Gait phase sensory guided lower limb motor imagery brain-computer interface method and system

Abstract

The gait phase feeling guided lower limb motor imagination brain-computer interface method and system belong to the technical field of brain-computer interface, and the present application is to solve the problem that the existing lower limb motor imagination technology is out of sync with the stimulation and motor imagination task. It comprises: an electrode patch for generating electrical stimulation; an electrical stimulator for changing the electrical stimulation level by changing the pulse width and sending the corresponding electrical stimulation signal to the electrode patch; an electroencephalogram data acquisition module for collecting electroencephalogram data during the application of electrical stimulation and preprocessing the collected data; and an upper computer module for sending electrical stimulation signal control instructions to the electrical stimulator and sequentially performing feature extraction, pattern recognition and linear discriminant analysis on the preprocessed electroencephalogram data to obtain lower limb recognition results. The present application is used for the brain-computer interface motion intention recognition of lower limb motor imagination.

Description

Technical Field

[0001] This invention relates to a brain-computer interface method and system for lower limb motor imagery guided by gait phase perception, belonging to the field of brain-computer interface technology. Background Technology

[0002] Brain-computer interface (BCI) is a technology that allows users to interact directly with computers or other external devices using brain signals. Classic BCI systems mainly include three paradigms: motor imagery (MI), event-related potentials (ERP), and steady-state visual evoked potentials (SSVEP).

[0003] For patients with lower limb motor disorders, especially lower limb paralysis, using MI-BCI for motor imagery-assisted training can reshape neural pathways. This allows the brain to directly control external mechanisms to traction the corresponding parts of the lower limbs, bringing significant benefits. Furthermore, rehabilitation medicine believes that rehabilitation training guided by active motor awareness is more effective than completely passive traction training. Therefore, lower limb MI-BCI has immense application value and practicality for patients. Clinical trials have shown a positive correlation between the level of motor function recovery and the decoding accuracy of the BCI rehabilitation training process, and incorrect feedback reduces patients' enthusiasm and participation in rehabilitation training.

[0004] Currently, motor imagery brain-computer interfaces (BCIs) primarily rely on the spatial distribution characteristics of ERD / ERS to distinguish different motor imagery tasks. For example, left- and right-hand motor imagery requires precise control of the upper limbs (such as the arms and fingers) to map onto the outer surface of the cerebral cortex, occupying a large area of ​​the motor cortex. The lower limbs, however, are mapped into cortical sulci, occupying less space. Furthermore, there are significant individual differences in motor imagery abilities among different subjects. Therefore, completing lower limb motor imagery tasks is difficult for some subjects. Both patients and healthy participants exhibit "BCI blindness," with decoding accuracy remaining below 70% even after training.

[0005] To improve the performance of MI-BCI, existing techniques include applying functional electrical stimulation (FES) to the lower limbs before imagining, allowing subjects to experience the sensation of muscle contraction and helping them to better perform motor imagery. However, a problem with this technique is that the stimulation and the motor imagery task are out of sync, making it difficult for subjects to associate the muscle illusion with the motor imagery task. Summary of the Invention

[0006] The purpose of this invention is to solve the problem of asynchronous stimulation and motor imagery tasks in existing lower limb motor imagery technologies, and to provide a brain-computer interface method and system for lower limb motor imagery guided by gait phase sensing.

[0007] The gait phase sensing-guided lower limb motor imagery brain-computer interface method of the present invention comprises:

[0008] S1. Place the two pairs of electrode patches on the superficial peroneal nerve and sural nerve of the lower limb, respectively;

[0009] S2. The sensory electrical stimulation level of the electrode patch is phase-coded according to the phase of the imagined gait movement. Sensory electrical stimulation is applied to the electrode patch according to the phase-coded sensory electrical stimulation level during the gait movement imagination.

[0010] S3. Under the guidance of sensory electrical stimulation, perform gait movement imagery, and collect and preprocess EEG data;

[0011] S4. Use the common space pattern algorithm to extract features and recognize patterns from the preprocessed EEG data to obtain the EEG feature vector for a single task.

[0012] S5. Input the single-task EEG feature vector into the linear discriminant analysis to train the classifier, and use the ten-fold cross-validation strategy to identify the lower limb movement intention through linear discriminant analysis.

