Portable wearable heterogeneous edge artificial intelligence electroencephalogram collection device and method

Abstract

The invention provides a portable wearable heterogeneous edge artificial intelligence electroencephalogram collection device and method. The electroencephalogram acquisition equipment comprises a central processing unit, at least two data processing units with different computing power architectures and an electroencephalogram analog front-end unit, wherein the central processing unit and the data processing units are artificial intelligence computing power units; and the central processing unit is used for acquiring equipment configuration information, acquiring electroencephalogram data from the electroencephalogram analog front-end unit, and allocating preprocessing and / or data analysis of the electroencephalogram data to the corresponding artificial intelligence computing power unit based on a user preset program and the equipment configuration information to obtain a target data processing result. According to the embodiment of the invention, through at least three artificial intelligence computing power units, data processing work such as preprocessing and data analysis of electroencephalogram data can be flexibly configured, and the edge heterogeneous computing power and the data processing capability of the portable wearable electroencephalogram acquisition equipment are effectively enhanced, so that the reliability, the real-time performance and the flexibility of the electroencephalogram equipment are enhanced; and the application scene is expanded.

Description

Technical Field

[0001] This application relates to the technical field of electroencephalogram (EEG) signal acquisition and processing, and in particular to a portable wearable heterogeneous edge AI EEG acquisition device and method. Background Technology

[0002] Electroencephalogram (EEG) signals are electrical potential signals manifested on the scalp surface after the postsynaptic potentials of pyramidal neurons in the cerebral cortex are superimposed within local neural networks. They reflect human brain neural activity. For example, during limb movement, desynchronization of neural rhythms occurs in the local neural networks controlling the corresponding parts of the body in the motor and sensory cortex, manifested as a decrease in signal power within a specific frequency range. Similarly, when a person notices a flickering sound at a specific frequency, the EEG signals in the primary visual cortex will also show corresponding frequency fluctuations. EEG technology has become an important tool for scientific research and medical diagnosis.

[0003] The acquisition of electroencephalogram (EEG) signals relies on EEG acquisition devices. Currently, commercially available EEG amplifiers are commonly used in laboratories and medical settings internationally. These devices have simple functions, providing only multi-channel EEG signal sampling, filtering, digitization, and external event labeling. The acquired data is typically transmitted via USB using a proprietary, closed protocol to dedicated data acquisition software on a host computer. This software then saves the data as computer files in a specific format for subsequent scientific research or medical diagnostic analysis.

[0004] However, most existing portable wearable heterogeneous edge AI EEG acquisition devices are based on the traditional model of "lower-level computer collecting EEG data and transmitting it to upper-level computer for processing". Since existing portable wearable heterogeneous edge AI EEG acquisition devices cannot process EEG data, the EEG data needs to be transmitted to external devices such as upper-level computers for processing, which limits the further application of EEG technology in neuroscience research, brain-computer interface, intelligent diagnosis, sports and mental rehabilitation and other fields. Summary of the Invention

[0005] In view of this, the present application provides a portable wearable heterogeneous edge AI EEG acquisition device and method to solve the problem that in the prior art, portable wearable heterogeneous edge AI EEG acquisition devices cannot process EEG data and need to transmit EEG data to external devices such as host computers for processing.

[0006] The first aspect of this application provides a portable wearable heterogeneous edge AI EEG acquisition device, including: a central processing unit, at least two data processing units with different computing power architectures and an EEG simulation front-end unit, wherein the central processing unit and the data processing units are both AI computing power units; The central processing unit is used to acquire device configuration information and EEG data from the EEG simulation front-end unit. Based on the user-preset program and device configuration information, it allocates the data preprocessing and / or data analysis of the EEG data to the corresponding artificial intelligence computing power unit. The EEG data is then preprocessed and / or analyzed by each artificial intelligence computing power unit to obtain the target data processing result.

[0007] In one possible implementation, at least two data processing units include: a graphics processor and a programmable gate array (PGA) circuit. The graphics processing unit is used to perform data preprocessing and / or data analysis on EEG data based on deep neural networks to obtain the first data processing result; Programmable gate array circuits are used to perform brain-like computing-based data preprocessing and / or data analysis on EEG data to obtain a second data processing result; The central processing unit is used to perform data preprocessing and / or data analysis on EEG data to obtain a third data processing result; and to obtain a target data processing result based on at least one of the first data processing result, the second data processing result, and the third data processing result.

[0008] In one possible implementation, the portable wearable heterogeneous edge AI EEG acquisition device also includes at least two of the following: A graphics processing unit is used to perform at least one of data preprocessing and data analysis on EEG data based on a loaded deep neural network model to obtain a first data processing result; the data preprocessing includes noise reduction processing, and the data analysis includes at least one of classification prediction and regression prediction; Programmable gate array circuits are used in loading-based neuromorphic computing circuits to perform at least one of data preprocessing and data analysis on EEG data to obtain a second data processing result; data preprocessing includes noise reduction processing, and data analysis includes at least one of classification prediction and regression prediction; The central processing unit is used to perform at least one of data bad channel repair and signal rereference on the EEG data to obtain a third data processing result; data preprocessing includes at least one of data bad channel repair and signal rereference.

[0009] In one possible implementation, the portable wearable heterogeneous edge AI EEG acquisition device also includes at least one of the following: A general-purpose input / output interface is used to output the target data processing result to an external device, so that the external device can perform corresponding operations based on the target data processing result; the target data processing result is a control command signal; The wireless communication module is used to send the target data processing results to other terminal devices.

