Brain-computer interface auxiliary rehabilitation system based on double-feedback closed-loop mechanism

Abstract

The invention provides a brain-computer interface auxiliary rehabilitation system based on a double-feedback closed-loop mechanism, and the system is characterized in that an online signal collection module collects an electroencephalogram signal and an electromyographic signal of a user in real time; the motion intention recognition model recognizes the motion intention of the user in real time according to the electroencephalogram signals collected in real time; the feedback optimization module generates a motion intention activation intensity display signal in real time according to the motion intention recognized in real time; the graphical user interface displays the signal according to the activation intensity of the motion intention and displays and updates the activation intensity of the motion intention in real time in a progress bar mode; the intervention execution module generates an intervention execution signal when the activation intensity of the motion intention reaches an activation intensity threshold value; the auxiliary exercise equipment assists the user in completing an exercise rehabilitation task according to the intervention execution signal. The intervention execution module also obtains brain-myoelectricity fusion rehabilitation index signals according to the collected brain electrical signals and myoelectricity signals, and regulates and controls intervention execution signals according to the rehabilitation index signals, thereby regulating and controlling the intervention of the auxiliary exercise equipment.

Description

Technical Field

[0001] The present invention belongs to the technical field of rehabilitation. Specifically, it relates to a brain-computer interface assisted rehabilitation system based on a dual-feedback closed-loop mechanism, which combines a functional electrical stimulation device and an exoskeleton device. Background Art

[0002] Neuroscience and clinical medicine research have shown that the restoration and reconstruction of nerve function are crucial for patients with motor dysfunction caused by stroke, spinal cord injury, or neurodegenerative diseases. Neurorehabilitation usually acts on the synergistic process of nerve signal transmission and muscle activation. A complete movement process can be divided into four stages: generation of movement intention, nerve signal transmission, muscle activation, and movement execution. It starts from the generation of movement intention in the motor cortex of the brain, passes through the spinal cord to transmit nerve signals to the target muscles, and then completes the movement through muscle activation. The generation of movement intention is characterized by specific movement-related brain waves (electroencephalograph, EEG) recorded by scalp electrodes; nerve signal transmission is characterized by the electrophysiological activities of the spinal cord or peripheral nerves; muscle activation is characterized by the muscle electrical activities recorded by electromyography (EMG).

[0003] Research has shown that active rehabilitation with autonomous training can significantly promote the development of neural plasticity by enhancing the synergistic effect between the generation of movement intention and muscle activation. However, the effect of traditional passive training is relatively general due to the lack of active stimulation of nerve activity. Although the mechanism of neural plasticity is not yet fully understood, some research results indicate that neural plasticity is related to the strengthening or reorganization of synaptic connections, and the precise decoding and real-time feedback of nerve signals may further promote the repair and reconstruction of neural circuits. Therefore, the closed-loop treatment with autonomous training can more effectively stimulate the nerve activity of patients, thereby accelerating the recovery of nerve function.

[0004] However, the existing technologies are usually passive rehabilitation. Passive rehabilitation methods mainly rely on the experience of therapists or external device-assisted training. Although they can provide certain rehabilitation assistance, the effects of these methods are often limited by the skills of therapists and the participation of patients. Therefore, passive training cannot stimulate the active participation of patients and the autonomous activities of the nervous system, resulting in a slow recovery rate of neural plasticity, unable to fully activate neural circuits, and thus limiting the improvement of rehabilitation effects. Summary of the Invention

[0005] In order to solve the above technical problems existing in the prior art, the present invention provides a brain-computer interface assisted rehabilitation system based on a dual-feedback closed-loop mechanism.

[0006] A brain-computer interface assisted rehabilitation system based on a dual-feedback closed-loop mechanism provided according to one aspect of the present invention includes: an online signal acquisition module for real-time acquisition of the user's electroencephalogram (EEG) signals; a motion intention recognition model for recognizing the user's motion intention according to the real-time acquired EEG signals; a feedback optimization module for generating and real-time updating a motion intention activation intensity display signal for the recognized motion intention, the motion intention activation intensity display signal being used to control the display of the activation intensity of the user's motion intention; an intervention execution module for generating an intervention execution signal when the activation intensity of the recognized motion intention reaches an activation intensity threshold; and an assistive motion device for assisting the user to complete a motion rehabilitation task according to the generated intervention execution signal.

