Motor imagery triggered electrical stimulation synchronous closed-loop multi-modal neuromodulation system

Abstract

The present application relates to a kind of based on motor imagination trigger's electric stimulation synchronous closed loop multimodal neuroregulation system, belong to medical rehabilitation engineering and neural engineering technical field.The system includes: electroencephalogram signal acquisition and processing module, for real-time acquisition user sensory motor cortex electroencephalogram signal, and extract the event-related desynchronization features of motor imagination task in the electroencephalogram signal by time-frequency analysis model;Motor intention recognition and trigger module generates synchronous trigger signal;Functional electric stimulation module, through multi-channel electrode device drive surface electrode output electric stimulation pulse to target area, drive target action;Transcutaneous auricular vagus nerve electric stimulation module, output and motor intention duration synchronous stimulation pulse;Central control and synchronization module, for coordination control timing.Formed " intention recognition-synchronous control-assisted movement" complete closed loop, provides a kind of efficient, accurate, individualized neural engineering solution.

Description

Technical Field

[0001] This invention belongs to the fields of medical rehabilitation engineering and neuroengineering technology, specifically relating to a synchronous closed-loop multimodal neuromodulation system based on motor imagery-triggered electrical stimulation. Background Technology

[0002] Stroke is the leading cause of long-term motor dysfunction in adults, often leaving users with severe upper limb motor impairments that significantly impact their ability to live independently. Traditional rehabilitation training methods often experience plateaus in effectiveness for users in the chronic phase.

[0003] Brain-computer interface (BCI) technology based on motor imagery offers new insights for stroke rehabilitation. MI-BCI technology identifies a user's motor intentions by decoding the electroencephalogram (EEG) signals generated during motor imagery (such as rhythmic event-related desynchronization). When combined with functional electrical stimulation (FES), a closed loop of "intention-action-feedback" (MI-BCI-FES) can be formed, where the FES drives actual or assisted movements in the paralyzed limbs when the user attempts motor imagery. This closed-loop feedback is believed to help activate damaged neural circuits, promote neural plasticity, and reconstruct motor function.

[0004] However, the effectiveness of MI-BCI-FES alone is still limited by the brain's own neuroplasticity potential after stroke. Research indicates that optimal neuroplasticity requires not only repeated activation of local cortical-specific activity but also the participation of whole-brain neuromodulation systems. For example, the release of neuromodulations such as norepinephrine and acetylcholine can create a "high plasticity window" conducive to synaptic remodeling. Percutaneous vagus nerve stimulation (PVS), as a non-invasive neuromodulation technique, can activate the locus coeruleus-norepinephrine system and the basal forebrain-acetylcholine system in the brain by stimulating the vagus nerve branches in the ear, promoting the release of these key neuromodulations and theoretically broadly enhancing the brain's learning and plasticity capabilities.

[0005] Current research has attempted to integrate taVNS with rehabilitation training, but primarily employs a continuous background stimulation pattern applied before, after, or during training. While this continuous stimulation can enhance baseline cortical excitability and user attention, its neuromodulation release is diffuse and persistent, lacking temporal specificity. It cannot precisely coincide temporally with the transient cortical-specific activation events induced by MI-BCI-FES and driven by specific motor intentions. This means that its diffuse and persistent neuromodulation effect is difficult to establish a precise temporal correlation with transient cortical-specific activation events, potentially wasting stimulation energy during non-learning periods and failing to achieve the 'neuromodulation gating' effect—that is, failing to provide the strongest gain signal at the moment when plasticity change is most needed.

