Upper and lower limb active and passive rehabilitation training method and system based on motor imagery brain-computer interface

Through the active passive rehabilitation training method of upper and lower limbs based on the sports imagination brain-computer interface, using EEG signals and myoelectric signals, combined with game experimental paradigms and classification models, the training parameters are adjusted, and the existing system is unable to effectively utilize brain nerve participation and The problem of poor rehabilitation of lower limbs has achieved more efficient rehabilitation results.

CN118986690AActive Publication Date: 2024-11-22UNIV OF ELECTRONICS SCI & TECH OF CHINA

Patent Information

Application Number
CN202411030443.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-30
Publication Date
2024-11-22
Estimated Expiration
2044-07-30

AI Technical Summary

Technical Problem

The existing upper and lower limb rehabilitation training system cannot effectively utilize brain nerves to participate, resulting in limited rehabilitation effects for stroke patients, especially the poor effect of lower limb rehabilitation training, and the system cannot adjust according to the patient's training situation.

Method used

Using the active passive rehabilitation training method of upper and lower limbs based on the brain-computer interface of the exercise imagination, we can create a game experimental paradigm of active mode and passive mode by collecting EEG signals and electromyography signals, and use the pre-trained exercise imagination classification model to judge the patient's movement. Intent, generate control instructions, adjust training parameters and auxiliary functions, including changing impedance and assist levels to improve the patient's subjective agency and muscle strength.

Benefits of technology

It significantly improves the overall rehabilitation effect of stroke patients, and improves the subjective initiative and rehabilitation effect by enhancing central nervous system stimulation feedback and muscle strength.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118986690A_ABST
    Figure CN118986690A_ABST
Patent Text Reader

Abstract

The invention relates to the field of rehabilitation, in particular to an upper and lower limb active and passive rehabilitation training method and system based on a motor imagery brain-computer interface, and the method comprises the steps: creating two game experiment normal forms which are matched with the method and are used for guiding a patient to train in an active mode and a passive mode; eEG signals and electromyographic signals are collected through a brain-computer interface, the motion intention of a stroke patient is extracted through the EEG signals, two game experiment paradigms are matched to guide the patient to conduct active or passive rehabilitation training, and the rehabilitation training process of the patient is accelerated. The motor imagery is evaluated by using the electromyographic signals in combination with expert experience, the parameter setting of the experimental normal form and the parameter setting of the motion assisting function in the corresponding mode are adjusted according to the evaluation result, the parameter setting comprises the step of gradually changing parameters such as the impedance level and the assisting level of the system in the training process, the subjective initiative of a patient is improved, and the training efficiency is improved. The muscle strength of the patient is enhanced while central nervous stimulation feedback of the patient is enhanced, and the overall rehabilitation effect of the patient is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of rehabilitation, and in particular to an active and passive rehabilitation training method and system for upper and lower limbs based on motor imagery brain-computer interface. Background Art

[0002] Stroke is the leading cause of death among Chinese residents. It has the characteristics of high morbidity, high mortality and disability rates. Its common symptoms include a sharp decline in the motor and sensory functions of one limb. How to help patients recover their limb functions is an important research topic in the field of medical rehabilitation. With the development of human-machine intelligence and robotics technology, the use of rehabilitation robots, especially upper and lower limb rehabilitation training robot systems, to help patients with limb rehabilitation training has gradually become an important means of rehabilitation treatment. However, the motor dysfunction of stroke patients is caused by damage to the patient's brain neural circuits. A separate upper and lower limb rehabilitation training system can only provide patients with passive rehabilitation training of the upper and lower limbs. Due to the lack of brain nerve participation (i.e., active rehabilitation training of patients), the rehabilitation effect of the rehabilitation robot system is limited.

[0003] A brain-computer interface is a direct path created between the human or animal brain and an external device that enables information exchange between the human brain and the device. The motor imagery brain-computer interface can collect EEG signals during the patient's motor imagery process to obtain the patient's motor intention, thereby controlling the rehabilitation robot device to assist the patient in completing the corresponding rehabilitation training movements. It has great application potential in the field of rehabilitation. Studies in recent years have shown that the motor imagery brain-computer interface technology combined with rehabilitation robot equipment for rehabilitation training can significantly promote the neural remodeling process of the patient's damaged parts, and its rehabilitation effect is far superior to traditional single motor rehabilitation methods.

[0004] The Chinese patent with publication number CN116059588A discloses "An Active and Passive Upper and Lower Limb Rehabilitation Training Device". The device mainly optimizes the structure of the patient's limb proportion defects to facilitate special patient groups such as polio patients to use the rehabilitation training device normally. It does not take into account the limitations of the upper and lower limb rehabilitation training system for functional recovery of patients with brain defects.