[0013] Preferably, the method described in S2 for phase encoding the sensory electrical stimulation level of the electrode patch based on the phase of the imagined gait movement, and for applying sensory electrical stimulation to the electrode patch during gait movement imagination based on the phase-encoded sensory electrical stimulation level, includes:

[0014] S2-1. Map different phases of imagined gait movements to electrode patches to obtain corresponding electrical stimulation levels;

[0015] S2-2. Scale the phase time of the imagined gait movements to obtain the time ratio corresponding to each phase between imagined gait movements;

[0016] S2-3. Based on the time ratio obtained in S2-2, the electrical stimulation level obtained in S2-1 is simultaneously applied to the electrode patch during gait imagination.

[0017] Preferably, the specific method for synchronously applying the electrical stimulation level obtained in S2-1 to the electrode patch during gait imagery includes: changing the electrical stimulation level by changing the pulse width of the electrical stimulator.

[0018] Preferably, the phases of the imagined gait movement described in S2 include: heel touchdown, full foot strike, heel liftoff, and swing.

[0019] Preferably, the specific method for gait movement visualization guided by sensory electrical stimulation as described in S3 includes:

[0020] Apply a weak electrical stimulation of level 1 to the electrode patch at the superficial peroneal nerve and a strong electrical stimulation of level 2 to the electrode patch at the sural nerve. Imagine the gait movement as the phase from a static standing position to the heel touching the ground.

[0021] Apply a strong electrical stimulation of level 2 to the electrode patch at the superficial peroneal nerve and apply a strong electrical stimulation of level 2 to the electrode patch at the sural nerve. Imagine the gait movement as a transition from the heel-to-ground phase to the full-foot-to-ground phase.

[0022] Apply a strong electrical stimulation of level 2 to the electrode patch at the superficial peroneal nerve and a weak electrical stimulation of level 1 to the electrode patch at the sural nerve. Imagine the gait movement as the transition from the full foot strike phase to the heel lift-off phase.

[0023] Apply a weak electrical stimulation of level 1 to the electrode patch at the superficial peroneal nerve and a weak electrical stimulation of level 1 to the electrode patch at the sural nerve. Imagine the gait movement as the transition from the heel-off phase to the swing phase.

[0024] Preferably, the specific method for preprocessing EEG data as described in S3 includes:

[0025] The acquired EEG data were filtered using a 5-40Hz bandpass filter.

[0026] A 50Hz notch filter is used to remove power frequency interference;

[0027] Common-average reference spatial filtering is used, and the frequency is downsampled to 200Hz;

[0028] Data from 0-5 seconds after the start of the task is extracted, and eye movement and muscle artifacts are removed using an independent component analysis algorithm to complete the preprocessing.

[0029] Preferably, the specific method for obtaining the EEG feature vector for a single task as described in S4 includes:

[0030] The common spatial pattern algorithm is used to calculate the common spatial pattern projection matrix W of the EEG data;

[0031] The EEG data X is transformed using a spatial filter to obtain the spatially filtered EEG data Z:

[0032] Z = W T X;

[0033] Furthermore, the EEG feature vector for a single task is obtained:

[0034]

[0035] Among them, f p This represents the eigenvector, var() represents calculating the variance, and k represents a constant k=1.

[0036] Z p Let p = 1, 2, ..., 2m, m = 3, where p represents the first m rows and the last m rows of the EEG signal after spatial filtering. The first m rows correspond to the m largest eigenvalues, and the last m rows correspond to the m smallest eigenvalues.

[0037] j represents an intermediate variable, j = 1, 2, ..., 2m, where 2 represents the pattern of gait movement imagination.

[0038] The gait phase sensing-guided lower limb motor imagery brain-computer interface system of the present invention comprises:

[0039] Electrode patches are used to generate electrical stimulation;

[0040] An electrical stimulator changes the level of electrical stimulation by altering the pulse width, and sends the corresponding level of electrical stimulation signal to the electrode pads.

[0041] The EEG data acquisition module collects EEG data during the application of electrical stimulation and preprocesses the collected data.

[0042] The host computer module sends electrical stimulation signal control commands to the electrical stimulator, and sequentially performs feature extraction, pattern recognition, and linear discriminant analysis on the preprocessed EEG data to obtain lower limb recognition results.