[0010] In one possible implementation, the central processing unit is also used to analyze the target data processing results, determine the output channel of the target data processing results, and output the target data processing results from a general-purpose input / output interface or a wireless communication module based on the output channel.

[0011] In one possible implementation, the central processing unit is also used to acquire infrared light synchronization signals, perform event tagging on the EEG data based on the infrared light synchronization signals, and time-align the EEG data with other data that have the same event tags based on the event tags of the EEG data.

[0012] In one possible implementation, the EEG simulation front-end unit is electrically connected to the central processing unit. The EEG simulation front-end unit is used to acquire EEG signals, perform predetermined operations on the EEG signals, and obtain EEG data. The EEG data is a digital signal, and the predetermined operations include converting the analog signal into a digital signal.

[0013] In one possible implementation, the portable wearable heterogeneous edge AI EEG acquisition device also includes several operation buttons for users to set device configuration information; Device configuration information includes at least one of the following: Data processing window length, preprocessing parameters, and the AI ​​model to be loaded; The preprocessing parameters include digital filter information, sampling rate information, bad channel repair information, and rereference information.

[0014] The second aspect of this application provides an electroencephalogram (EEG) acquisition system, including: an EEG cap assembly, and the portable wearable heterogeneous edge AI EEG acquisition device of the first aspect; The EEG cap assembly includes multiple acquisition electrodes for collecting the user's electroencephalogram (EEG) signals; The EEG cap assembly is used to connect to a portable wearable heterogeneous edge AI EEG acquisition device to transmit the EEG signals from each acquisition electrode to the portable wearable heterogeneous edge AI EEG acquisition device.

[0015] A third aspect of this application provides a method for processing electroencephalogram (EEG) data, applied to the portable wearable heterogeneous edge AI EEG acquisition device of the first aspect, the method comprising: Acquire device configuration information and obtain EEG data from the EEG simulation front-end unit; Based on user-preset programs and device configuration information, the data preprocessing and / or data analysis of EEG data are assigned to the corresponding artificial intelligence computing units; The EEG data is preprocessed and / or analyzed by various artificial intelligence computing units to obtain the target data processing results.

[0016] Compared with the prior art, the embodiments of this application have at least the following technical effects: The portable wearable heterogeneous edge AI EEG acquisition device of the first aspect of this application includes a central processing unit (CPU), at least two data processing units with different computing power architectures, and an EEG simulation front-end unit. Both the CPU and the data processing units are AI computing power units, forming a heterogeneous edge AI computing power architecture to accelerate computation for different types of AI models. Simultaneously, the CPU can acquire device configuration information and EEG data from the EEG simulation front-end unit. Based on user-preset programs and device configuration information, it allocates the preprocessing and / or data analysis of the EEG data to the corresponding AI computing power units. Each AI computing power unit then performs preprocessing and / or data analysis on the EEG data to obtain the target data processing result. Therefore, the portable wearable heterogeneous edge AI EEG acquisition device of this application, through at least three types of AI computing power units, enables flexible configuration of EEG data preprocessing and data analysis, breaking through the traditional brain-computer interface application model of "lower-level machine acquiring EEG data and transmitting it to upper-level machine for processing." This effectively enhances the reliability, real-time performance, and flexibility of EEG technology, expanding its application scenarios.

[0017] It is understood that the beneficial effects of the second and third aspects mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the structure of an electroencephalogram (EEG) acquisition system provided in an embodiment of this application; Figure 2 This is a schematic diagram of the framework of a portable wearable heterogeneous edge AI EEG acquisition device provided in an embodiment of this application; Figure 3 This is a schematic diagram of the framework of another portable wearable heterogeneous edge AI EEG acquisition device provided in the embodiments of this application; Figure 4 This is a schematic diagram of one side of the circuit board of a portable wearable heterogeneous edge AI EEG acquisition device provided in an embodiment of this application; Figure 5This is a schematic diagram of the other side of the circuit board of a portable wearable heterogeneous edge AI EEG acquisition device provided in an embodiment of this application; Figure 6 This is a flowchart of a method for processing electroencephalogram (EEG) data provided in an embodiment of this application.

[0020] Figure label: 1-EEG cap assembly; 2-EEG acquisition device, 21-Central processing unit, 22-Graphics processor, 23-Programmable gate array circuit, 24-Input / output interface, 25-Communication module, 26-EEG simulation front-end unit, 27-Operation button; 101-EEG cap, 102-Collection electrodes, 103-EEG signal transmission line, 104-EEG cap connector, 105-EEG cap socket, 106-Outer shell, 107-Heat dissipation holes, 108-First operation button, 109-Second operation button, 110-Third operation button, 111-Fourth operation button, 112-Fifth operation button, 113-Sixth operation button, 115-Infrared light synchronization receiver, 116-Display; 201-First Core Board, 202-First Core Board Connector, 203-Circuit Board Mounting Hole, 204-Second Core Board Connector, 205-Second Core Board, 207-Power Supply / Charging Port, 208-Supporting Components, 209-EEG Simulation Front-End Unit Connector, 211-Battery, 212-First Button Component, 213-Second Button Component, 214-Third Button Component, 215-Fourth Button Component, 216-Fifth Button Component, 217-Sixth Button Component, 218-Gigabit Wireless Network Card, 219-Gigabit Wireless Network Card Connector, 220-Solid State Drive Connector, 221-Solid State Drive, 224-Display Connector, 225-Display Cable. Detailed Implementation

[0021] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0022] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0023] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0024] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0025] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0026] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0027] Research has revealed the following problems with existing portable wearable heterogeneous edge AI EEG acquisition devices for collecting EEG signals: 1. Existing technologies rely on communication with and software from a host computer, and cannot complete signal analysis within portable, wearable, heterogeneous edge AI EEG acquisition devices.