[0005] The Chinese patent application with publication number CN116386807A discloses the "Rehabilitation Training Method and System Based on Motor Imagery and Event-Related Potentials", which uses the transfer learning algorithm to reduce the time and frequency of the patient's initial model training, while reducing the impact of patient specificity on the system's classification and recognition accuracy, helping them quickly enter an effective rehabilitation training state. The mirror training simulated in the VR virtual reality environment optimizes the patient's immersive experience, while the detection of P300 potentials expands the patient's autonomous instruction set, allowing them to independently choose rehabilitation training modes and improve the patient's enthusiasm for training.

[0006] A Chinese patent application with publication number CN115944851A discloses a VR-based immersive electrical stimulation motor imagery rehabilitation training method. This method changes the rehabilitation mode and stimulation level according to the setting of the rehabilitation mode to achieve the goal of staged rehabilitation. First, it is determined whether the patient's condition belongs to the level of rehabilitation. At the current stage, motor imagination is provided and mirror movements are performed in the game scene to help the patient perform bionic learning and collaborative learning of the corresponding postures through the game screen, and a classification model is established. Then, in the online stage, the patient's imagination feedback is determined to be correct based on the classification model, and corresponding stimulation is performed according to different training requirements of different levels. This method can repair damaged brain nerves faster and enable the patient to consolidate muscle memory to achieve the goals of muscle strength training at each stage, thereby achieving the purpose of rehabilitation.

[0007] Both of the above methods guide patients to perform motor imagery by designing different games and provide feedback on the results. However, their main rehabilitation training targets are the patients' upper limbs, and lower limb rehabilitation training is not considered. In addition, the system cannot be adjusted according to the patient's training situation, which affects the patient's overall rehabilitation effect. Summary of the invention

[0008] The purpose of the present invention is to provide an active and passive rehabilitation training method and system for upper and lower limbs based on motor imagery brain-computer interface, which extracts the movement intention of stroke patients through the motor imagery brain-computer interface, thereby controlling the upper and lower limb rehabilitation training system to assist patients in completing the synchronous training of upper and lower limbs, and gradually changes the impedance level and assistance level of the system during the training process, thereby improving the patient's subjective initiative, strengthening the patient's central nervous stimulation feedback, and enhancing the patient's muscle strength, thereby greatly improving the patient's overall rehabilitation effect.

[0009] To achieve the above object, the present invention adopts the following technical solution:

[0010] A method for active and passive rehabilitation training of upper and lower limbs based on motor imagery brain-computer interface, comprising the following steps:

[0011] Step 1: Create two game experimental paradigms: active mode guided patient training and passive mode guided patient training;

[0012] Step 2: Collect the EEG signals and EMG signals generated by the patients when performing the two training game paradigms, and assign labels to each EEG signal;

[0013] Step 3: preprocessing the collected EEG signals and EMG signals;

[0014] Step 4: Use the pre-trained motor imagery classification model to classify and identify the pre-processed EEG signal to obtain a motor imagery classification result for judging the patient's intention;

[0015] Step 5, generating a control instruction according to the obtained motor imagery classification result, and executing the motor assistance function under the control of the passive mode guided patient training game experimental paradigm or the motor assistance function under the control of the active mode guided patient training game experimental paradigm according to the control instruction;

[0016] The movement assistance under the control of the passive mode guided patient training game experimental paradigm is: using the passive mode guided patient training game experimental paradigm to drive the patient to perform synchronous upper and lower limb movements to complete the movement rehabilitation training;

[0017] The motion assistance under the active mode control logic is: based on the patient's motor imagination, the patient is enabled to autonomously perform upper and lower limb rehabilitation training synchronized with the active mode guided patient training experimental paradigm, and the electromyographic signals of the patient's movements are recorded;

[0018] Step 6: After the patient completes the staged training, the patient's muscle strength and motor imagery are evaluated by combining the preprocessed electromyographic signals with expert experience, and the game experiment paradigm parameters and motor assistance function parameters in the corresponding mode are adjusted according to the evaluation results.

[0019] Furthermore, in step 1, the passive mode guided patient training game experimental paradigm adopts a racing game, and the active mode guided patient training game experimental paradigm adopts a cycling game. The game interfaces of the two game experimental paradigms include a patient simulation role and a guiding screen. After setting the initial speed of the racing car or bicycle, the patient is guided to perform corresponding movement imagination to generate EEG signals by setting different changing game scenes and related props.

[0020] Furthermore, the racing game design of the passive mode guided patient training game experimental paradigm is as follows:

[0021] The racing game scene is constructed with a car representing the patient, multiple NPC cars with different speeds, the shape of the left foot brake pedal and the shape of the accelerator pedal. The car representing the patient and the multiple NPC cars travel in the same direction. The car representing the patient races and avoids obstacles to guide the patient to perform left / right foot movement imagination to slow down / accelerate. After the patient's movement intention is determined according to the movement imagination classification results, the corresponding pedal shape dynamic is pressed, and the vehicle speed changes at the same time to strengthen the feedback on the patient's movement imagination;

[0022] The design of the cycling game in the active mode guided patient training game experimental paradigm is as follows:

[0023] The cycling game scene includes a character representing the patient and a bicycle, left / right foot pedaling animations, and three driving sections, namely a flat section, an uphill section, and a downhill section. Each driving section is configured as a different impedance mode (impedance and assistance) for assisted movement according to the patient's muscle strength. The impedance and assistance levels of the flat section are set to 0, the impedance level of the uphill section is increased, and the assistance level of the downhill section is increased. The patient imagines triggering the character to ride forward / stop through left / right foot movement, and the probability of appearance of various sections in the game is changed by detecting the patient's upper and lower limb muscle strength.