[0006] Therefore, providing a closed-loop neuromodulation system and method based on synchronous triggering of motor intention to precisely couple the global neuromodulation effect of taVNS with the local specific activation of MI-BCI-FES in time to maximize synergistic effect is a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0007] To address the aforementioned problems in the existing technology, this invention provides a synchronous closed-loop multimodal neuromodulation system based on motor imagery-triggered electrical stimulation. The objective of this invention can be achieved through the following technical solutions: A synchronous closed-loop multimodal neuromodulation system based on motor imagery-triggered electrical stimulation, comprising: The EEG signal acquisition and processing module is used to acquire EEG signals from the user's sensorimotor cortex in real time and extract event-related desynchronization features of the motor imagery task from the EEG signals through a time-frequency analysis model. The motion intention recognition and triggering module is connected to the EEG signal acquisition and processing module. It presets the recognition threshold of the event-related desynchronization feature. When the signal strength of the event-related desynchronization feature exceeds the recognition threshold, it is determined to be a valid motion intention and generates a synchronization trigger signal to be sent to the functional electrical stimulation module and the transcutaneous vagus nerve electrical stimulation module. The functional electrical stimulation module receives the synchronous trigger signal from the motion intention recognition and triggering module, and drives the surface electrodes to output electrical stimulation pulses to the target area through the multi-channel electrode device to drive the target movement. The percutaneous vagus nerve electrical stimulation module is activated after receiving the synchronous trigger signal from the motor intention recognition and triggering module, and outputs stimulation pulses synchronized with the duration of the motor intention. The central control and synchronization module is used to coordinate the control timing. Based on the hardware synchronization triggering mechanism, the synchronization triggering signal can simultaneously activate the functional electrical stimulation module and the transcutaneous vagus nerve electrical stimulation module.

[0008] Specifically, the time-frequency analysis model includes: using short-time Fourier transform combined with a common spatial mode algorithm to perform time-frequency decomposition and spatial filtering on the EEG signal, and extracting event-related desynchronization features of μ rhythm and β rhythm; the time-frequency analysis model also includes individualized calibration, collecting EEG data of the user in resting state and motor imagery state during the basic training stage, and establishing individualized spatial filters and feature weight matrices.

[0009] Specifically, the motor imagery task is the user's motor imagery of extending the wrist, clenching the fist, or extending the fingers of the upper limbs; the motor imagery task is triggered by visual or auditory cues and includes a time sequence design that alternates between task periods and rest periods.

[0010] Specifically, the event-related desynchronization features include the magnitude or percentage decrease in the power of the user's sensorimotor cortex of the μ rhythm and / or β rhythm relative to the resting baseline during motor imagery, wherein the magnitude or percentage decrease in power is calculated in real time and used as a quantitative indicator for motor intention recognition.

[0011] Specifically, the synchronization trigger signal is a transistor-to-transistor logic level pulse signal or a digital trigger signal, which is generated by the motion intention recognition and triggering module when determining a valid motion intention, and distributed to the functional electrical stimulation module and the transcutaneous vagus nerve electrical stimulation module through the central control and synchronization module.

[0012] Specifically, the multi-channel electrode device includes at least one pair of surface electrode patches, which are arranged according to the movement points of the target muscle and connected to the output port of the functional electrical stimulation module via wires. The multi-channel electrode device supports at least two independent output channels for simultaneously or alternately stimulating different muscle groups. Each channel independently controls the stimulation intensity and pulse timing. The multi-channel electrode device also has a real-time impedance monitoring function, which automatically alarms and pauses output when the electrode-skin contact impedance exceeds a preset threshold.

[0013] Specifically, the selection of the target area is individualized based on the user's functional impairment type and applicable stage.

[0014] Specifically, the target movements include wrist dorsiflexion, finger extension, or thumb abduction; the range of motion of the target movements is controlled by the output parameters of the functional electrical stimulation module, including stimulation frequency, pulse width, and stimulation intensity, to achieve auxiliary, non-completely substitutive motor output.

[0015] Specifically, the duration of the motor intention is monitored in real time by the motor intention recognition and triggering module, starting from the moment when the event-related desynchronization feature signal exceeds the recognition threshold and ending when the event-related desynchronization feature signal falls back below the threshold or reaches the maximum allowable duration; the stimulation output time of the percutaneous ear vagus nerve electrical stimulation module is completely synchronized with the duration of the motor intention.