[0028] Currently, the processing of EEG signals involves using a simple embedded microcontroller to control analog amplification circuits and analog-to-digital conversion circuits. The read EEG signals are then forwarded to a host computer system via a wired or wireless communication port, where further processing is performed in the host computer software. Within this traditional framework of "lower-level computer collecting EEG data and transmitting it to the host computer for processing," portable wearable heterogeneous edge AI EEG acquisition devices do not perform EEG signal analysis. This makes application systems based on these devices overly reliant on the communication and computing power of the host computer, resulting in lower reliability, portability, and higher complexity.

[0029] 2. Existing technologies lack heterogeneous computing power suitable for edge deployment of artificial intelligence models. Some portable wearable heterogeneous edge AI EEG acquisition devices only perform simple data forwarding or low-level digital filtering. These devices lack internal AI computing power and cannot deploy AI models for more complex signal analysis. Currently, AI technology has developed rapidly, greatly improving the preprocessing and data analysis of EEG signals. AI models rely on different types of computing architectures: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and Neural Processing Unit (NPU). Existing portable wearable heterogeneous edge AI EEG acquisition devices have built-in edge heterogeneous AI computing platforms, which hinder the application of edge AI technology in various EEG-based systems, resulting in lower system real-time performance, reliability, and stability, making it difficult to work independently offline.

[0030] The portable wearable heterogeneous edge AI EEG acquisition device and method provided in this application are intended to solve the above-mentioned technical problems of the prior art.

[0031] The technical solution of this application and how it solves the above-mentioned technical problems are described in detail below with specific embodiments. It should be noted that the following embodiments can be referenced, borrowed, or combined with each other, and the same terms, similar features, and similar implementation steps in different embodiments will not be described again.

[0032] See Figure 1 As shown in the diagram, this application provides a schematic diagram of the structure of an electroencephalogram (EEG) acquisition system. Figure 1 As shown, the EEG acquisition system includes: an EEG cap assembly 1, and a portable wearable heterogeneous edge AI EEG acquisition device 2 according to an embodiment of this application.

[0033] The EEG cap assembly 1 includes multiple acquisition electrodes 102 for acquiring the user's EEG signals.

[0034] The EEG cap assembly 1 is used to connect to the portable wearable heterogeneous edge AI EEG acquisition device 2 and send the EEG signals of each acquisition electrode 102 to the portable wearable heterogeneous edge AI EEG acquisition device 2.

[0035] See Figure 1 As shown, the EEG cap assembly 1 includes an EEG cap 101, multiple acquisition electrodes 102, EEG signal transmission lines 103, and an EEG cap connector 104.

[0036] Optionally, the EEG cap component 1 can be configured with 32-64 EEG signal channels depending on the application scenario requirements.

[0037] See Figure 2 As shown, this application provides a schematic diagram of the framework of a portable wearable heterogeneous edge AI EEG acquisition device 2. Figure 2 As shown, this application embodiment provides a portable wearable heterogeneous edge AI EEG acquisition device 2, including: a central processing unit 21, at least two data processing units with different computing power architectures and an EEG simulation front-end unit 26, wherein the central processing unit 21 and the data processing units are both AI computing power units.

[0038] The central processing unit 21 is used to acquire device configuration information and acquire EEG data from the EEG simulation front-end unit 26. Based on the user preset program and device configuration information, it allocates the data preprocessing and / or data analysis of the EEG data to the corresponding artificial intelligence computing power unit, and the EEG data is preprocessed and / or analyzed by each artificial intelligence computing power unit to obtain the target data processing result.

[0039] Among them, device configuration information refers to the device usage parameters set by the user.

[0040] The portable wearable heterogeneous edge AI EEG acquisition device 2 of this application embodiment includes a central processing unit 21, at least two data processing units with different computing power architectures, and an EEG simulation front-end unit 26. Both the central processing unit 21 and the data processing units are AI computing power units, forming a heterogeneous edge AI computing power architecture. Simultaneously, the central processing unit 21 can acquire device configuration information and EEG data from the EEG simulation front-end unit 26. Based on user-preset programs and device configuration information, it allocates data preprocessing and / or data analysis of the EEG data to the corresponding AI computing power units. Each AI computing power unit then performs data preprocessing and / or data analysis on the EEG data to obtain the target data processing result.

[0041] Therefore, the portable wearable heterogeneous edge AI EEG acquisition device 2 of this application embodiment realizes data processing work such as preprocessing and data analysis of EEG data that can be flexibly configured through at least three AI computing power units. It breaks through the traditional brain-computer interface application mode of "lower-level machine collects EEG data and transmits it to upper-level machine for processing", effectively enhancing the reliability, real-time performance and flexibility of EEG technology and expanding its application scenarios.

[0042] This application relates to the field of EEG signal acquisition and processing, particularly to edge heterogeneous intelligent wearable EEG acquisition systems applied in fields such as neuroscience, brain-computer interfaces, and sports and mental rehabilitation.

[0043] See Figure 2 As shown, at least two data processing units include: a graphics processor 22 and a programmable gate array circuit 23; that is, the portable wearable heterogeneous edge AI EEG acquisition device 2 of this application embodiment includes: a central processing unit 21, a graphics processor 22 and a programmable gate array circuit 23.