[0024] Furthermore, the preprocessing of the collected EEG signals in step 3 includes: channel selection, re-referencing, filtering, downsampling, and epochs segmentation; the preprocessing of the collected electromyographic signals includes: filtering, downsampling, and epochs segmentation to obtain EEG signal sample data.

[0025] Furthermore, the pre-trained motor imagery classification model training method used in step 3 comprises the following steps:

[0026] (1) The EEG signals of the patients performing the two game experimental paradigms are collected in real time and labeled as sample data; the sample data are preprocessed using the same method as the EEG signal preprocessing in step 3 to obtain EEG signal sample data;

[0027] (2) Using the common spatial pattern algorithm to extract features from the EEG signal sample data, the motor imagery category features and label information of each segment of the EEG signal sample data are obtained;

[0028] (3) calculating the normalized covariance matrix of each type of motor imagery sample data according to the motor imagery category characteristics of each segment of EEG signal sample data, and then calculating the mixed space covariance matrix;

[0029] (4) using orthogonal whitening transformation to obtain the mixed space covariance matrix to obtain the whitened feature matrix, then using the whitening matrix to transform the covariance matrix of each type of motor imagery, and then performing principal vector decomposition to obtain the feature vector matrix of each type of motor imagery sample;

[0030] (5) multiplying the whitened feature matrix by the feature vector matrix obtained in (4) to obtain a spatial filter for constructing all motion imagery samples;

[0031] (6) multiplying each motor imagery sample by the corresponding spatial filter and performing variance normalization to obtain a feature vector of each EEG signal sample data; generating a Gaussian kernel-based support vector machine classifier for each patient based on the feature vector and the corresponding label;

[0032] (7) All categories of motor imagery sample data are completely separated by minimizing the classification error and maximizing the classification interval, and the model parameters of the support vector machine classifier are trained; thereby, a motor imagery classification model for judging the patient's motor intention is obtained.

[0033] An upper and lower limb active and passive rehabilitation training system based on motor imagery brain-computer interface, comprising:

[0034] an experimental paradigm module for presenting patients with either an active-mode guided or passive-mode guided training game paradigm;

[0035] A data acquisition module is used to collect the EEG signals and EMG signals generated by the patient in the training game paradigm in real time, and after assigning labels to the EEG signals, synchronously transmit them to the data preprocessing module;

[0036] The data preprocessing module preprocesses the received EEG signals and EMG signals respectively; transmits the preprocessed EEG signals to the online testing module, and transmits the preprocessed EMG signals to the evaluation module;

[0037] The online test module uses the pre-trained motor imagery classification model to classify the pre-processed EEG signals and obtains the motor imagery classification results used to judge the patient's intention; the patient's motor imagery classification results are broadcast through the UDP protocol and conveyed to the motor control module and the experimental paradigm module in the same network frequency band;

[0038] The motion control module is a rehabilitation training system for upper and lower limbs that operates independently, including active mode control logic and passive mode control logic; it generates control instructions based on the received motor imagery classification results, and executes the motor assistance function under the control of the passive mode guided patient training game experimental paradigm or the motor assistance function under the control of the active mode guided patient training game experimental paradigm according to the control instructions. The motor assistance function under the passive mode guided patient training game experimental paradigm is: using the passive mode guided patient training game experimental paradigm to drive the patient to perform synchronous movements of the upper and lower limbs to complete motor rehabilitation training; the motor assistance function under the active mode control logic is: based on the patient's motor imagery, the patient is allowed to autonomously perform upper and lower limb rehabilitation training synchronized with the active mode guided patient training experimental paradigm, and record the patient's electromyographic signals of the movement, which are transmitted to the evaluation module after passing through the acquisition module and the preprocessing module in turn;

[0039] The evaluation module is used to evaluate the patient's muscle strength and motor imagery after the patient completes the staged training, combining the received electromyographic signals and expert experience, and adjust the parameter settings of the experimental paradigm and the parameter settings of the motor assistance function in the corresponding mode according to the evaluation results.

[0040] Furthermore, the active and passive rehabilitation training system for upper and lower limbs based on motor imagery brain-computer interface is also provided with a model training module, which uses a common space mode to extract features from the received EEG signal sample data to obtain motor imagery category features and label information of the EEG signal sample data at each end; and trains a motor imagery classification model based on the motor imagery category features and label information.