[0043] The gait phase sensing-guided lower limb motor imagery brain-computer interface method and system proposed in this invention have the following advantages:

[0044] 1. A gait cycle of the lower limb is divided into four gait action phases. Sequential sensory electrical stimulation is applied based on the gait phase sequence encoding obtained from different gait action phases, which provides guidance in the process of gait imagination, so that the corresponding imagined gait action is completed during the electrical stimulation.

[0045] 2. Under the guidance of sequential sensory electrical stimulation, the accuracy of classification was improved, with an average of 82%. Attached Figure Description

[0046] Figure 1 This is a flowchart of the brain-computer interface method and system for lower limb motor imagery guided by gait phase sensing as described in this invention;

[0047] Figure 2 This is a schematic diagram showing the placement position of the electrode patch described in this invention;

[0048] Figure 3This is a schematic diagram of gait movement visualization under sensory electrical stimulation;

[0049] Figure 4 This is a schematic diagram showing the classification results of lower limb motor imagery of eight test subjects under three paradigms based on the gait phase proposed in this invention, traditional textual cues, and constant sensory electrical stimulation applied during motor imagery. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0052] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but this is not intended to limit the scope of the invention.

[0053] Example 1:

[0054] The following is combined with Figures 1-3 This embodiment describes a gait phase sensing-guided lower limb motor imagery brain-computer interface method, which includes:

[0055] S1. Place the two pairs of electrode patches on the superficial peroneal nerve and sural nerve of the lower limb, respectively;

[0056] S2. The sensory electrical stimulation level of the electrode patch is phase-coded according to the phase of the imagined gait movement. Sensory electrical stimulation is applied to the electrode patch according to the phase-coded sensory electrical stimulation level during the gait movement imagination.

[0057] S3. Under the guidance of sensory electrical stimulation, perform gait movement imagery, and collect and preprocess EEG data;

[0058] S4. Use the common space pattern algorithm to extract features and recognize patterns from the preprocessed EEG data to obtain the EEG feature vector for a single task.

[0059] S5. Input the single-task EEG feature vector into the linear discriminant analysis to train the classifier, and use the ten-fold cross-validation strategy to identify the lower limb movement intention through linear discriminant analysis.

[0060] Furthermore, the method described in S2 for phase encoding the sensory electrical stimulation level of the electrode patch based on the phase of the imagined gait movement, and for applying sensory electrical stimulation to the electrode patch during gait movement imagination based on the phase-encoded sensory electrical stimulation level, includes:

[0061] S2-1. Map different phases of imagined gait movements to electrode patches to obtain corresponding electrical stimulation levels;

[0062] S2-2. Scale the phase time of the imagined gait movements to obtain the time ratio corresponding to each phase between imagined gait movements;

[0063] S2-3. Based on the time ratio obtained in S2-2, the electrical stimulation level obtained in S2-1 is simultaneously applied to the electrode patch during gait imagination.

[0064] Furthermore, the specific method described in S2-3 for synchronously applying the electrical stimulation level obtained in S2-1 to the electrode patch during gait motion visualization includes: changing the electrical stimulation level by changing the pulse width of the electrical stimulator.

[0065] Furthermore, the phases of the imagined gait described in S2 include: heel strike, full foot strike, heel lift-off, and swing.

[0066] Furthermore, the specific method for gait imagery guided by sensory electrical stimulation, as described in S3, includes:

[0067] Apply a weak electrical stimulation of level 1 to the electrode patch at the superficial peroneal nerve and a strong electrical stimulation of level 2 to the electrode patch at the sural nerve. Imagine the gait movement as the phase from a static standing position to the heel touching the ground.

[0068] Apply a strong electrical stimulation of level 2 to the electrode patch at the superficial peroneal nerve and apply a strong electrical stimulation of level 2 to the electrode patch at the sural nerve. Imagine the gait movement as a transition from the heel-to-ground phase to the full-foot-to-ground phase.

[0069] Apply a strong electrical stimulation of level 2 to the electrode patch at the superficial peroneal nerve and a weak electrical stimulation of level 1 to the electrode patch at the sural nerve. Imagine the gait movement as the transition from the full foot strike phase to the heel lift-off phase.

[0070] Apply a weak electrical stimulation of level 1 to the electrode patch at the superficial peroneal nerve and a weak electrical stimulation of level 1 to the electrode patch at the sural nerve. Imagine the gait movement as the transition from the heel-off phase to the swing phase.