[0007] In an example of the brain-computer interface assisted rehabilitation system based on a dual-feedback closed-loop mechanism provided in the above aspect, the online signal acquisition module is further configured to acquire the user's electromyogram (EMG) signals; the intervention execution module is further configured to obtain an EEG-EMG fusion rehabilitation index signal according to the acquired EEG signals and EMG signals, and is further configured to regulate the generated intervention execution signal according to the obtained EEG-EMG fusion rehabilitation index signal.

[0008] In an example of the brain-computer interface assisted rehabilitation system based on a dual-feedback closed-loop mechanism provided in the above aspect, the assistive motion device includes: a functional electrical stimulation device for performing electrical stimulation intervention on the user according to the electrical stimulation parameters set by the generated intervention execution signal; and an exoskeleton device for performing assistive motion intervention on the user according to the motion control parameters set by the generated intervention execution signal.

[0009] In an example of the brain-computer interface assisted rehabilitation system based on a dual-feedback closed-loop mechanism provided in the above aspect, the motion intention recognition model includes: a primary convolutional layer, a depthwise separable convolutional layer, and a fully connected layer; wherein, the primary convolutional layer is used to extract the time-domain features of the real-time acquired EEG signals through a local receptive field, the depthwise separable convolutional layer is used to further extract the spatial features, and the fully connected layer maps the extracted features to specific motion intention categories and outputs the motion intentions of specific categories.

[0010] In an example of the brain-computer interface assisted rehabilitation system based on a dual-feedback closed-loop mechanism provided in the above aspect, the motion intention recognition model further includes: a preprocessing layer for performing preprocessing of filtering, denoising, and segmentation on the real-time acquired EEG signals.

[0011] In an example of the brain-computer interface assisted rehabilitation system based on a dual-feedback closed-loop mechanism provided in the above aspect, the motion intention recognition model further includes: a postprocessing layer for performing probability threshold screening processing on the motion intention prediction values output by each sliding window.

[0012] In an example of the brain-computer interface assisted rehabilitation system based on the dual-feedback closed-loop mechanism provided in the above aspect, the brain-computer interface assisted rehabilitation system based on the dual-feedback closed-loop mechanism further includes: an offline calibration module, configured to acquire the training EEG signals of the user, and train a motion intention recognition model for this user according to the acquired training EEG signals.

[0013] In an example of the brain-computer interface assisted rehabilitation system based on the dual-feedback closed-loop mechanism provided in the above aspect, the offline calibration module includes: an offline signal acquisition module, configured to acquire the training EEG signals of the user; a model training module, configured to train a motion intention recognition model for this user according to the acquired training EEG signals.

[0014] In an example of the brain-computer interface assisted rehabilitation system based on the dual-feedback closed-loop mechanism provided in the above aspect, the brain-computer interface assisted rehabilitation system based on the dual-feedback closed-loop mechanism further includes: a graphical user interface, configured to display signals according to the generated activation intensity of the motion intention and update and display the activation intensity of its motion intention in real time in the form of a progress bar.

[0015] In an example of the brain-computer interface assisted rehabilitation system based on the dual-feedback closed-loop mechanism provided in the above aspect, the graphical user interface is further configured to display electrical stimulation parameters, EEG signals, EMG signals, experimental parameters, experimental paradigms, and system status to the user.