[0041] The active and passive rehabilitation training method and system of upper and lower limbs based on motor imagery brain-computer interface provided by the present invention is based on the combination of brain-computer interface rehabilitation method and upper and lower limb active and passive rehabilitation training method, creating two game experimental paradigms that match active mode to guide patient training and passive mode to guide patient training; collect EEG signals and electromyographic signals through brain-computer interface, extract the movement intention of stroke patients by using EEG signals, and guide patients to actively or passively conduct rehabilitation training in conjunction with two game experimental paradigms, thereby accelerating the rehabilitation training process of patients. Motor imagery is evaluated by using electromyographic signals combined with expert experience, and the parameter settings of the experimental paradigm and the parameter settings of the motor assistance function in the corresponding mode are adjusted according to the evaluation results. The parameter settings include gradually changing the impedance level, power level and other parameters of the system during the training process, thereby improving the subjective initiative of patients, strengthening the patient's central nervous stimulation feedback, and enhancing the patient's muscle strength, thereby greatly improving the overall rehabilitation effect of patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a block diagram of an upper and lower limb active and passive rehabilitation training system based on motor imagery brain-computer interface provided in an embodiment;

[0043] Figure 2 is a schematic diagram of an upper and lower limb active and passive rehabilitation training system based on motor imagery brain-computer interface provided in an embodiment;

[0044] Figure 3 It is an active and passive rehabilitation training system for upper and lower limbs based on motor imagery brain-computer interface provided by an embodiment, a training game paradigm and schematic diagram guided in passive mode;

[0045] Figure 4 It is an active and passive rehabilitation training system for upper and lower limbs based on motor imagery brain-computer interface provided by an embodiment, a training game paradigm and schematic diagram guided in active mode;

[0046] Figure 5 It is a schematic diagram of the EEG acquisition channel and the EMG acquisition channel of the upper and lower limb active and passive rehabilitation training system based on the motor imagery brain-computer interface provided in the embodiment, wherein a is a schematic diagram of EEG signal acquisition and b is a schematic diagram of EMG signal acquisition. DETAILED DESCRIPTION

[0047] The technical solution of the present invention is described in detail below in conjunction with the accompanying drawings.

[0048] The present embodiment provides a method for active and passive rehabilitation training of upper and lower limbs based on motor imagery brain-computer interface, comprising the following steps:

[0049] Step 1. Create two game experimental paradigms: active mode guided patient training and passive mode guided patient training. The passive mode guided patient training game experimental paradigm uses a racing game, and the active mode guided patient training game experimental paradigm uses a cycling game. Both game experimental paradigms were made by E-PRIME and Unity software. The user subjects were free racing and bicycle model resources from the Asset Store imported into Unity. The visual interface was set with patient simulation characters and guidance screens. After setting the initial speed of the racing car or bicycle, the patient was guided to perform corresponding motor imagination to generate EEG signals by setting different game scenes and related props.

[0050] The passive mode guided patient training game experimental paradigm is set up in Figure 3 . The car marked with an arrow in the center of the visual interface represents the patient, and the arrow indicates the patient's forward direction. Multiple NPC cars marked with different speeds are driving on both sides of the car representing the patient and moving forward with the car representing the patient. The left foot brake pedal shape is set at the lower left of the visual interface, and the accelerator pedal shape is set at the lower right. The vehicle racing obstacle avoidance on the visual interface guides the patient to perform left / right foot movement imagination to slow down / accelerate. After determining the patient's movement intention according to the movement imagination classification result, press the corresponding pedal shape dynamic, and the vehicle speed changes at the same time to strengthen the feedback of the patient's movement imagination. In this embodiment, the initial speed of the vehicle representing the patient is set to 15 revolutions per minute (rpm), and the speed of other NPC cars is set to an integer between 20 and 35 rpm. Under the passive model, the movement imagination paradigm lasts a total of 12 seconds. 0 to 2 seconds are used to prompt the patient to perform left / right foot movement imagination through visual images and voice, 2 to 6 seconds are used to perform movement imagination according to the prompts, 6 to 12 seconds are used to feedback the movement imagination classification results through visual images and voice, and the movement assists the joint speed adjustment to complete a single trial experiment. The acceleration / deceleration step size was 5 rpm each time. The E-PRIME experimental paradigm consisted of 3 rounds, each round contained 5 blocks, each block contained 10 experimental trials, and the duration of a single experimental trial was 12 s.

[0051] The active mode guided patient training game experimental paradigm is set up in Figure 4 .