[0071] Furthermore, the specific methods for preprocessing EEG data described in S3 include:

[0072] The acquired EEG data were filtered using a 5-40Hz bandpass filter.

[0073] A 50Hz notch filter is used to remove power frequency interference;

[0074] Common-average reference spatial filtering is used, and the frequency is downsampled to 200Hz;

[0075] Data from 0-5 seconds after the start of the task is extracted, and eye movement and muscle artifacts are removed using an independent component analysis algorithm to complete the preprocessing.

[0076] Furthermore, the specific method for obtaining the EEG feature vector for a single task as described in S4 includes:

[0077] The common spatial pattern algorithm is used to calculate the common spatial pattern projection matrix W of the EEG data;

[0078] The EEG data X is transformed using a spatial filter to obtain the spatially filtered EEG data Z:

[0079] Z = W T X;

[0080] Furthermore, the EEG feature vector for a single task is obtained:

[0081]

[0082] Among them, f p This represents the eigenvector, var() represents calculating the variance, and k represents a constant k=1.

[0083] Z p Let p = 1, 2, ..., 2m, m = 3, where p represents the first m rows and the last m rows of the EEG signal after spatial filtering. The first m rows correspond to the m largest eigenvalues, and the last m rows correspond to the m smallest eigenvalues.

[0084] j represents an intermediate variable, j = 1, 2, ..., 2m, where 2 represents the pattern of gait movement imagination.

[0085] Example 2:

[0086] The gait phase sensing-guided lower limb motor imagery brain-computer interface system described in this embodiment is used to implement the gait phase sensing-guided lower limb motor imagery brain-computer interface method, and it includes:

[0087] Electrode patches are used to generate electrical stimulation;

[0088] An electrical stimulator changes the level of electrical stimulation by altering the pulse width, and sends the corresponding level of electrical stimulation signal to the electrode pads.

[0089] The EEG data acquisition module collects EEG data during the application of electrical stimulation and preprocesses the collected data.

[0090] The host computer module sends electrical stimulation signal control commands to the electrical stimulator, and sequentially performs feature extraction, pattern recognition, and linear discriminant analysis on the preprocessed EEG data to obtain lower limb recognition results.

[0091] In this invention, two pairs of electrode patches are placed on the superficial peroneal nerve and the sural nerve of the lower limb, respectively, such as... Figure 2 As shown, the pulse width of the electrical stimulation was adjusted based on the subject's feedback, and two distinguishable levels of electrical stimulation were found at each electrode. The electrical stimulation frequency was 100 Hz, and the level of electrical stimulation was adjusted by the pulse width, which could be set to be adjustable from 50 μs to 1000 μs. The waveform was a bidirectional square wave, and the current intensity was set to 10 mA.

[0092] Subjects performed motor imagery guided by gait phase sequence encoded sensory electrical stimulation, and EEG data were collected. The experimental paradigm is as follows:

[0093] During the experiment, participants sat quietly in a chair about 1 meter away from the screen, keeping their hands and feet relaxed and minimizing eye movements. The task mode was as follows: Figure 2 As shown, it is divided into 4 stages, with a total time of 12 seconds.

[0094] The first stage is the preparation period. A "+" sign appears in the center of the screen and lasts for 3 seconds, indicating that the experiment has begun. During this period, the subject remains relaxed and motionless.

[0095] The second stage is the cueing period. When the "+" sign appears, the "motor imagery" / "resting" cue appears, and the subject is given a pre-stimulation. A weak Level 1 stimulus is given at the two electrode patches for 1 second, prompting the subject to prepare for the motor imagery task to begin and to adapt to the sensation of electrical stimulation.

[0096] The third stage is the imagination phase: In the case of a "motor imagination" task, the time of one gait cycle is stretched to 5 seconds according to the proportion of gait phase time. During the stimulation period, sensory electrical stimulation coded with gait phase sequence is given as guidance, and at the same time, the screen provides guidance with numbers 1, 2, 3, 4 according to the sequence. The subject follows the sensory electrical stimulation guidance to imagine the corresponding gait phase movements. Figure 3 As shown, the details are as follows:

[0097] “1”: 0-1s, apply a weak stimulus of level 1 to the superficial peroneal nerve and a strong stimulus of level 2 to the sural nerve, imagine the movement from a static standing position to the heel touching the ground phase;

[0098] “2”: 1-1.6s, apply strong stimulation of grade 2 at the superficial peroneal nerve and strong stimulation of grade 2 at the sural nerve, imagine the movement from the heel-to-ground phase to the full-foot-to-ground phase.