[0016] Specifically, the hardware synchronization triggering mechanism is executed by a synchronization controller, which is electrically connected to the motion intention recognition and triggering module, the functional electrical stimulation module, and the transcutaneous vagus nerve stimulation module. The synchronization controller receives a single trigger pulse from the motion intention recognition and triggering module and forwards the single trigger pulse to both the functional electrical stimulation module and the transcutaneous vagus nerve stimulation module simultaneously through hardware-level signal distribution, thereby achieving synchronous activation of the two modules. Signal distribution is achieved using a parallel port or a dedicated synchronization interface box. Both the functional electrical stimulation module and the transcutaneous vagus nerve stimulation module have external trigger input interfaces and support edge triggering mode.

[0017] Specifically, the setting and adaptive adjustment of the recognition threshold are implemented using a dynamic gating mechanism, which includes a fixed threshold mode in the basic training phase and a dynamic adaptive mode in the reinforcement training phase. The basic training phase collects users' resting-state EEG data and motor imagery task data. The mean and standard deviation of event-related desynchronization features are calculated in the μ rhythm 8-13Hz and / or β rhythm 13-30Hz frequency bands, respectively. The initial recognition threshold is set according to the mean, standard deviation and the weight of each frequency band set by the user's feature differences. The reinforcement training phase uses a sliding recognition window to correct the recognition threshold when the number of training attempts exceeds a preset upper limit. Specific correction methods include: threshold correction based on the number of successful triggers, compensation based on the non-stationarity of EEG signals, and adaptive optimization based on the signal-to-noise ratio.

[0018] Furthermore, the recognition threshold also forms a closed-loop interaction with the output of the transcutaneous vagus nerve electrical stimulation module: when the signal-to-noise ratio of the event-related desynchronization feature in a single frequency band is continuously lower than the set threshold, the system automatically switches to the multimodal feature fusion mode, introduces motion-related cortical potentials or slow cortical potentials as auxiliary recognition features, constructs a multidimensional feature space, and uses a support vector machine or linear discriminant analysis classifier to generate the fusion threshold, thereby maintaining the accuracy of motion intention recognition.

[0019] The beneficial effects of this invention are as follows: This multimodal neuromodulation system, triggered by motor imagery and synchronized with electrical stimulation, achieves precise temporal coupling between transcutaneous vagal nerve electrical stimulation and functional electrical stimulation at the moment of motor intention generation. This effectively solves the problem of temporal misalignment between neuromodulation and cortical-specific activation events in existing technologies. The system utilizes a hardware-level synchronous triggering mechanism to ensure that the functional electrical stimulation module and the transcutaneous vagal nerve electrical stimulation module activate synchronously within milliseconds of receiving the motor intention recognition signal. This allows for precise temporal overlap between global neuromodulation effects and local motor cortical activation, maximizing the "neuromodulation gating" effect and providing optimal neurochemical environmental support during the window of synaptic plasticity when it is most likely to occur.

[0020] By employing a dynamic gating mechanism to adaptively adjust the recognition threshold, the system balances stability during basic training and adaptability during intensive training. A multimodal feature fusion strategy effectively addresses the non-stationarity of EEG signals, ensuring reliable operation across different rehabilitation stages and user groups. The percutaneous vagus nerve electrical stimulation module's stimulation output duration is perfectly synchronized with the duration of motor intent, avoiding energy waste and non-specific effects caused by continuous stimulation and achieving precise delivery of neuromodulation resources.