[0052] The character in the center of the visual interface represents the patient, and the arrow indicates the direction in which the patient is moving. Different sections are set in the game, such as flat sections, uphill sections, and downhill sections. The left / right foot pedaling animation is displayed on both sides below the visual interface; the patient triggers the character to ride forward / stop by imagining the left / right foot movement; the probability of the sections in the game is changed by detecting the patient's upper and lower limb muscle strength. Different section designs represent different impedance modes (power assist level and impedance level) of the sports assist module. When on flat ground, the power assist level and impedance level are both 0. The impedance level increases on the uphill section, and the power assist level increases on the downhill section. The impedance and power assist level depend on the patient's muscle strength assessment. When the impedance level increases on the uphill section, the screen prompts the patient to increase the strength of the limbs to complete the rehabilitation training movements within the specified time. When the power assist level increases on the downhill section, the screen prompts the patient to relax the limbs and relieve limb fatigue. The total duration of the motor imagery paradigm in the active mode is 12 seconds. The patient is prompted by visual images and voice to imagine the left / right foot movement for 0-2 seconds, the patient performs motor imagery according to the prompts for 2-6 seconds, and the motor imagery classification results are fed back through visual images and voice for 6-12 seconds. At the same time, the patient is reminded to try his best to complete the prescribed rehabilitation training movements independently. At the same time, the data acquisition module records the patient's upper and lower limb electromyographic signals during this period to achieve a single trial experiment. The E-PRIME experimental paradigm consists of four rounds, each round contains 5 blocks, each block contains 10 experimental trials, and a single experimental trial lasts 12 seconds.

[0053] Step 2: Collect the EEG signals and EMG signals generated by the patient in the two training game paradigms in real time, and assign a label to each EEG signal.

[0054] The device used in this embodiment to collect EEG signals is the Eego manufactured by ANT neuro, a German company. TM mylab 32-lead device. The device used to collect myoelectric signals is the high-performance myoelectric signal collector myoMUSCLE produced by Noraxon, USA. The EEG acquisition equipment includes a 32-channel EEG cap, a 32-channel 16kHz amplifier, and a terminal notebook equipped with eego64 EEG signal acquisition software. EEG signal acquisition and processing software eego64 is used to present and process the received EEG signals in real time. It also has the event marking function, that is, assigning labels. The time position of the event mark is shown in Figure 2 and Figure 3 The EEG cap is connected to the amplifier by wire, and is used to transmit the collected EEG data to the terminal notebook through the amplifier.

[0055] The electromyographic acquisition device of this embodiment consists of three parts: electromyographic acquisition processing software, a data receiver, and an electromyographic data sensor. The electromyographic data sensor has a sampling frequency of up to 4000Hz and is used to collect and transmit electromyographic signals in real time. The electromyographic data sensor and the data receiver transmit data wirelessly. The electromyographic acquisition processing software is used to present and process the electromyographic signals received by the data receiver in real time.

[0056] Step 3: pre-process the collected EEG signals and EMG signals.

[0057] The preprocessing of EEG signals includes: channel selection, re-referencing, filtering, downsampling and epochs segmentation processing. Among them:

[0058] Channel selection processing: The EEG signal channels selected in this embodiment are the following electrode channels in the international 10-20 standard lead system: Fp1, Fp2, Fpz, C3, C4, Cz, T7, T8, FC1, FC2, Cp1, Cp2, Cp5, Cp6. Figure 5 (a) is shown. The re-reference processing is used to avoid the interference of multi-channel EEG signal activity on single channel signals caused by volume conduction effect, as well as the loss of single channel EEG signal information. In this embodiment, the M1 and M2 bilateral mastoid electrode channels are selected as reference channels, and the average value of the bilateral mastoid channel signals is subtracted from each segment of the EEG electrode channel signal to achieve EEG signal re-reference. The filtering processing uses a Butterworth bandpass filter to filter the EEG signal to retain the EEG signal in a specific (8-28Hz) frequency band. In this embodiment, the upper limit frequency is 40Hz and the lower limit frequency is 5Hz. The downsampling processing is used to reduce the length of the collected EEG signal to remove redundant data. In this embodiment, the sampling rate is reduced from 1000Hz to 250Hz. The epochs segmentation processing is used to divide the collected EEG signal into EEG signal data of different experimental times according to the label. The data of 2-6s in the experimental paradigm in the active / passive mode is taken as the EEG signal data of a single trail.

[0059] The preprocessing of electromyographic signals includes filtering, downsampling, and epochs segmentation processing. Among them:

[0060] Filtering is used to retain the EMG signal data in a specific (8-28Hz) frequency band. The filtering operation uses a Butterworth bandpass filter with an upper frequency limit of 450Hz and a lower frequency limit of 10Hz; a 50Hz filtering operation is performed at the same time to remove AC interference. The downsampling operation is used to reduce the length of the collected EMG signal by reducing the sampling rate from 2000Hz to 250Hz. Epochs segmentation is used to divide the collected EMG signal into EMG signal data of different experimental times according to the marking points. In this embodiment, the data of 6-12s in the experimental paradigm under the active / passive mode is taken as the EMG signal data of a single trail.