[0016] Beneficial effects: The present invention performs real-time brain-computer fusion regulation, and through physiological signal induction - brain state-based control - electrical stimulation device and exoskeleton-assisted movement - further physiological signal-induced closed-loop operation, to achieve visual feedback and rehabilitation index feedback, thereby realizing an online closed-loop feedback mechanism. In addition, based on the brain state control command activation, positive visual feedback process, and induced electrical stimulation and exoskeleton intervention in the present invention, users can be enabled to actively participate in autonomous training. Further, in the present invention, the electrical stimulation device and the exoskeleton adopt complementary collaborative control, that is, the electrical stimulation device preferentially activates muscle proprioception, and the exoskeleton provides precise torque assistance. The two avoid over-reliance on a single mode through a weight distribution algorithm; and the intervention effects of the two are verified by multiple means such as electroencephalogram signals (enhanced electroencephalogram frequency band energy), electromyogram signals (muscle activation degree), and brain-electromyogram fusion rehabilitation indexes, realizing the integration of "stimulation-assisted evaluation". Further, the present invention defines a corticomuscular coordination index, which comprehensively combines the movement-related cortical potential of the electroencephalogram signal and the muscle coordination activation time sequence of the electromyogram signal to quantify the rehabilitation progress; the corticomuscular coordination index, as the core feedback parameter, drives the system to adjust the stimulation frequency and exoskeleton stiffness in real time to form a personalized rehabilitation plan. The present invention uses a lightweight deep learning network for online decoding (i.e., recognition), supports millisecond-level online decoding, and ensures that the delay from collecting electroencephalogram signals to electrical stimulation / exoskeleton intervention is less than 500 ms. The wireless hardware platform of the present invention uses a local area network wireless protocol, supports users to move freely in multiple scenarios, and breaks through the spatial limitations of traditional wired systems. Further, the integrated software platform of the present invention integrates the data streams of multiple modules through a unified timing scheduling framework to ensure the strict synchronization of algorithms, control instructions, and feedback signals. Brief Description of the Drawings

[0017] Through the following description with reference to the accompanying drawings, the above and other aspects, features, and advantages of the embodiments of the present invention will become more apparent, in which:

[0018] Figure 1 is a schematic diagram of the graphical user interface of a brain-computer interface-assisted rehabilitation system based on a dual feedback closed-loop mechanism according to an embodiment of the present invention;

[0019] Figure 2 is a schematic diagram of the communication hardware adopted by a brain-computer interface-assisted rehabilitation system based on a dual feedback closed-loop mechanism according to an embodiment of the present invention;

[0020] Figure 3 is a block diagram of a brain-computer interface-assisted rehabilitation system based on a dual feedback closed-loop mechanism according to an embodiment of the present invention;

[0021] Figure 4 is an architecture diagram of a movement intention recognition model according to an embodiment of the present invention. Detailed Description of the Embodiments

[0022] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts shall fall within the protection scope of the present application.

[0023] Those skilled in the art of the present technology can understand that unless specifically stated, the singular forms "a", "an", "the" and "said" used herein may also include the plural forms. It should be further understood that the term "comprising" used in the specification of the present invention means the presence of the described features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or their groups. It should be understood that when the present application states that an element is "connected" or "coupled" to another element, it can be directly connected or coupled to other elements, or there may also be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The phrase "and / or" used herein includes all or any unit and all combinations of one or more related listed items.

[0024] Those skilled in the art of the present technology can understand that unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as the general understanding of those of ordinary skill in the art to which the present invention belongs. It should also be understood that terms such as those defined in a general dictionary should be understood to have a meaning consistent with the meaning in the context of the prior art, and will not be interpreted with an idealized or overly formal meaning unless specifically defined as here.

[0025] Hereinafter, a graphical user interface adopted by a brain-computer interface assisted rehabilitation system based on a dual feedback closed-loop mechanism according to an embodiment of the present invention will be described in detail. Figure 1 is a schematic diagram of a graphical user interface of a brain-computer interface assisted rehabilitation system based on a dual feedback closed-loop mechanism according to an embodiment of the present invention.