[0044] The graphics processor 22 is used to perform data preprocessing and / or data analysis on EEG data based on deep neural networks to obtain a first data processing result; The programmable gate array circuit 23 is used to perform brain-like computing-based data preprocessing and / or data analysis on EEG data to obtain a second data processing result; The central processing unit 21 is used to perform data preprocessing and / or data analysis on the EEG data to obtain a third data processing result; and to obtain a target data processing result based on at least one of the first data processing result, the second data processing result, and the third data processing result.

[0045] Optionally, the central processing unit 21 performs data preprocessing and / or data analysis on the EEG data based on process control, numerical calculation, and logical analysis.

[0046] This application provides a portable wearable heterogeneous edge AI EEG acquisition device 2 based on a heterogeneous edge AI computing power architecture. The portable wearable heterogeneous edge AI EEG acquisition device 2 of this application integrates an embedded central processing unit (CPU), a graphics processing unit (GPU), and a field-programmable gate array (FPGA), enabling it to perform flexibly configurable preprocessing and analysis of EEG signals based on edge AI computing power within the acquisition device.

[0047] In some embodiments, the portable wearable heterogeneous edge AI EEG acquisition device 2 further includes at least one of the following: The graphics processor 22 is used to perform at least one of data preprocessing and data analysis on EEG data based on a loaded deep neural network model to obtain a first data processing result; the data preprocessing includes noise reduction processing, and the data analysis includes at least one of classification prediction and regression prediction.

[0048] The programmable gate array circuit 23 is used for loading-based neuromorphic computing circuits to perform at least one of data preprocessing and data analysis on EEG data to obtain a second data processing result; the data preprocessing includes noise reduction processing, and the data analysis includes at least one of classification prediction and regression prediction.

[0049] The central processing unit 21 is used to perform at least one of data bad channel repair and signal rereference on the EEG data to obtain a third data processing result; the data preprocessing includes at least one of data bad channel repair and signal rereference.

[0050] Optionally, the graphics processor 22, programmable gate array circuit 23, and central processing unit 21 in this embodiment of the application can preprocess and / or analyze EEG data based on user-preset programs and device configuration information.

[0051] Optionally, taking the deep neural network-based noise reduction processing of the graphics processor 22 as an example: Electroencephalogram (EEG) signals are often interfered with by physiological signals such as electrooculogram (EOG), motion noise, and electromyography (EMG). The graphics processor 22 can load a trained deep neural network noise reduction model to estimate the noise-free EEG components from noisy EEG data, or to estimate the noise components from noisy EEG data and subtract them from the original EEG signal.

[0052] Optionally, taking the programmable gate array circuit 23 as an example of neuromorphic computing-based noise reduction processing: Electroencephalogram (EEG) signals are often interfered with by physiological signals such as electrooculogram (EOG), motion noise, and electromyography (EMG). The programmable gate array (PGA) circuit 23 can be loaded with a pre-designed neuromorphic computing noise reduction circuit to estimate the noise-free EEG components from noisy EEG data, or to estimate the noise components from noisy EEG data and subtract them from the original EEG signal.

[0053] Optionally, taking the graphics processor 22 performing classification prediction based on a deep neural network as an example: Electroencephalogram (EEG) signals reflect the functional state of the human brain. For example, different emotional states produce different EEG responses. The graphics processor 22 can load a trained deep neural network classification model to identify the wearer's emotional state, and the resulting target data classification results are output through a general input / output interface 24 to control devices such as emotional embodied intelligent robots.

[0054] Optionally, taking the programmable gate array circuit 23 for classification prediction based on neuromorphic computing as an example: Electroencephalogram (EEG) signals reflect the functional state of the human brain. For example, different emotional states produce different EEG responses. The programmable gate array (PGA) circuit 23 can be loaded with a trained neuromorphic computing circuit classification model to identify the wearer's emotional state. The resulting target data classification processing results are output through the general input / output interface 24 to control devices such as emotional embodied intelligent robots.

[0055] Optionally, taking the graphics processor 22 performing regression prediction based on a deep neural network as an example: Electroencephalography (EEG) reflects the functional state of the human brain. For example, limb motor imagery includes limb movement planning and control information, which can help patients with motor dysfunction control prostheses and perform assisted rehabilitation. The graphics processor 22 can load a trained deep neural network regression decoding model to predict limb movement signals from EEG to control prostheses or perform assisted rehabilitation.

[0056] Optionally, taking the programmable gate array circuit 23 for regression prediction based on neuromorphic computing as an example: Electroencephalography (EEG) reflects the functional state of the human brain. For example, limb motor imagery includes limb movement planning and control information, which can help patients with motor dysfunction control prostheses and perform assisted rehabilitation. Programmable gate array circuit 23 can be loaded with a trained brain-like computational regression decoding model to predict limb movement signals from EEG in order to control prostheses or perform assisted rehabilitation.

[0057] Optionally, taking the central processing unit 21 performing data corruption channel repair on EEG data as an example: During EEG acquisition, some channel data may remain unusable due to unexpected and continuous large-scale noise interference or deterioration of the contact impedance of the channel electrodes. To ensure the structure of the acquired data for subsequent analysis, the central processing unit 21 can perform spatial interpolation, a numerical calculation, based on the data of neighboring channels to repair the data of that channel.

[0058] Optionally, taking the signal rereference of EEG data by the central processing unit 21 as an example: The essence of EEG acquisition is to collect the voltage between two electrodes. Modern devices with multi-channel EEG acquisition capabilities typically employ a "reference layout" electrode arrangement. Specifically, one electrode on the scalp serves as the reference electrode, while the remaining electrodes act as acquisition electrodes, collecting the voltage signals of each electrode relative to this single reference electrode. During acquisition, the location of the reference electrode can be arbitrarily chosen. After acquisition, the data can be rereferenced as needed. The central processing unit 21 can select rereference schemes such as "average reference" and "bilateral mastoid reference" according to user configuration, performing rereference processing on the EEG data through numerical calculations. Different reference schemes have different application scenarios and can also affect the performance of classification and regression predictions under different cognitive and motor tasks.