[0061] Step 4: Use the pre-trained motor imagery classification model to classify and identify the pre-processed EEG signal to obtain a motor imagery classification result for judging the patient's intention. The pre-trained motor imagery classification model training method used in step 3 of this embodiment includes the following steps:

[0062] The EEG signals of the patients performing the two game experimental paradigms were collected in real time and labeled as sample data; the sample data was preprocessed using the same method as the EEG signal preprocessing in step 3, and after preprocessing, the EEG signal sample data of a single experimental trial was divided with a time window of 1000ms and a window overlap rate of 50%. Then, the common spatial pattern algorithm was used to extract features from the divided EEG signal segments and train the spatial filter.

[0063] Assume that X1 and X2 are the EEG data sets under the two types of motor imagination of the left and right feet, that is, and C is the number of EEG channels, S is the number of samples collected for each channel, and m is the number of samples for each type of motor imagery. is the set of real numbers.

[0064] For each type of motor imagery sample, the normalized covariance matrix is ​​calculated, and then the mixed space covariance matrix is ​​obtained, as shown in formulas (1) and (2):

[0065]

[0066] R=R1+R2 (2)

[0067] Since the mixed space covariance matrix R is a positive definite matrix, the eigendecomposition is performed according to the singular value decomposition theorem:

[0068] R=UΛU T (3)

[0069] U is the eigenvector matrix, Λ is the diagonal matrix of the corresponding eigenvalues, arranged in descending order by eigenvalue, and the whitened value matrix is:

[0070]

[0071] The covariance matrix of each type of motion imagery is transformed using the whitening matrix, and then the principal vector decomposition is performed to obtain the eigenvector matrix of each type of sample, as shown in formulas (5) and (6).

[0072] S1=PR1P T ,S2=PR2P T (5)

[0073]

[0074] From equations (5) and (6), we can see that the eigenvector matrix of matrix S1 is equal to the eigenvector matrix of matrix S2, that is, B1 = B2 = B. Then the eigenvector corresponding to the maximum eigenvalue of matrix S1 makes matrix S2 have the minimum eigenvalue, and vice versa. Therefore, the projection matrix W is:

[0075] W=BP (7)

[0076] The projection matrix W is the spatial filter corresponding to the CSP algorithm. For each motor imagery sample, the feature vector of the sample can be obtained through the projection matrix W. The feature vector obtained from a single trial is used as a sample (a 1×T vector). The feature vectors extracted from all experimental CSPs and the corresponding motor imagery labels are used to generate a Gaussian kernel-based support vector machine classifier for each patient. The two types of samples are completely separated by minimizing the classification error and maximizing the classification interval. The model parameters W of the support vector machine classifier are obtained by training. and B

[0077] During implementation, the feature F extracted by the common spatial pattern CSP is first calculated for the received preprocessed EEG signal X, and the patient's left / right foot motor imagination is judged by comparing the size of FW+B with 0. If FW+B>0, the output 1 represents the left foot motor imagination, otherwise the output -1 represents the right foot motor imagination.

[0078] Step 5: Generate control instructions based on the obtained motor imagery classification results, and execute the motor assistance function under the control of the passive mode guided patient training game experimental paradigm or the motor assistance function under the control of the active mode guided patient training game experimental paradigm according to the control instructions.

[0079] In passive mode, the initial speed setting of the motion control module is 15rpm. The motor imagery paradigm time is 12s. The system extracts CSP features and uses the support vector machine classifier to judge the instructions, and displays the acceleration / deceleration animation of the car in the interface. At the same time, the speed of the motion control module increases / decreases synchronously.

[0080] In the active mode, the patient first performs motor imagery, and then performs upper and lower limb rehabilitation training exercises autonomously with the greatest strength. The motor imagery paradigm time is 12s. During motor imagery, the system rotation joints are limited and cannot rotate. The system extracts CSP features and the support vector machine classifier determines the instructions. If the motor imagery is successful, the system joint movement limit is cancelled, and the patient performs autonomous rehabilitation training. At the same time, the bicycle in the interface starts to move forward at the same joint rotation speed and continues until the next motor imagery. During the patient's autonomous rehabilitation training, the electromyographic signals of the rectus femoris, biceps femoris, tibialis anterior, gastrocnemius, biceps brachii and triceps brachii are recorded, such as Figure 5 After completing the phased training, the patient's muscle strength is evaluated based on the patient's electromyographic signal and expert experience. Based on the muscle strength evaluation results, the probability of uphill and downhill scene changes and the impedance and assistance level corresponding to the uphill and downhill scenes are introduced.

[0081] Step 6: After the patient completes the staged training, the patient's muscle strength and motor imagery are evaluated by combining electromyographic signals with expert experience, and the game experiment paradigm parameters and motor assistance function parameters in the corresponding mode are adjusted according to the evaluation results.