[0021] This invention organically integrates brain-computer interface decoding technology, functional electrical stimulation-assisted movement technology, and vagus nerve electrical stimulation neuromodulation technology to form a complete closed loop of "intention recognition-synchronous regulation-assisted movement". It provides an efficient, precise, and individualized neuroengineering solution for rehabilitation training of upper limb motor dysfunction after stroke, and has significant clinical application value and promotion prospects. Attached Figure Description

[0022] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0023] Figure 1 This is a schematic diagram of the data flow of the present invention. Detailed Implementation

[0024] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided to make this disclosure more comprehensive and complete, and to fully convey the concept of example embodiments to those skilled in the art. Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a full understanding of embodiments of this disclosure. However, those skilled in the art will recognize that the technical solutions of this disclosure can be practiced without one or more of the specific details, or other methods, components, apparatuses, steps, etc., can be employed. In other instances, well-known methods, apparatuses, implementations, or operations are not shown or described in detail to avoid obscuring aspects of this disclosure. The blocks shown in the drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices. The flowcharts shown in the drawings are merely illustrative and do not necessarily include all contents and operations / steps, nor do they necessarily have to be performed in the order described. For example, some operations / steps can be broken down, while others can be combined or partially combined. Therefore, the actual execution order may change depending on the actual situation.

[0025] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0026] Please see Figure 1 A closed-loop multimodal neuromodulation system based on motor imagery-triggered electrical stimulation, comprising: The EEG signal acquisition and processing module is used to acquire EEG signals from the user's sensorimotor cortex in real time and extract event-related desynchronization features of the motor imagery task from the EEG signals through a time-frequency analysis model. The motion intention recognition and triggering module is connected to the EEG signal acquisition and processing module. It presets the recognition threshold of the event-related desynchronization feature. When the signal strength of the event-related desynchronization feature exceeds the recognition threshold, it is determined to be a valid motion intention and generates a synchronization trigger signal to be sent to the functional electrical stimulation module and the transcutaneous vagus nerve electrical stimulation module. The functional electrical stimulation module receives the synchronous trigger signal from the motion intention recognition and triggering module, and drives the surface electrodes to output electrical stimulation pulses to the target area through the multi-channel electrode device to drive the target movement. The percutaneous vagus nerve electrical stimulation module is activated after receiving the synchronous trigger signal from the motor intention recognition and triggering module, and outputs stimulation pulses synchronized with the duration of the motor intention. The central control and synchronization module is used to coordinate the control timing. Based on the hardware synchronization triggering mechanism, the synchronization triggering signal can simultaneously activate the functional electrical stimulation module and the transcutaneous vagus nerve electrical stimulation module.

[0027] Specifically, the time-frequency analysis model includes: using short-time Fourier transform combined with a common spatial mode algorithm to perform time-frequency decomposition and spatial filtering on the EEG signal, and extracting event-related desynchronization features of μ rhythm and β rhythm; the time-frequency analysis model also includes individualized calibration, collecting EEG data of the user in resting state and motor imagery state during the basic training stage, and establishing individualized spatial filters and feature weight matrices.

[0028] In this embodiment, EEG signal acquisition and preprocessing are performed using an EEG acquisition rehabilitation training device. An 8-channel electrode cap is used to acquire EEG signals from the user's sensorimotor cortex (C3, C4, C2, and surrounding extended electrodes). The sampling frequency is set to 256Hz, and the signals are preprocessed using a 0.5-50Hz bandpass filter to remove power line interference (50Hz notch filter) and baseline drift.

[0029] Short-time Fourier Transform (STFT) time-frequency decomposition was performed on the preprocessed EEG signal. The time-domain EEG signal was converted into a time-frequency spectrum using STFT, and the time-frequency energy distribution of the 8-13Hz (μ rhythm) and 13-30Hz (β rhythm) frequency bands was extracted.