[0059] As an example, based on user-preset programs and device configuration information, the central processing unit 21 can perform data bad channel repair and signal rereference on the EEG data to obtain a third data processing result; the graphics processing unit 22 can use a deep neural network or the programmable gate array circuit 23 can use a neuromorphic computing circuit to perform data noise reduction on the third data processing result to obtain a second data processing result; the graphics processing unit 22 can use a loaded deep neural network model or the programmable gate array circuit 23 can use a neuromorphic computing circuit to perform at least one of classification prediction and regression prediction on the second data processing result to obtain a first data processing result; the first data processing result is used as the target data processing result.

[0060] This application proposes a portable wearable heterogeneous edge AI EEG acquisition device 2 that integrates an edge heterogeneous AI computing architecture. The main function of the portable wearable heterogeneous edge AI EEG acquisition device 2 is to acquire EEG signals through a simulated front end, and according to the developer's wishes, send the EEG data to the most suitable computing resources for processing based on the characteristics of signal preprocessing and data analysis to obtain results such as noise reduction, classification prediction, and regression prediction, thereby completing heterogeneous computational analysis of the signals at the signal acquisition edge.

[0061] As an example, see Figure 1 As shown, the portable wearable heterogeneous edge AI EEG acquisition device 2 also includes an EEG cap port 105, a shell 106, a heat dissipation hole 107, a first operation button 108, a second operation button 109, a third operation button 110, a fourth operation button 111, a fifth operation button 112, a sixth operation button 113, an infrared light synchronization receiver 115, a display 116, and an input / output interface 24.

[0062] See Figure 3 As shown in the figure, this application provides a schematic diagram of the framework of another portable wearable heterogeneous edge AI EEG acquisition device 2. Figure 3As shown, the central processing unit 21 is connected to the graphics processor 22 and the programmable gate array (FPGA) circuit 23. The central processing unit 21 can be an embedded microprocessor, the graphics processor 22 can be an embedded graphics processor, and the FPGA circuit 23 can be a field-programmable gate array (FPGA). The central processing unit 21, the graphics processor 22, and the FPGA circuit 23 form a heterogeneous edge artificial intelligence control, processing, and analysis architecture 30.

[0063] The central processing unit 21 is connected to the memory 31 to read the computer program 31a therein; the central processing unit 21 is also connected to the communication module 25 to transmit control commands or forward EEG data when needed. The communication module 25 can be a gigabit wireless network card.

[0064] In some embodiments, memory 31 may be an internal storage unit, such as a hard disk or RAM. Memory 31 may be a removable / non-removable, volatile / non-volatile computer system storage medium; for example, memory 31 may be a non-volatile memory used for reading and writing non-volatile magnetic media. In other embodiments, memory 31 may also be an external storage device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Memory 31 is used to store the operating system, applications, a bootloader, data, and other programs, such as the program code of computer program 31a. Memory 31 may also be used to temporarily store data that has been output or will be output.

[0065] See Figure 3 As shown, the portable wearable heterogeneous edge AI EEG acquisition device 2 also includes at least one of the following: The general-purpose input / output interface 24 is used to output the target data processing result to an external device, so that the external device can perform a corresponding operation based on the target data processing result; the target data processing result is a control command signal. The wireless communication module 25 is used to send the target data processing results to other terminal devices.

[0066] Optionally, the external device can be a prosthesis, and the movements of the prosthesis can be controlled via control command signals. The terminal device can display the target data processing results or perform further analysis on the target data processing results.

[0067] Optionally, the external device can be a drone, whose activities can be controlled via control commands. The terminal device can display the target data processing results or perform further analysis on the target data processing results.

[0068] Optionally, the external device can be an emotional embodied intelligent device, through which the wearer's emotional state is transmitted via control commands to facilitate closed-loop intervention for emotional disorders. The terminal device can display the target data processing results or perform further analysis on the target data processing results.

[0069] Optionally, the wireless communication module 25 can provide communication solutions for applications on network devices, including wireless local area networks (WLANs) (such as Wi-Fi networks), Bluetooth, Zigbee, mobile communication networks, etc. The wireless communication module 25 can be one or more devices integrating at least one communication processing module. The wireless communication module 25 can include an antenna, which can have a single element or be an antenna array with multiple elements. The wireless communication module 25 can receive electromagnetic waves through the antenna, frequency-modulate and filter the electromagnetic wave signals, and send the processed signal to the processor. The wireless communication module 25 can also receive signals to be transmitted from the processor, frequency-modulate and amplify them, and then convert them into electromagnetic waves for radiation via the antenna.

[0070] As an example, the wireless communication module 25 in this embodiment of the application may employ a gigabit wireless network card.

[0071] In some embodiments, the central processing unit 21 is further configured to analyze the target data processing result, determine the output channel of the target data processing result, and output the target data processing result from the general-purpose input / output interface 24 or the wireless communication module 25 based on the output channel.

[0072] Optionally, embodiments of this application can analyze the target data processing results, determine the data type of the target data processing results, and determine the output channel of the target data processing results based on the data type. If it is a control command signal, it can be output to the general input / output interface 24; if it is other data processing results that have undergone data preprocessing or data analysis, it can be sent to the terminal device through the wireless communication module 25.