[0099] “3”: 1.6-3.4s, apply strong stimulation of grade 2 to the superficial peroneal nerve and weak stimulation of grade 1 to the sural nerve, imagine the movement from the full foot landing phase to the heel lift-off phase;

[0100] “4”: 3.4-5s, apply a weak level 1 stimulus to the superficial peroneal nerve and a weak level 1 stimulus to the sural nerve, imagine the movement from the heel-off phase to the swing phase.

[0101] If it is a "resting" task, the sensory electrical stimulation is still applied according to the above coding, but there are no numerical prompts on the screen. Subjects should try not to pay attention to the sensory electrical stimulation on their feet and remain at rest.

[0102] The above process consisted of 5 sets of experiments, each containing 20 tasks, with 10 repetitions each of lower limb motor imagery and resting tasks. A total of 100 EEG signals were collected from these 100 experimental tasks, resulting in 100 EEG data samples. To ensure the quality of the collected EEG signals, a rest period of approximately 3 minutes was included in each set of experiments to reduce the impact of fatigue.

[0103] In this invention, the specific method for collecting EEG data is as follows: using the Neuroscan64 EEG acquisition system, with the forehead as the ground and the tip of the nose as the reference, the EEG sampling frequency is 1000Hz, and the bandpass filter is 0.5-100Hz.

[0104] In this invention, the original data is filtered using a 5-40Hz bandpass filter and a 50Hz notch filter to remove power frequency interference. Then, common average reference space filtering is applied, and the data is downsampled to 200Hz. The data X from 0-5s after the start of the task is extracted for subsequent processing.

[0105] The common spatial pattern algorithm is used to calculate the common spatial pattern projection matrix W on the EEG data, and the preprocessed EEG signal is transformed by a spatial filter to obtain Z = W. T X, X represents preprocessed EEG data, and T represents transpose. Where Z... p (p = 1, 2, ..., 2m) is the vector of the first m rows (corresponding to the m largest eigenvalues) and the last m rows (corresponding to the m smallest eigenvalues) in the signal after spatial filtering. The feature extraction for a single task is shown in the following formula:

[0106]

[0107] Where VAR represents the solution of variance; fp is the extracted eigenvector; m = 3; k = 1, 2 represent the imagined pattern.

[0108] In this invention, a ten-fold cross-validation strategy is used to complete classification and recognition through linear discriminant analysis, thereby improving performance.

[0109] like Figure 4 As shown, the horizontal axis represents the classification results of lower limb motor imagery guided by gait phase sequence encoding sensory electrical stimulation for eight subjects. Compared with the traditional text-based motor imagery paradigm and the control group which applied constant sensory electrical stimulation during motor imagery, the experimental results show that the classification accuracy guided by gait phase sequence encoding sensory electrical stimulation is the highest, with a mean of 82%, which is about 6% higher than the traditional text-based paradigm. One-way repeated measures ANOVA showed that there was a significant difference in classification accuracy among the three paradigms (F(2,14)=3.982, p=0.043), and the new method was significantly different from the traditional text-based motor imagery paradigm (p=0.016).

[0110] While the invention has been described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the invention. Therefore, it should be understood that many modifications can be made to the exemplary embodiments, and other arrangements can be designed without departing from the spirit and scope of the invention as defined by the appended claims. It should be understood that different dependent claims and features described herein can be combined in ways different from those described in the original claims. It is also understood that features described in conjunction with individual embodiments can be used in other described embodiments.