[0026] Refer to Figure 1 , in an embodiment according to the present invention, a graphical user interface (GUI) integrating acquisition and control is designed through PyQT5, which is at least used to display the user's electroencephalogram (EEG), electromyogram (EMG), activation intensity of movement intention, and electrical stimulation parameters in real time. The design of this graphical user interface fully considers the convenience of user operation and the clarity of information presentation, enabling users to efficiently perform real-time data monitoring and parameter adjustment, thereby enhancing the practicality of the system and the user experience.

[0027] The graphical user interface includes multiple functional areas, specifically including the following parts:

[0028] Experimental parameter setting area: Located in the upper left corner of the graphical user interface, it provides multiple input boxes and adjustment sliders for configuring parameters such as device connection, stage selection, decoding control model configuration, and saving records of subject information data, and sets start and stop switches. With one-key operation, the process can be started or paused, greatly improving the convenience and controllability of operation.

[0029] Electrical stimulation parameter adjustment area: Users can precisely set the key parameters of the functional electrical stimulation device (FES) through these controls, such as stimulation intensity, frequency, and pulse width. These parameters directly affect the treatment effect, so users can flexibly adjust them according to the specific needs of patients to achieve personalized rehabilitation intervention.

[0030] Experimental paradigm display window: Used to display the activation intensity of the user's movement intention to the user in the form of a progress bar and display visual stimulation prompts for the current task to help the user intuitively understand the experimental process.

[0031] Electromyogram signal display window: Presents the real-time electromyogram signal waveform to facilitate users to observe and analyze their muscle activity status and provide a basis for adjusting electrical stimulation parameters.

[0032] Electromyogram device control area: This area is mainly used to configure the electromyogram device, including: the "Scan" button under the connected electromyogram device can search for nearby electromyogram devices, establish connection pairing with the scanned electromyogram devices, set the conditions for triggering start and stop, and the data acquisition process of the electromyogram device can be set to start and stop.

[0033] Electroencephalogram signal display window: Presents the real-time electroencephalogram signal waveform to help users evaluate and understand the characteristics of electroencephalogram activities and provide support for decoding the user's movement intention.

[0034] System status monitoring area: Located on the right side of the graphical user interface, it real-time displays the current device connection status, statistical information of data acquisition, and model decoding information, such as the number of connected sensors, the number of data frames collected, the system operation status, and the model decoding rate, etc. These information ensure that users can instantly understand the operation situation of the system, so as to discover and solve problems in a timely manner.

[0035] Through the reasonable layout and design of the above functional areas, the graphical user interface according to the embodiment of the present invention realizes the integrated integration of data acquisition, signal real-time display, parameter control, and system status display. Users can intuitively monitor the real-time changes of electroencephalogram signals and electromyogram signals through the graphical user interface and adjust the electrical stimulation parameters for each individual as needed, so as to achieve precise individualized regulation of the rehabilitation process.

[0036] Next, the hardware of the brain-computer interface assisted rehabilitation system based on the dual-feedback closed-loop mechanism according to an embodiment of the present invention will be described in detail. Figure 2 It is a schematic diagram of the communication hardware adopted by the brain-computer interface assisted rehabilitation system based on the dual-feedback closed-loop mechanism according to an embodiment of the present invention.

[0037] Refer to Figure 2 , the communication hardware is the basic module of the brain-computer interface assisted rehabilitation system based on the dual-feedback closed-loop mechanism according to an embodiment of the present invention, and mainly provides data support for the training of the personalized motion intention recognition model and the online intervention process. These communication hardwares provide comprehensive physiological signal inputs for subsequent real-time intervention and personalized training through the combination of high-precision electroencephalogram acquisition, electromyogram acquisition, electrical stimulation devices and exoskeleton devices.

[0038] In terms of electroencephalogram signal acquisition, a 64-channel wireless electroencephalogram amplifier is used according to an embodiment of the present invention. This amplifier has the characteristics of high precision and high sampling rate, and can collect the electrical activity signals of the cerebral cortex in real time. The design of 64 channels enables the device to capture the neural activities of different regions of the brain more comprehensively, effectively reduce signal noise, and improve the resolution of the signals. During data acquisition, the device is connected to the amplifier through electroencephalogram electrodes, and uses wireless data transmission technology to transmit the collected EEG signals to the computer for processing in real time. To ensure the stability and accuracy of the signals, the electroencephalogram amplifier is also equipped with an A / D converter to convert the analog signals into digital signals for further analysis.