[0030] A co-space mode spatial filter is constructed for both μ and β rhythms. The CSP algorithm extracts the most discriminative spatial features by maximizing the variance ratio of the two task classes (resting state and motion-imagined state). Resting state covariance matrix: ; Motion imagination state covariance matrix: ; Where R0 is the spatial covariance matrix of the EEG signal of the user in a resting state (without motor imagery task), N0 is the total number of resting-state EEG data segments collected during the basic training phase, and X... 0,i Let X be the EEG data matrix for the i-th resting state trial, R1 be the spatial covariance matrix of the EEG signals of the user in the motor imagery task state, N1 be the total number of EEG data segments collected in the motor imagery task during the basic training phase, and X be the total number of segments collected in the basic training phase. 1,i Let i be the sample index and T be the transpose matrix identifier; Eigenvalue decomposition is performed on the composite covariance matrix R = R0 + R1 to construct the whitening transformation matrix; Joint diagonalization is performed on the whitened covariance matrix to extract the spatial filter matrix (selecting the first 3 pairs of co-space modes, for a total of 6 spatial filters).

[0031] The steps of feature extraction and event-related desynchronization quantization include: Project the original electrical signal through the CSP spatial filter: Where Z(t) is the CSP spatial filter projection, W T Let X(t) be the transpose of the spatial filter matrix, and let X(t) be the original electrical signal. Calculate the instantaneous power of the projected signal and compare it with the individual resting baseline power to obtain the event-related desynchronization (ERD) percentage: , where P baseline The resting-state average power, P, collected during the basic training phase task Real-time task power; During individualized calibration, based on the collected data, the system calculates an individualized CSP spatial filter matrix and feature weight vector to establish a user-specific time-frequency analysis model. This model is used consistently in subsequent reinforcement training phases, or it can be dynamically optimized according to the adaptive threshold mechanism described in weight 12.

[0032] During the intensive training phase, the time-frequency analysis model processes EEG signals in real time with a 125ms sliding window, updating the ERD feature value every 0.5 seconds. When the ERD% of μ rhythm and / or β rhythm exceeds the preset recognition threshold (determined by weight 12), it is determined as a valid motor intention, triggering the output of a synchronization control signal to the functional electrical stimulation module and the transcutaneous vagus nerve electrical stimulation module.

[0033] Specifically, the motor imagery task is the user's motor imagery of extending the wrist, clenching the fist, or extending the fingers of the upper limbs; the motor imagery task is triggered by visual or auditory cues and includes a time sequence design that alternates between task periods and rest periods.

[0034] In this embodiment, the motor imagery task employs a combination of visual and auditory cues to enhance user focus and consistency. Specifically, the visual cues are presented on the display screen and include the following sequence: at the end of the rest period, a "+" sign is displayed in the center of the screen for one second as a preparation prompt, followed by animated icons corresponding to the target movement (e.g., wrist extension animation showing the back of the hand rising, fist clenching animation showing finger flexion, and finger extension animation showing finger opening). The animated icons are displayed for four seconds to mark the task period. The auditory cues are broadcast through headphones or an external speaker, synchronized with the visual cues, including a preparation prompt "beep" and voice commands during the task period (e.g., "imagine wrist extension," "imagine fist clenching," or "imagine finger extension").

[0035] The timing design for the alternation of task and rest periods is as follows: in a single training session, the task period lasts for 4 seconds, and the rest period lasts for 8 seconds, with the task and rest periods alternating in a 1:2 time ratio. During the task period, the user is required to continuously visualize the corresponding movement, without performing any actual physical actions; during the rest period, the user is required to relax all muscles, gaze at a static image on the screen, and refrain from any visual visualization or physical activity.

[0036] To accommodate different users' cognitive abilities and fatigue levels, the duration of the task period and rest period can be individually adjusted, with the task period adjustable from 2 to 6 seconds and the rest period from 5 to 12 seconds. During the basic training phase, the system collects user performance data. If a user repeatedly fails to reach the preset recognition threshold within the task period, the system automatically extends the task period by 1 second or the rest period by 2 seconds to reduce cognitive load. Conversely, if a user repeatedly reaches the recognition threshold early within the task period and the EEG signal quality is stable, the system automatically shortens the task period by 0.5 seconds to increase training challenge.