Claims

1. A brain-computer interface method for lower limb motor imagery guided by gait phase sensing, characterized in that it include: S1. Place the two pairs of electrode patches on the superficial peroneal nerve and sural nerve of the lower limb, respectively; S2. The sensory electrical stimulation level of the electrode patch is phase-coded according to the phase of the imagined gait movement. Sensory electrical stimulation is applied to the electrode patch according to the phase-coded sensory electrical stimulation level during the gait movement imagination. S3. Under the guidance of sensory electrical stimulation, perform gait movement imagery, and collect and preprocess EEG data; S4. Use the common space pattern algorithm to extract features and recognize patterns from the preprocessed EEG data to obtain the EEG feature vector for a single task. S5. Input the single-task EEG feature vector into the linear discriminant analysis to train the classifier, and use the ten-fold cross-validation strategy to identify the lower limb movement intention through linear discriminant analysis. The method described in S2 for phase encoding the sensory electrical stimulation level of the electrode patch based on the phase of the imagined gait movement, and for applying sensory electrical stimulation to the electrode patch during gait movement imagination based on the phase-encoded sensory electrical stimulation level, includes: S2-1. Map different phases of imagined gait movements to electrode patches to obtain corresponding electrical stimulation levels; S2-2. Scale the phase time of the imagined gait movements to obtain the time ratio corresponding to each phase between imagined gait movements; S2-3. Based on the time ratio obtained in S2-2, the electrical stimulation level obtained in S2-1 is simultaneously applied to the electrode patch during gait imagination. The phases of the imagined gait movement described in S2 include: heel strike, full foot strike, heel lift-off, and swing; The specific methods for gait imagery guided by sensory electrical stimulation, as described in S3, include: Apply a weak electrical stimulation of level 1 to the electrode patch at the superficial peroneal nerve and a strong electrical stimulation of level 2 to the electrode patch at the sural nerve. Imagine the gait movement as the phase from standing still to the heel touching the ground. Apply a strong electrical stimulation of level 2 to the electrode patch at the superficial peroneal nerve and apply a strong electrical stimulation of level 2 to the electrode patch at the sural nerve. Imagine the gait movement as the transition from the heel-to-ground phase to the full-foot-to-ground phase. Apply a strong electrical stimulation of level 2 to the electrode patch at the superficial peroneal nerve and a weak electrical stimulation of level 1 to the electrode patch at the sural nerve. Imagine the gait movement as the transition from the full foot strike phase to the heel lift-off phase. Apply a weak electrical stimulation of level 1 to the electrode patch at the superficial peroneal nerve and a weak electrical stimulation of level 1 to the electrode patch at the sural nerve. Imagine the gait movement as the transition from the heel-off phase to the swing phase.

2. The brain-computer interface method for lower limb motor imagery guided by gait phase sensing according to claim 1, characterized in that, The specific method described in S2-3 for synchronously applying the electrical stimulation level obtained in S2-1 to the electrode patch during gait imagery includes: changing the electrical stimulation level by changing the pulse width of the electrical stimulator.

3. The brain-computer interface method for lower limb motor imagery guided by gait phase sensing according to claim 1, characterized in that, The specific methods for preprocessing EEG data described in S3 include: The acquired EEG data were filtered using a 5-40Hz bandpass filter. A 50Hz notch filter is used to remove power frequency interference; Common-average reference spatial filtering is used, and the frequency is downsampled to 200Hz; Data from 0-5 seconds after the start of the task is extracted, and eye movement and muscle artifacts are removed using an independent component analysis algorithm to complete the preprocessing.

4. The brain-computer interface method for lower limb motor imagery guided by gait phase sensing according to claim 3, characterized in that, The specific methods for obtaining single-task EEG feature vectors as described in S4 include: The common spatial pattern algorithm is used to calculate the common spatial pattern projection matrix of the obtained EEG data. ; Spatial filters are used to process EEG data The EEG data was transformed to obtain spatially filtered data. : ； Furthermore, the EEG feature vector for a single task is obtained: ； in, Represents the eigenvector. This indicates the solution for variance. Represents a constant , This indicates the anterior part of the EEG signal after spatial filtering. Line and after The vector of rows, , ,forward The row vector corresponds to the largest Each feature value, then The row vector corresponds to the smallest 1 eigenvalue, Indicates intermediate variables. ,2 represents the pattern of gait movement imagination.

5. A gait phase sensing-guided lower limb motor imagery brain-computer interface system, wherein the system is used to implement the gait phase sensing-guided lower limb motor imagery brain-computer interface method of claim 1, characterized in that, It includes: Electrode patches are used to generate electrical stimulation; An electrical stimulator changes the level of electrical stimulation by altering the pulse width, and sends the corresponding level of electrical stimulation signal to the electrode pads. The EEG data acquisition module collects EEG data during the application of electrical stimulation and preprocesses the collected data. The host computer module sends electrical stimulation signal control commands to the electrical stimulator, and sequentially performs feature extraction, pattern recognition, and linear discriminant analysis on the preprocessed EEG data to obtain lower limb recognition results.

Images