[0039] In terms of electromyogram signal acquisition, a wireless surface electromyogram system is used according to an embodiment of the present invention. This system uses highly sensitive sensors to accurately capture the electrical activity signals of the muscles. This wireless surface electromyogram system records the contraction and relaxation states of the muscles by attaching electrode sensors to the target muscle areas of the user, and reflects the performance of the muscles during exercise in real time.

[0040] The functional electrical stimulation device according to an embodiment of the present invention adopts a multi-channel programmable electrical stimulation device. This multi-channel programmable electrical stimulation device is a high-performance electrical stimulator, which is widely used in neuromodulation and electrical stimulation therapy. In the embodiment of the present invention, this multi-channel programmable electrical stimulation device activates the target muscles of the user through electrical stimulation to assist in completing the motion tasks. This device supports a variety of electrical stimulation modes and adjustment parameters, and can accurately control the stimulation intensity, frequency and duration according to personalized training needs to ensure effective muscle activation.

[0041] In an embodiment according to the present invention, the exoskeleton device is mainly used to assist a user (e.g., a person with motor disabilities) in performing lower limb-related movements. The exoskeleton device starts the exoskeleton system to adaptively configure corresponding movement patterns, including: standing, walking, climbing slopes, etc., by receiving in real time control commands triggered by the user's movement intention, so as to assist the user to complete corresponding movement actions. Among them, the exoskeleton device is also connected to the data communication hardware platform in a wireless manner.

[0042] The communication module provides signal input for the subsequent online intervention module by synchronously collecting electroencephalogram (EEG) signals, electromyogram (EMG) signals and electrical stimulation parameters. In this communication module, the signal acquisition of all devices is transmitted through wireless communication to ensure the real-time and accuracy of data, and the wireless communication method provides convenience and feasibility for lower limb motor rehabilitation intervention, and the subject can move in a free space. Through the combination of these high-precision devices, the communication module provides a solid foundation for the training of the personalized movement intention recognition model by collecting rich neural signal data, thus laying a support for the efficient implementation of real-time movement intention recognition and interventions such as functional electrical stimulation (FES) and exoskeleton assistance.

[0043] Next, a brain-computer interface assisted rehabilitation system based on a dual-feedback closed-loop mechanism according to an embodiment of the present invention will be described in detail. Figure 3 is a block diagram of a brain-computer interface assisted rehabilitation system based on a dual-feedback closed-loop mechanism according to an embodiment of the present invention.

[0044] Referring to Figure 3 , a brain-computer interface assisted rehabilitation system based on a dual-feedback closed-loop mechanism according to an embodiment of the present invention includes: an online signal acquisition module 31, a movement intention recognition model 32, a feedback optimization module 33, a graphical user interface 34, an intervention execution module 35 and an assisted movement device 36. Among them, the online signal acquisition module 31, the movement intention recognition model 32, the feedback optimization module 33, the graphical user interface 34, the intervention execution module 35 and the assisted movement device 36 constitute an online intervention module 30.

[0045] Specifically, the online signal acquisition module 31 is used to collect the electroencephalogram signals of the user in real time. In one example, the online signal acquisition module 31 includes the above-mentioned 64-channel wireless EEG amplifier, wireless surface EMG system, etc.

[0046] The movement intention recognition model 32 is used to recognize the movement intention of the user according to the electroencephalogram signals collected in real time. Here, the specific architecture of the movement intention recognition model 32 will be described below.

[0047] The feedback optimization module 33 is used to generate and update in real time a display signal for the activation intensity of the recognized motion intention, wherein the display signal for the activation intensity of the motion intention is used to control the display of the activation intensity of the user's motion intention.