[0082] In this embodiment, 30 rounds are defined as one stage. For the experimental paradigm module, the patient's motor imagination success rate is defined as the ratio of the number of successful motor imaginations according to the guidance picture within the specified motor imagination time to the total number of experiments. In passive and active modes, if the patient's staged motor imagination accuracy is greater than or equal to 90%, the motor imagination time in the experimental paradigm in the corresponding mode is shortened by 0.5s until the patient's motor imagination time is finally 1s. For the motion assistance module, its parameter adjustment involves the impedance and assistance level of the system's rotating joints.

[0083] In active mode, the patient's muscle strength level is defined as 0 to 5, which is consistent with the definition of medical muscle strength level. According to the amplitude, time and maximum muscle strength of the electromyographic signal of the patient's staged completion of autonomous rehabilitation training, combined with expert experience, the overall level of the patient's limb muscle strength is evaluated. The system joint impedance and power level of the motion assist module are defined as 0 to 5 levels, and the impedance level is defined as 0%, 10%, 20%, 30%, 40%, 50% of the patient's current maximum muscle strength, and the power level corresponds to 0%, 20%, 40%, 60%, 80%, 100% of the patient's current maximum muscle strength. If the muscle strength level is defined as level a, the system's uphill scene rehabilitation training joint impedance is defined as level a, and the downhill scene rehabilitation training joint power level is defined as level 5-a. The probability of flat ground, uphill and downhill is 50%, (10a)%, (50-10a)%.

[0084] Based on the above method, this embodiment also provides an upper and lower limb active and passive rehabilitation training system based on motor imagery brain-computer interface, such as Figure 1 and Figure 2 As shown, including:

[0085] Experimental paradigm module for presenting patients with either active-mode guided or passive-mode guided training game paradigms.

[0086] The data acquisition module is used to collect the EEG signals and EMG signals generated by the patient in the training game paradigm in real time, assign labels to the EEG signals, and then synchronously transmit them to the data preprocessing module.

[0087] The data preprocessing module is used to preprocess the EEG signal and the EMG signal, transmit the preprocessed EEG signal to the online testing module, and transmit the preprocessed EMG signal to the evaluation module.

[0088] The online test module uses a pre-trained motor imagery classification model to classify the preprocessed EEG signals to obtain motor imagery classification results for judging the patient's intentions; the patient's motor imagery classification results are broadcast through the UDP protocol and conveyed to the motor control module and the experimental paradigm module in the same network frequency band.

[0089] The motion control module is a rehabilitation training system for upper and lower limbs that operates independently, including active mode control logic and passive mode control logic; it generates control instructions according to the received motor imagery classification results, and executes the motor assistance function under the control of the passive mode guided patient training game experimental paradigm or the motor assistance function under the control of the active mode guided patient training game experimental paradigm according to the control instructions. The motor assistance function under the passive mode guided patient training game experimental paradigm is: using the passive mode guided patient training game experimental paradigm to drive the patient to perform synchronous upper and lower limb movements to complete motor rehabilitation training; the motor assistance function under the active mode control logic is: based on the patient's motor imagery, the patient is allowed to autonomously perform upper and lower limb rehabilitation training synchronized with the active mode guided patient training experimental paradigm, and record the patient's electromyographic signals of the movement, which are transmitted to the evaluation module after passing through the acquisition module and the preprocessing module.

[0090] The evaluation module is used to evaluate the patient's muscle strength and motor imagery after the patient completes the staged training, combining the received electromyographic signals and expert experience, and adjust the parameter settings of the experimental paradigm and the parameter settings of the motor assistance function in the corresponding mode according to the evaluation results.

[0091] Since the EEG signals of each subject vary greatly from person to person, in order to obtain the best matching model with the user, this embodiment sets up a model training model for the above-mentioned upper and lower limb active and passive rehabilitation training system. The model training module receives the preprocessed EEG signals, and then uses a common space mode to extract features from the received EEG signal sample data to obtain the motor imagination category features and label information of each EEG signal sample data; and trains the motor imagination classification model based on the motor imagination category features and label information.

[0092] It is to be understood that the present invention is described by some embodiments, and it is known to those skilled in the art that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present invention. In addition, under the teachings of the present invention, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the scope of protection of the present invention.

Claims

1. A method for active and passive rehabilitation training of upper and lower limbs based on motor imagery brain-computer interface, comprising the following steps: Step 1: Create two game experimental paradigms: active mode guided patient training and passive mode guided patient training; Step 2: Collect the EEG signals and EMG signals generated by the patients when performing the two training game paradigms, and assign labels to each EEG signal; Step 3: preprocessing the collected EEG signals and EMG signals; Step 4: Use the pre-trained motor imagery classification model to classify and identify the pre-processed EEG signal to obtain a motor imagery classification result for judging the patient's intention; Step 5, generating a control instruction according to the obtained motor imagery classification result, and executing the motor assistance function under the control of the passive mode guided patient training game experimental paradigm or the motor assistance function under the control of the active mode guided patient training game experimental paradigm according to the control instruction; The movement assistance under the control of the passive mode guided patient training game experimental paradigm is: using the passive mode guided patient training game experimental paradigm to drive the patient to perform synchronous upper and lower limb movements to complete the movement rehabilitation training; The motion assistance under the active mode control logic is: based on the patient's motor imagination, the patient is enabled to autonomously perform upper and lower limb rehabilitation training synchronized with the active mode guided patient training experimental paradigm, and the electromyographic signals of the patient's movements are recorded; Step 6: After the patient completes the staged training, the patient's muscle strength and motor imagery are evaluated by combining the preprocessed electromyographic signals with expert experience, and the game experiment paradigm parameters and motor assistance function parameters in the corresponding mode are adjusted according to the evaluation results.