[0037] Furthermore, to enhance the effectiveness of the motor imagery task, this embodiment introduces mirror feedback assistance during the basic training phase: during the task period, the display screen simultaneously shows a real-time mirror video of the user's affected limb. When the user performs motor imagery, the system presents a mirrored image of the affected limb in an ideal motor state, helping the user establish a clearer representation of their motor intentions. In the intensive training phase, the mirror feedback is linked to the output of the functional electrical stimulation module. When a synchronous trigger signal is generated, the corresponding limb in the mirror image simultaneously produces a visual motion effect, forming a multimodal closed loop of "motor imagery - visual feedback - electrical stimulation feedback - neuromodulation".

[0038] Through the above-mentioned motor imagery task design, the system can effectively guide users to generate stable and identifiable EEG event-related desynchronization features, laying the foundation for subsequent motor intention recognition and synchronous triggering.

[0039] Specifically, the event-related desynchronization features include the magnitude or percentage decrease in the power of the user's sensorimotor cortex of the μ rhythm and / or β rhythm relative to the resting baseline during motor imagery, wherein the magnitude or percentage decrease in power is calculated in real time and used as a quantitative indicator for motor intention recognition.

[0040] Specifically, the synchronization trigger signal is a transistor-to-transistor logic level pulse signal or a digital trigger signal, which is generated by the motion intention recognition and triggering module when determining a valid motion intention, and distributed to the functional electrical stimulation module and the transcutaneous vagus nerve electrical stimulation module through the central control and synchronization module.

[0041] In this embodiment, the synchronization controller is implemented using an Arduino Mega 2560 microcontroller in conjunction with a dedicated synchronization interface box: Input interface: Receives TTL pulse signals (5V level, pulse width 10-50ms, rising edge trigger) from the motion intention recognition and triggering module. Signal distribution: Electrical isolation is achieved through an opto-isolator (6N137), with branched outputs to the FES module and taVNS module. Time accuracy: Hardware-level signal distribution delay <1ms, time difference between two outputs <5ms Synchronous monitoring: Real-time monitoring of the actual activation time of FES and taVNS, and correction of deviations through feedback loop.

[0042] Specifically, the multi-channel electrode device includes at least one pair of surface electrode patches, which are arranged according to the movement points of the target muscle and connected to the output port of the functional electrical stimulation module via wires. The multi-channel electrode device supports at least two independent output channels for simultaneously or alternately stimulating different muscle groups. Each channel independently controls the stimulation intensity and pulse timing. The multi-channel electrode device also has a real-time impedance monitoring function, which automatically alarms and pauses output when the electrode-skin contact impedance exceeds a preset threshold.

[0043] Specifically, the selection of the target area is individualized based on the user's functional impairment type and applicable stage.

[0044] Specifically, the target movements include wrist dorsiflexion, finger extension, or thumb abduction; the range of motion of the target movements is controlled by the output parameters of the functional electrical stimulation module, including stimulation frequency, pulse width, and stimulation intensity, to achieve auxiliary, non-completely substitutive motor output.

[0045] Specifically, the duration of the motor intention is monitored in real time by the motor intention recognition and triggering module, starting from the moment when the event-related desynchronization feature signal exceeds the recognition threshold and ending when the event-related desynchronization feature signal falls back below the threshold or reaches the maximum allowable duration; the stimulation output time of the percutaneous ear vagus nerve electrical stimulation module is completely synchronized with the duration of the motor intention.

[0046] Specifically, the hardware synchronization triggering mechanism is executed by a synchronization controller, which is electrically connected to the motion intention recognition and triggering module, the functional electrical stimulation module, and the transcutaneous vagus nerve stimulation module. The synchronization controller receives a single trigger pulse from the motion intention recognition and triggering module and forwards the single trigger pulse to both the functional electrical stimulation module and the transcutaneous vagus nerve stimulation module simultaneously through hardware-level signal distribution, thereby achieving synchronous activation of the two modules. Signal distribution is achieved using a parallel port or a dedicated synchronization interface box. Both the functional electrical stimulation module and the transcutaneous vagus nerve stimulation module have external trigger input interfaces and support edge triggering mode.