[0048] The graphical user interface 34 is used to display and update in real time the activation intensity of its motion intention according to the generated display signal for the activation intensity of the motion intention in the form of a progress bar. Specifically, the graphical user interface 34 is visually presented in the form of a progress bar to intuitively reflect the intensity of the electroencephalogram activity of the user under the motor imagery (MI) task. Therefore, in one example, this brain-computer interface-assisted rehabilitation system based on a dual-feedback closed-loop mechanism adopts a sliding window analysis strategy, sets 2 seconds as the basic recognition time window, and takes 0.5 seconds as the step size to continuously collect and analyze the electroencephalogram signal data. In a 10-second test cycle, 17 consecutive sliding time windows can be generated. When an activation signal of the motion intention is detected in a certain time window, a blue square will be displayed at the corresponding position of the progress bar interface on the graphical user interface 34 as an indication; if the activation state is detected in multiple consecutive time windows, the blue squares will be cumulatively superimposed, forming a visual feedback reinforcement effect on the user, so as to encourage the user to induce the brain to continuously generate the motion intention state in the same way; on the contrary, if a non-activation state is detected in a certain time window, the progress bar will be reset to the initial state.

[0049] The intervention execution module 35 is used to generate an intervention execution signal when the activation intensity of the recognized motion intention reaches the activation intensity threshold. In one example, when the intervention execution module 35 recognizes that, for example, the stable motion intention activation state appears in 5 consecutive time windows, the cumulative degree of the progress bar reaches the preset threshold and turns red. At this time, the activation intensity of the motion intention recognized by the intervention execution module 35 reaches the activation intensity threshold, so that the intervention execution module 35 generates an intervention execution signal.

[0050] The assisted motion device 36 is used to assist the user to complete the motion rehabilitation task according to the generated intervention execution signal. In one example, the assisted motion device 36 includes the above-mentioned functional electrical stimulation device and exoskeleton device, wherein the functional electrical stimulation device performs electrical stimulation intervention on the user according to the electrical stimulation parameters set by the generated intervention execution signal, and the exoskeleton device is used to perform assisted motion intervention on the user according to the motion control parameters set by the generated intervention execution signal.

[0051] In addition, in an embodiment according to the present invention, the online signal acquisition module 31 is further configured to acquire the myoelectric signals of the user, and the intervention execution module 35 is further configured to obtain a brain-myolectric fusion rehabilitation index signal (such as the corticomuscular coherence index (CMC)) according to the acquired electroencephalogram signals and myoelectric signals, and is further configured to regulate the generated intervention execution signal according to the obtained brain-myolectric fusion rehabilitation index signal. Therefore, after regulating the intervention execution signal according to the brain-myolectric fusion rehabilitation index signal, the electrical stimulation parameters of the functional electrical stimulation device and the motion control parameters of the exoskeleton device are also adjusted accordingly, so as to adaptively optimize the rehabilitation strategy to meet the individual needs of different rehabilitation stages.

[0052] Continuing to refer to Figure 3 , the brain-computer interface assisted rehabilitation system based on a dual feedback closed-loop mechanism according to an embodiment of the present invention further includes: an offline calibration module 40. Specifically, the offline calibration module 40 includes: an offline signal acquisition module 41, configured to acquire the training electroencephalogram signals of the user; and a model training module 42, configured to train a motion intention recognition model for the user according to the acquired training electroencephalogram signals. In the offline training stage, the model training module 42 is trained by a large number of labeled electroencephalogram signals acquired by the offline signal acquisition module 41 to learn the mapping relationship from electroencephalogram signals to motion intentions. The training data (i.e., the training electroencephalogram signals used for training) includes the electroencephalogram signals of the user when performing different motion tasks, as well as the corresponding motion intention labels. Through the backpropagation algorithm and gradient descent optimization, the model training module 42 gradually adjusts the model parameters to minimize the classification error, thereby improving the accuracy of motion intention recognition.

[0053] Next, the architecture of the motion intention recognition model 32 according to an embodiment of the present invention will be described in detail. Figure 4 It is an architecture diagram of the motion intention recognition model according to an embodiment of the present invention.

