network includes a seventh convolutional layer, a seventh normalization layer, a seventh activation layer, an eighth convolutional layer, an eighth normalization layer, an eighth activation layer, a ninth convolutional layer, a ninth normalization layer, and a ninth activation layer connected in sequence.
[0050] In one embodiment, the feature extraction model further includes a feature reconstruction network. The output channel of the feature reconstruction network is 1. The model structure of the feature reconstruction network is the same as that of the feature output network. The output of the feature reconstruction network is used to calculate the loss function.
[0051] In one embodiment, a computer device is provided. The computer device can be a server, and its internal structure diagram can be as Figure 6 shown. The computer device includes a processor, a memory, a network interface, and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external client through a network connection. When the computer program is executed by the processor, it realizes the functions or steps on the server side of a feature extraction method.
[0052] In one embodiment, a computer device is provided. The computer device can be a client, and its internal structure diagram can be as shown in Figure 7 The computer device includes a processor, a memory, a network interface, a display screen, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external server through a network connection. When the computer program is executed by the processor, it realizes the functions or steps on the client side of a feature extraction method.
[0053] In one embodiment, a computer device is proposed, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are realized:
[0054] Obtain a functional imaging video;
[0055] Based on the functional imaging video and a trained feature extraction model, obtain target features, where the feature extraction model includes a feature extraction network, a feature compression and filtering network, and a feature output network connected in sequence;
[0056] Register the functional imaging video based on the target features.
[0057] In the present invention, a deep learning-based feature extraction model is used to extract features from a functional imaging video, and then the extracted features are used to register the functional imaging video frame by frame. The extracted features are more conducive to registration and are less affected by noise, improving the anti-interference ability of the registration method.
[0058] In one embodiment, a computer-readable storage medium is proposed. The computer-readable storage medium stores a computer program. When the computer program is executed by the processor, the following steps are realized:
[0059] Obtain a functional imaging video;
[0060] Based on the functional imaging video and a trained feature extraction model, obtain target features, where the feature extraction model includes a feature extraction network, a feature compression and filtering network, and a feature output network connected in sequence;
[0061] Register the functional imaging video based on the target features.
[0062] In the present invention, a feature extraction model of deep learning is adopted to extract features from a functional imaging video, and then the extracted features are used to perform frame-to-frame registration on the functional imaging video. The extracted features are more conducive to registration and are less affected by noise, improving the anti-interference ability of the registration method.
[0063] It should be noted that for the functions or steps that can be realized by the above-mentioned computer-readable storage medium or computer device, reference can be made to the relevant descriptions on the server side and the client side in the foregoing method embodiments. To avoid repetition, they will not be described in detail here.
[0064] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above method embodiments. Among them, any reference to a memory, storage, database, or other medium used in the various embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.
[0065] Those skilled in the art can clearly understand that for the convenience and brevity of description, only the above-mentioned division of each functional unit and module is used as an example. In actual applications, the above functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above.
[0066] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included within the protection scope of the present invention.
Claims
1. A feature extraction method, characterized in that, The described feature extraction method includes: Obtain a functional imaging video; Based on the functional imaging video and a trained feature extraction model, obtain target features, where the feature extraction model includes a feature extraction network, a feature compression and filtering network, and a feature output network connected in sequence; Register the functional imaging video based on the target features.
2. The feature extraction method according to claim 1, wherein The feature extraction model includes a feature extraction network, a feature compression and filtering network, and a feature output network. The output end of the feature extraction network is connected to the input end of the feature compression and filtering network, and the output end of the feature compression and filtering network is connected to the input end of the feature output network.
3. The feature extraction method according to claim 2, wherein The feature extraction network adopts a Unet network.
4. The feature extraction method according to any one of claims 2 or 3, characterized in that, The feature compression and filtering network includes n encoding modules and n decoding modules connected in sequence, where n is an integer greater than or equal to 1. Among them, the encoding module includes a first convolutional layer, a first normalization layer, a first activation layer, a second convolutional layer, a second normalization layer, a second activation layer, a third convolutional layer, a third normalization layer, and a third activation layer connected in sequence. The decoding module includes a fourth convolutional layer, a fourth normalization layer, a fourth activation layer, a fifth convolutional layer, a fifth normalization layer, a fifth activation layer, a deconvolutional layer, a sixth normalization layer, and a sixth activation layer connected in sequence.
5. The feature extraction method according to claim 4, wherein n is 4.
6. The feature extraction method according to claim 5, wherein The feature output network includes a seventh convolutional layer, a seventh normalization layer, a seventh activation layer, an eighth convolutional layer, an eighth normalization layer, an eighth activation layer, a ninth convolutional layer, a ninth normalization layer, and a ninth activation layer connected in sequence.
7. The feature extraction method according to claim 6, wherein The feature extraction model further includes a feature reconstruction network. The output channel of the feature reconstruction network is 1. The model structure of the feature reconstruction network is the same as that of the feature output network. The output of the feature reconstruction network is used to calculate the loss function.
8. A feature extraction device, characterized in that, The described feature extraction device includes: An acquisition module for acquiring a functional imaging video; A feature extraction module for obtaining target features based on the functional imaging video and a trained feature extraction model, where the feature extraction model includes a feature extraction network, a feature compression and filtering network, and a feature output network connected in sequence; A registration module for registering the functional imaging video based on the target features.
9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the feature extraction method according to any one of claims 1 to 7.
10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the feature extraction method according to any one of claims 1 to 7.