[0047] Specifically, the setting and adaptive adjustment of the recognition threshold are implemented using a dynamic gating mechanism, which includes a fixed threshold mode in the basic training phase and a dynamic adaptive mode in the reinforcement training phase. The basic training phase collects users' resting-state EEG data and motor imagery task data. The mean and standard deviation of event-related desynchronization features are calculated in the μ rhythm 8-13Hz and / or β rhythm 13-30Hz frequency bands, respectively. The initial recognition threshold is set according to the mean, standard deviation and the weight of each frequency band set by the user's feature differences. The reinforcement training phase uses a sliding recognition window to correct the recognition threshold when the number of training attempts exceeds a preset upper limit. Specific correction methods include: threshold correction based on the number of successful triggers, compensation based on the non-stationarity of EEG signals, and adaptive optimization based on the signal-to-noise ratio.

[0048] Furthermore, the recognition threshold also forms a closed-loop interaction with the output of the transcutaneous vagus nerve electrical stimulation module: when the signal-to-noise ratio of the event-related desynchronization feature in a single frequency band is continuously lower than the set threshold, the system automatically switches to the multimodal feature fusion mode, introduces motion-related cortical potentials or slow cortical potentials as auxiliary recognition features, constructs a multidimensional feature space, and uses a support vector machine or linear discriminant analysis classifier to generate the fusion threshold, thereby maintaining the accuracy of motion intention recognition.

[0049] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A closed-loop multimodal neural modulation system based on motor imagery-triggered electrical stimulation, characterized in that, include: The EEG signal acquisition and processing module is used to acquire EEG signals from the user's sensorimotor cortex in real time and extract event-related desynchronization features of the motor imagery task from the EEG signals through a time-frequency analysis model. The motion intention recognition and triggering module is connected to the EEG signal acquisition and processing module. It presets the recognition threshold of the event-related desynchronization feature. When the signal strength of the event-related desynchronization feature exceeds the recognition threshold, it is determined to be a valid motion intention and generates a synchronization trigger signal to be sent to the functional electrical stimulation module and the transcutaneous vagus nerve electrical stimulation module. The functional electrical stimulation module receives the synchronous trigger signal from the motion intention recognition and triggering module, and drives the surface electrodes to output electrical stimulation pulses to the target area through the multi-channel electrode device to drive the target movement. The percutaneous vagus nerve electrical stimulation module is activated after receiving the synchronous trigger signal from the motor intention recognition and triggering module, and outputs stimulation pulses synchronized with the duration of the motor intention. The central control and synchronization module is used to coordinate the control timing. Based on the hardware synchronization triggering mechanism, the synchronization triggering signal can simultaneously activate the functional electrical stimulation module and the transcutaneous vagus nerve electrical stimulation module.

2. The system according to claim 1, characterized in that, The time-frequency analysis model includes: using short-time Fourier transform combined with a common spatial mode algorithm to perform time-frequency decomposition and spatial filtering on the EEG signal, and extracting event-related desynchronization features of μ rhythm and β rhythm; the time-frequency analysis model also includes individualized calibration, collecting EEG data of the user in resting state and motor imagery state during the basic training stage, and establishing individualized spatial filters and feature weight matrices.

3. The system according to claim 1, characterized in that, The motor imagery task involves imagining the user's upper limbs extending their wrists, clenching their fists, or extending their fingers. The motor imagery task is triggered by visual or auditory cues and includes a time-series design that alternates between task periods and rest periods.

4. The system according to claim 1, characterized in that, The event-related desynchronization features include the magnitude or percentage decrease in the power of the user's sensorimotor cortex of the μ rhythm and / or β rhythm relative to the resting baseline during motor imagery, wherein the magnitude or percentage decrease in power is calculated in real time and used as a quantitative indicator for motor intention recognition.