[0054] Referring to Figure 4 , the motion intention recognition model 32 according to an embodiment of the present invention includes: a preprocessing layer 321, a primary convolutional layer 322, a depthwise separable convolutional layer 323, a fully connected layer 324, and a postprocessing layer 325.

[0055] Specifically, the motion intention recognition model 32 according to an embodiment of the present invention is a lightweight and efficient deep learning network, which is used to process multi-channel electroencephalogram signals and can extract features related to motion intentions from complex electroencephalogram activities, so as to achieve high-precision motion intention recognition. Therefore, all layers of the motion intention recognition model 32 work together to effectively capture the highly representative spatio-temporal features in the electroencephalogram signals and classify them.

[0056] Further, the preprocessing layer 321 receives the input multi-channel EEG signals. After signal preprocessing such as filtering, denoising, and segmentation, the preprocessed EEG signals are output to the initial convolutional layer 322. The initial convolutional layer 322 extracts the time-domain features of the EEG signals through local receptive fields, and then further extracts the spatial features through the depthwise separable convolutional layer 323, thus realizing the joint modeling of the spatio-temporal features of the EEG signals. Finally, the fully connected layer 324 maps the extracted high-dimensional features to specific movement intention categories (such as ankle dorsiflexion movement, etc.) and outputs the movement intention of specific categories (i.e., the classification result). In addition, to improve the robustness of recognition, the post-processing layer 325 performs probability threshold screening on the movement intention prediction values output by each sliding window to further reduce the misrecognition rate. The advantage of the movement intention recognition model 32 lies in its lightweight structure and high efficiency, which can achieve high-precision real-time recognition of movement intentions with limited computing resources. In addition, the movement intention recognition model 32 also has strong generalization ability, can adapt to different EEG signal characteristics, and train a personalized movement intention recognition model for each user, thus supporting personalized rehabilitation training.

[0057] In summary, the present application performs real-time brain-computer fusion regulation and realizes an online closed-loop feedback mechanism through a closed-loop operation of physiological signal induction-based brain state control-electrical stimulation device and exoskeleton-assisted movement-further physiological signal induction to achieve visual feedback and rehabilitation index feedback. In addition, based on the brain state control command activation, positive visual feedback process, and evoked electrical stimulation and exoskeleton intervention in the present application, users can actively participate in autonomous training. In addition, in the present application, the electrical stimulation device and the exoskeleton adopt complementary collaborative control, that is, the electrical stimulation device preferentially activates muscle proprioception, and the exoskeleton provides precise torque assistance. The two avoid over-reliance on a single mode through a weight distribution algorithm; and the intervention effects of the two are verified by multiple factors such as EEG signals (enhanced EEG frequency band energy), EMG signals (muscle activation degree), and brain-EMG fusion rehabilitation indexes to achieve "stimulation-assisted evaluation" integration. Further, the present application defines the corticomuscular coordination index (CMC), synthesizes the movement-related cortical potential of EEG signals and the muscle coordination activation time series of EMG signals, and quantifies the rehabilitation progress; CMC, as the core feedback parameter, drives the system to adjust the stimulation frequency and exoskeleton stiffness in real time to form a personalized rehabilitation plan. The present application uses a lightweight deep learning network for online decoding (i.e., recognition), supports millisecond-level online decoding, and ensures that the delay from EEG signal acquisition to electrical stimulation / exoskeleton intervention is less than 500 ms. The wireless hardware platform of the present application adopts a local area network wireless protocol, supports users to move freely in multiple scenarios, and breaks through the space limitation of traditional wired systems. Further, the integrated software platform of the present application integrates the data streams of multiple modules through a unified timing scheduling framework to ensure the strict synchronization of algorithms, control instructions, and feedback signals.

[0058] The above are only the embodiments of the present application, and do not limit the patent scope of the present application accordingly. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present application, or directly or indirectly applied in other related technical fields, shall similarly be included in the patent protection scope of the present application.