[0073] In some embodiments, the central processing unit 21 is further configured to acquire an infrared light synchronization signal, perform event tagging on the EEG data based on the infrared light synchronization signal, and time-align the EEG data with other data having the same event tag based on the event tagging of the EEG data.

[0074] See Figure 3 As shown, the central processing unit 21 is connected to the infrared light synchronization receiver 115 to provide a real-time response with high timing accuracy to external synchronization trigger signals. The central processing unit 21 is also connected to the display 116 to display necessary information.

[0075] See Figure 3 As shown, the EEG simulation front-end unit 26 is electrically connected to the central processing unit 21. The EEG simulation front-end unit 26 is used to acquire EEG signals, perform predetermined operations on the EEG signals, and obtain EEG data. The EEG data is a digital signal, and the predetermined operations include converting the analog signal into a digital signal.

[0076] See Figure 3 As shown, the central processing unit 21 is connected to the EEG simulation front-end unit 26 to control the EEG simulation front-end unit 26 to read EEG data; the central processing unit 21 is connected to the input / output interface 24 to control other external devices through secondary development programs in application scenarios where needed.

[0077] See Figure 3 As shown, the portable wearable heterogeneous edge AI EEG acquisition device 2 also includes several operation buttons 27 for users to set device configuration information.

[0078] The device configuration information includes at least one of the following: data processing window length, preprocessing parameters, and the artificial intelligence model to be loaded.

[0079] The preprocessing parameters include digital filter information, sampling rate information, bad channel repair information, and rereference information.

[0080] Optionally, the sampling rate information includes the frequency of the original sampling rate and the resampling rate, the bad channel repair information includes the bad channel repair status and repair method, the repair method can be linear interpolation, the rereference information is used for signal rereference, and the preprocessing parameters may also include noise reduction model information.

[0081] See Figure 3 As shown, the central processing unit 21 is connected to the operation button 27 so that the user can make necessary settings for the device; the EEG simulation front-end unit 26 is connected to the acquisition electrode 102 so as to amplify, sample, quantize and encode the EEG signal.

[0082] See Figure 4 As shown in the diagram, this application embodiment provides a schematic diagram of one side of the circuit board of a portable wearable heterogeneous edge AI EEG acquisition device 2. The circuit board of the portable wearable heterogeneous edge AI EEG acquisition device 2 of this application embodiment can install and connect various electronic components, including a first core board 201, a first core board connector 202, a circuit board fixing hole 203, a second core board connector 204, a second core board 205, an EEG cap socket 105, a power supply / charging port 207, supporting operating components 208, an EEG simulation front-end unit connector 209, an EEG simulation front-end unit 26, and a battery 211.

[0083] The first core board 201 includes a central processing unit 21 and a graphics processor 22, and the second core board 205 includes a programmable gate array circuit 23.

[0084] See Figure 5 As shown in the diagram, this application embodiment provides a schematic diagram of the other side of the circuit board of a portable wearable heterogeneous edge AI EEG acquisition device 2. The circuit board of the portable wearable heterogeneous edge AI EEG acquisition device 2 also includes: a first button element 212, a second button element 213, a third button element 214, a fourth button element 215, a fifth button element 216, a sixth button element 217, a gigabit wireless network card 218, a gigabit wireless network card connector 219, a solid-state drive connector 220, a solid-state drive 221, an input / output interface 24, an infrared light synchronization receiver 115, a display connector 224, a display cable 225, and a display 116.

[0085] The GPIO (General Purpose Input Output) settings are configured for input / output interface 24. The analysis model is the artificial intelligence model that needs to be loaded for data analysis.

[0086] This application provides a portable wearable heterogeneous edge AI EEG acquisition device 2 that integrates an edge heterogeneous AI computing architecture. The portable wearable heterogeneous edge AI EEG acquisition device 2 of this application embodiment acquires EEG signals through an EEG simulation front-end unit 26, and then transmits the acquired signals to computing chips with different architectures via onboard communication. The device performs edge-deployed EEG preprocessing and data analysis within itself, and finally outputs the target data processing results. The purpose of the portable wearable heterogeneous edge AI EEG acquisition device 22 of this application embodiment is described as follows: 1. Solve the problem of existing technologies' dependence on host computer communication and host computer software. Traditional portable wearable heterogeneous edge AI EEG acquisition devices rely on host computer communication primarily to transmit large amounts of EEG data to the host computer, where EEG signal preprocessing and analysis are performed. The portable wearable heterogeneous edge AI EEG acquisition device 2 of this application moves the EEG data preprocessing and analysis from the host computer to within the device itself, providing users with a customizable edge-end signal processing and data analysis computing platform. This avoids the dependence on host computer communication and software found in existing technologies.

[0087] 2. Address the issue of existing technologies lacking built-in heterogeneous edge AI computing power. Some existing portable wearable heterogeneous edge AI EEG acquisition devices may only have microcontrollers with limited computing power, lacking edge AI computing capabilities. This application's embodiments design a main control system based on a heterogeneous computing framework, allowing users to allocate different data processing and analysis tasks to computing resources with different architectures through secondary development, tailored to the characteristics of EEG data preprocessing and analysis. For example, signal decoding and control transmission can be completed by the CPU; signal filtering and domain transformation can be completed by designing dedicated circuits in an FPGA; and signal noise reduction, classification, and prediction tasks can be completed by the GPU using artificial neural network model inference.

[0088] By addressing the two key issues mentioned above, the embodiments of this application can resolve the problems of low overall reliability and portability, and high complexity of existing EEG technology application systems, thereby improving the real-time performance and stability of the application system and providing assurance for the system's offline independent and intelligent operation.