5. The system according to claim 1, characterized in that, The synchronization trigger signal is a transistor-to-transistor logic level pulse signal or a digital trigger signal, which is generated by the motion intention recognition and triggering module when a valid motion intention is determined, and distributed to the functional electrical stimulation module and the transcutaneous vagus nerve electrical stimulation module through the central control and synchronization module.

6. The system according to claim 1, characterized in that, The multi-channel electrode device includes at least one pair of surface electrode patches, which are arranged according to the movement points of the target muscle and connected to the output port of the functional electrical stimulation module via wires. The multi-channel electrode device supports at least two independent output channels for simultaneously or alternately stimulating different muscle groups. Each channel independently controls the stimulation intensity and pulse timing. The multi-channel electrode device also has a real-time impedance monitoring function, which automatically alarms and pauses output when the electrode-skin contact impedance exceeds a preset threshold.

7. The system according to claim 1, characterized in that, The selection of the target area is based on the user's functional impairment type and applicable stage, and is configured individually.

8. The system according to claim 1, characterized in that, The target movements include wrist dorsiflexion, finger extension, or thumb abduction; the range of motion of the target movements is controlled by the output parameters of the functional electrical stimulation module, including stimulation frequency, pulse width, and stimulation intensity, to achieve auxiliary, non-completely substitutive motor output.

9. The system according to claim 1, characterized in that, The duration of the motor intention is monitored in real time by the motor intention recognition and triggering module. The timing starts when the event-related desynchronization feature signal exceeds the recognition threshold and ends when the event-related desynchronization feature signal falls back below the threshold or reaches the maximum allowable duration. The stimulation output time of the percutaneous ear vagus nerve electrical stimulation module is completely synchronized with the duration of the motor intention.

10. The system according to claim 1, characterized in that, The hardware synchronization triggering mechanism is executed by a synchronization controller, which is electrically connected to the motion intention recognition and triggering module, the functional electrical stimulation module, and the transcutaneous vagus nerve stimulation module. The synchronization controller receives a single trigger pulse from the motion intention recognition and triggering module and forwards the single trigger pulse to both the functional electrical stimulation module and the transcutaneous vagus nerve stimulation module simultaneously through hardware-level signal distribution, thereby achieving synchronous activation of the two modules. Signal distribution is achieved using a parallel port or a dedicated synchronization interface box. Both the functional electrical stimulation module and the transcutaneous vagus nerve stimulation module have external trigger input interfaces and support edge triggering mode.

11. The system according to claim 1, characterized in that, The setting and adaptive adjustment of the recognition threshold are achieved through a dynamic gating mechanism, specifically including a fixed threshold mode in the basic training phase and a dynamic adaptive mode in the reinforcement training phase. The basic training phase collects users' resting-state EEG data and motor imagery task data. The mean and standard deviation of event-related desynchronization features are calculated in the μ rhythm 8-13Hz and / or β rhythm 13-30Hz frequency bands, respectively. The initial recognition threshold is set according to the mean, standard deviation and the weight of each frequency band set by the user's feature differences. The reinforcement training phase uses a sliding recognition window to correct the recognition threshold when the number of training attempts exceeds a preset upper limit. Specific correction methods include: threshold correction based on the number of successful triggers, compensation based on the non-stationarity of EEG signals, and adaptive optimization based on the signal-to-noise ratio.

12. The system according to claim 11, characterized in that, The recognition threshold also forms a closed-loop interaction with the output of the transcutaneous vagus nerve electrical stimulation module: when the signal-to-noise ratio of the event-related desynchronization feature in a single frequency band is continuously lower than the set threshold, the system automatically switches to the multimodal feature fusion mode, introduces motion-related cortical potentials or slow cortical potentials as auxiliary recognition features, constructs a multidimensional feature space, and uses a support vector machine or linear discriminant analysis classifier to generate the fusion threshold, thereby maintaining the accuracy of motion intention recognition.

Images