Claims

1. A brain-computer interface assisted rehabilitation system based on a double feedback closed-loop mechanism, characterized in that, The brain-computer interface assisted rehabilitation system based on a dual-feedback closed-loop mechanism includes: An online signal acquisition module for real-time acquisition of the user's electroencephalogram (EEG) signals; A movement intention recognition model for recognizing the user's movement intention based on the real-time acquired EEG signals; A feedback optimization module for generating a movement intention activation intensity display signal in real time for the real-time recognized movement intention, wherein the movement intention activation intensity display signal is used to control the display of the activation intensity of the user's movement intention; An intervention execution module for generating an intervention execution signal when the activation intensity of the recognized movement intention reaches an activation intensity threshold; An assisted movement device for assisting the user to complete a movement rehabilitation task according to the generated intervention execution signal.

2. The brain-computer interface assisted rehabilitation system based on a dual feedback closed-loop mechanism according to claim 1, wherein The online signal acquisition module is also used to acquire the user's electromyogram (EMG) signals; The intervention execution module is also used to obtain a brain-EMG fusion rehabilitation index signal based on the acquired EEG signals and EMG signals, and is also used to regulate the generated intervention execution signal according to the obtained brain-EMG fusion rehabilitation index signal.

3. The brain-computer interface assisted rehabilitation system based on a double feedback closed-loop mechanism according to claim 1 or 2, characterized in that, The assisted device includes: A functional electrical stimulation device for performing electrical stimulation intervention on the user according to the electrical stimulation parameters set by the generated intervention execution signal; An exoskeleton device for performing assisted movement intervention on the user according to the movement control parameters set by the generated intervention execution signal.

4. The brain-computer interface assisted rehabilitation system based on the double feedback closed-loop mechanism according to claim 1, characterized in that, The movement intention recognition model includes: a primary convolutional layer, a depthwise separable convolutional layer, and a fully connected layer; Among them, the primary convolutional layer is used to extract the time-domain features of the real-time acquired EEG signals through a local receptive field, the depthwise separable convolutional layer is used to further extract spatial features, and the fully connected layer maps the extracted features to specific movement intention categories and outputs the movement intention of specific categories.

5. The brain-computer interface assisted rehabilitation system based on a double feedback closed-loop mechanism according to claim 4, characterized in that, The movement intention recognition model also includes: a preprocessing layer for performing preprocessing such as filtering, denoising, and segmentation on the real-time acquired EEG signals.

6. The brain-computer interface assisted rehabilitation system based on a dual-feedback closed-loop mechanism according to claim 4 or 5, characterized in that The movement intention recognition model also includes: a postprocessing layer for performing probability threshold screening processing on the movement intention prediction values output by each sliding window.

7. The brain-computer interface assisted rehabilitation system based on the double feedback closed-loop mechanism according to claim 1, wherein The brain-computer interface assisted rehabilitation system based on a dual-feedback closed-loop mechanism further includes: An offline calibration module for acquiring the user's training EEG signals and training a movement intention recognition model for this user based on the acquired training EEG signals.

8. The brain-computer interface assisted rehabilitation system based on a dual feedback closed-loop mechanism according to claim 7, characterized in that, The offline calibration module includes: An offline signal acquisition module for acquiring the user's training EEG signals; A model training module for training a movement intention recognition model for this user based on the acquired training EEG signals.

9. The brain-computer interface assisted rehabilitation system based on a double feedback closed-loop mechanism according to claim 1, characterized in that, The brain-computer interface assisted rehabilitation system based on a dual-feedback closed-loop mechanism further includes: A graphical user interface for displaying and real-time updating the activation intensity of the movement intention in the form of a progress bar according to the generated movement intention activation intensity display signal.

10. The brain-computer interface assisted rehabilitation system based on a dual feedback closed-loop mechanism according to claim 9, characterized in that, The graphical user interface is also used to display the electrical stimulation parameters, EEG signals, EMG signals, experimental parameters, experimental paradigms, and system status to the user.

Images