[0089] The beneficial effects of the portable wearable heterogeneous edge AI EEG acquisition device 2 in this application embodiment include: by innovatively realizing heterogeneous edge AI computing power, functions that previously required data transmission and upper computer software processing can be realized in the local edge heterogeneous computing power platform. This breaks through the traditional brain-computer interface application mode of "lower computer collecting EEG data and transmitting it to upper computer for processing", which can enhance the reliability, real-time performance, portability and intelligence of various EEG technology applications and reduce the complexity of related systems.

[0090] See Figure 6 As shown, this application provides a flowchart of a method for processing electroencephalogram (EEG) data. This EEG data processing method is applied to the portable wearable heterogeneous edge AI EEG acquisition device 2 of this application. Figure 6 As shown, the method for processing EEG data includes steps S601 to S603.

[0091] S601, Obtain device configuration information and acquire EEG data from the EEG simulation front-end unit.

[0092] S602. Based on user-preset programs and device configuration information, the data preprocessing and / or data analysis of EEG data are assigned to the corresponding artificial intelligence computing unit.

[0093] S603. The EEG data is preprocessed and / or analyzed by each artificial intelligence computing unit to obtain the target data processing result.

[0094] Optionally, the EEG data is preprocessed and / or analyzed by each AI computing unit to obtain the target data processing result, including: obtaining a first data processing result, a second data processing result, and a third data processing result; the first data processing result is obtained by the graphics processor 22 through data preprocessing and / or data analysis of the EEG data, the second data processing result is obtained by the programmable gate array circuit 23 through data preprocessing and / or data analysis of the EEG data, and the third data processing result is obtained by the central processing unit 21 through data preprocessing and / or data analysis of the EEG data. The target data processing result is obtained based on at least one of the first data processing result, the second data processing result, and the third data processing result.

[0095] Optionally, the EEG data is preprocessed and / or analyzed by various artificial intelligence computing units to obtain the target data processing result, including at least one of the following: Based on the loaded deep neural network model, at least one of data preprocessing and data analysis is performed on the EEG data to obtain a first data processing result; the data preprocessing includes noise reduction processing, and the data analysis includes at least one of classification prediction and regression prediction. Based on the loaded neuromorphic computing circuit, at least one of data preprocessing and data analysis is performed on the EEG data to obtain a second data processing result; the data preprocessing includes noise reduction processing, and the data analysis includes at least one of classification prediction and regression prediction. The EEG data is subjected to at least one of data bad channel repair and signal rereference to obtain a third data processing result; the preprocessing includes at least one of data bad channel repair and signal rereference.

[0096] Optionally, after preprocessing and / or analyzing the EEG data through each AI computing unit to obtain the target data processing result, the method further includes: analyzing the target data processing result, determining the output channel of the target data processing result, and outputting the target data processing result from the general input / output interface 24 or the wireless communication module 25 based on the output channel.

[0097] Optionally, the method for processing EEG data further includes: acquiring an infrared light synchronization signal, marking the EEG data for events based on the infrared light synchronization signal, and aligning the EEG data with other data having the same event markers in time based on the event markers of the EEG data.

[0098] As an example, this application embodiment provides a working principle of a portable wearable heterogeneous edge AI EEG acquisition device 2. After the user presses the "newline / switch" button on the casing 106, the portable wearable heterogeneous edge AI EEG acquisition device 2 is activated. After the central processing unit 21 loads the embedded operating system, it runs the EEG data processing method of this application embodiment.

[0099] Step 1: Read the device configuration information and store it in the buffer window.

[0100] Step 2: Acquire EEG data and store it in a buffer window.

[0101] Step 3: Based on the user-preset program and device configuration information, the sub-steps in preprocessing and data analysis are allocated to the central processing unit 21, the graphics processing unit 22, and the programmable gate array circuit 23, respectively, according to the required optimal computing resources. These are the three heterogeneous computing power architecture processing steps.

[0102] Step 4: The central processing unit 21 is used to perform computational tasks such as micro-control, data bad channel repair, and signal re-reference; the graphics processing unit 22 supports the operation of deep neural network models and is used for computational tasks such as data noise reduction, prediction, and classification; the programmable gate array circuit 23 completes computational tasks such as time-frequency transformation of signals and extraction of certain features in the form of dedicated circuits.

[0103] Step 5: Based on the user-preset program, the processing and analysis results of the three heterogeneous computing architectures are integrated. On the one hand, control command signals are output to the general input / output interface 24; on the other hand, with the relevant functions and parameters configured, data and analysis results are forwarded to the wireless communication module 25. Finally, the program begins data processing for the next buffer window using the same steps.

[0104] Step 6, the EEG data processing method, also includes an interrupt service routine designed for the central processing unit 21, used to handle button operations and infrared light synchronization signal triggering. Button operations can configure the data processing window length, EEG signal preprocessing parameters, and the analysis model to be loaded based on screen feedback. Infrared light synchronization triggering receives real-time event trigger signals transmitted by other devices via wireless infrared light encoding and marks the event information, such as the event name and time, into the EEG data for use in EEG event-related analysis functions.

[0105] It should be noted that the information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0106] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0107] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps in the above-described method embodiments.

[0108] If the integrated units described above are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a device / terminal equipment, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0109] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk drive (HDD), or solid-state drive (SSD), etc. The storage medium can also include combinations of the above types of memory.

[0110] This application provides a computer program product that, when run on a processor, enables the processor to execute the steps described in the various method embodiments above.

[0111] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0112] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0113] In the embodiments provided in this application, it should be understood that the disclosed apparatus / network devices and methods can be implemented in other ways. For example, the apparatus / network device embodiments described above are merely illustrative. For instance, the division of modules or units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0114] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0115] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A portable wearable heterogeneous edge artificial intelligence electroencephalogram acquisition device, characterized in that, Comprise: A central processor, at least two data processing units with different computing power architectures, and an electroencephalogram simulation front-end unit, the central processor and the data processing unit are artificial intelligence computing units; The central processor is used to obtain device configuration information and electroencephalogram data from the electroencephalogram simulation front-end unit, based on user preset programs and device configuration information, the data preprocessing and / or data analysis of the electroencephalogram data is assigned to the corresponding artificial intelligence computing unit, and the electroencephalogram data is preprocessed and / or analyzed by each artificial intelligence computing unit to obtain the target data processing result.

2. The portable wearable heterogeneous edge artificial intelligence electroencephalogram acquisition device according to claim 1, characterized in that, At least two data processing units include: a graphics processor and a programmable gate array circuit; The graphics processor is used to preprocess and / or analyze the electroencephalogram data based on a deep neural network to obtain a first data processing result; The programmable gate array circuit is used to preprocess and / or analyze the electroencephalogram data based on brain-like computing to obtain a second data processing result; The central processor is used to preprocess and / or analyze the electroencephalogram data to obtain a third data processing result; based on at least one of the first data processing result, the second data processing result and the third data processing result, the target data processing result is obtained.

3. The portable wearable heterogeneous edge artificial intelligence electroencephalogram acquisition device according to claim 2, characterized in that, Further comprising at least one of: The graphics processor is used to preprocess and analyze at least one of the electroencephalogram data based on a loaded deep neural network model to obtain a first data processing result; the data preprocessing includes noise reduction processing, and the data analysis includes at least one of classification prediction and regression prediction; The programmable gate array circuit is based on a loaded brain-like computing circuit to preprocess and analyze at least one of the electroencephalogram data to obtain a second data processing result; the data preprocessing includes noise reduction processing, and the data analysis includes at least one of classification prediction and regression prediction; The central processor is used to preprocess and analyze at least one of the electroencephalogram data to obtain a third data processing result; The data preprocessing includes at least one of data bad channel repair and signal reference.

4. The portable wearable heterogeneous edge artificial intelligence electroencephalogram acquisition device according to claim 1, wherein, The portable wearable heterogeneous edge artificial intelligence electroencephalogram acquisition device further comprises at least one of: A general input / output interface is used to output the target data processing result to an external device, so that the external device performs corresponding operations based on the target data processing result; the target data processing result is a control instruction signal; A wireless communication module is used to send the target data processing result to other terminal devices.

5. The portable wearable heterogeneous edge artificial intelligence electroencephalogram acquisition device according to claim 4, characterized in that, The central processor is also used to analyze the target data processing result and determine the output channel of the target data processing result, and based on the output channel, the target data processing result is output from the general input / output interface or the wireless communication module.

6. The portable wearable heterogeneous edge artificial intelligent electroencephalogram acquisition device according to any one of claims 1-5, characterized in that, The central processor is also configured to acquire an infrared light synchronization signal, mark events of the electroencephalogram data based on the infrared light synchronization signal, and time-align the electroencephalogram data with other data having the same event mark based on the event mark of the electroencephalogram data.

7. The portable wearable heterogeneous edge artificial intelligent electroencephalogram acquisition device according to any one of claims 1-5, characterized in that, The electroencephalogram analog front-end unit is electrically connected to the central processor, and is configured to acquire an electroencephalogram signal, perform a predetermined operation on the electroencephalogram signal, and obtain the electroencephalogram data; the electroencephalogram data is a digital signal, and the predetermined operation includes converting an analog signal into a digital signal.

8. The portable wearable heterogeneous edge artificial intelligent electroencephalogram acquisition device according to any one of claims 1-5, characterized in that, The portable wearable heterogeneous edge artificial intelligence electroencephalogram acquisition device further comprises a plurality of operation buttons for a user to set the device configuration information. The device configuration information comprises at least one of the following: a data processing window length, a preprocessing parameter, and an artificial intelligence model to be loaded. The preprocessing parameter comprises digital filter information, sampling rate information, data bad channel repair information, and re-reference information.

9. An electroencephalography acquisition system, characterized in that, The portable wearable heterogeneous edge artificial intelligence electroencephalogram acquisition device comprises: an electroencephalogram cap assembly, and the portable wearable heterogeneous edge artificial intelligence electroencephalogram acquisition device according to any one of claims 1-8; The electroencephalogram cap assembly comprises a plurality of acquisition electrodes for acquiring electroencephalogram signals of a user. The electroencephalogram cap assembly is configured to be connected to the portable wearable heterogeneous edge artificial intelligence electroencephalogram acquisition device, and transmit the electroencephalogram signals of each acquisition electrode to the portable wearable heterogeneous edge artificial intelligence electroencephalogram acquisition device.

10. A method of processing electroencephalographic data, characterized by, The method is applied to the portable wearable heterogeneous edge artificial intelligence electroencephalogram acquisition device according to any one of claims 1-8, and comprises: acquiring device configuration information and acquiring electroencephalogram data from an electroencephalogram analog front-end unit; based on a user preset program and the device configuration information, assigning data preprocessing and / or data analysis of the electroencephalogram data to corresponding artificial intelligence computing units; performing data preprocessing and / or data analysis on the electroencephalogram data by each artificial intelligence computing unit to obtain a target data processing result.

Images