2. The method for active and passive rehabilitation training of upper and lower limbs based on motor imagery brain-computer interface according to claim 1, characterized in that: In the step 1, the passive mode guided patient training game experimental paradigm adopts a racing game, and the active mode guided patient training game experimental paradigm adopts a cycling game, and the game interfaces of the two game experimental paradigms both include a patient simulation role and a guiding screen; After setting the initial speed of the racing car or bicycle, the patient is guided to perform corresponding movement imagination to generate EEG signals by setting different game scenes and related props.

3. The method for active and passive rehabilitation training of upper and lower limbs based on motor imagery brain-computer interface according to claim 2, characterized in that: The design of the racing game in the passive mode guided patient training game experimental paradigm is as follows: The racing game scene is constructed with a car representing the patient, multiple NPC cars with different speeds, the shape of the left foot brake pedal and the shape of the accelerator pedal. The car representing the patient and the multiple NPC cars travel in the same direction. The car representing the patient races and avoids obstacles to guide the patient to perform left / right foot movement imagination to slow down / accelerate. After the patient's movement intention is determined according to the movement imagination classification results, the corresponding pedal shape dynamic is pressed, and the vehicle speed changes at the same time to strengthen the feedback on the patient's movement imagination; The design of the cycling game in the active mode guided patient training game experimental paradigm is as follows: The cycling game scene includes a character representing the patient and a bicycle, left / right foot pedaling animations, and three driving sections, namely a flat section, an uphill section, and a downhill section. Each driving section is configured as a different impedance mode (impedance and assistance) for assisted movement according to the patient's muscle strength. The impedance and assistance levels of the flat section are set to 0, the impedance level of the uphill section is increased, and the assistance level of the downhill section is increased. The patient imagines triggering the character to ride forward / stop through left / right foot movement, and the probability of appearance of various sections in the game is changed by detecting the patient's upper and lower limb muscle strength.

4. The method for active and passive rehabilitation training of upper and lower limbs based on motor imagery brain-computer interface according to claim 1, characterized in that: In the step 3, the preprocessing of the collected EEG signal includes: channel selection, re-referencing, filtering, downsampling, and epochs segmentation; the preprocessing of the collected electromyographic signal includes: filtering, downsampling, and epochs segmentation to obtain EEG signal sample data.

5. The method for active and passive rehabilitation training of upper and lower limbs based on motor imagery brain-computer interface according to claim 1, characterized in that: The pre-trained motor imagery classification model training method used in step 3 comprises the following steps: (1) The EEG signals of the patients performing the two game experimental paradigms are collected in real time and labeled as sample data; the sample data are preprocessed using the same method as the EEG signal preprocessing in step 3 to obtain EEG signal sample data; (2) Using the common spatial pattern algorithm to extract features from the EEG signal sample data, the motor imagery category features and label information of each segment of the EEG signal sample data are obtained; (3) calculating the normalized covariance matrix of each type of motor imagery sample data according to the motor imagery category characteristics of each segment of EEG signal sample data, and then calculating the mixed space covariance matrix; (4) using orthogonal whitening transformation to obtain the mixed space covariance matrix to obtain the whitened feature matrix, then using the whitening matrix to transform the covariance matrix of each type of motor imagery, and then performing principal vector decomposition to obtain the feature vector matrix of each type of motor imagery sample; (5) multiplying the whitened feature matrix by the feature vector matrix obtained in (4) to obtain a spatial filter for constructing all motion imagery samples; (6) multiplying each motor imagery sample by the corresponding spatial filter and performing variance normalization to obtain a feature vector of each EEG signal sample data; generating a Gaussian kernel-based support vector machine classifier for each patient based on the feature vector and the corresponding label; (7) All categories of motor imagery sample data are completely separated by minimizing the classification error and maximizing the classification interval, and the model parameters of the support vector machine classifier are trained; thereby, a motor imagery classification model for judging the patient's motor intention is obtained.

6. An active and passive rehabilitation training system for upper and lower limbs based on motor imagery brain-computer interface, comprising: an experimental paradigm module for presenting patients with either an active-mode guided or passive-mode guided training game paradigm; A data acquisition module is used to collect the EEG signals and EMG signals generated by the patient in the training game paradigm in real time, and after assigning labels to the EEG signals, synchronously transmit them to the data preprocessing module; A data preprocessing module preprocesses the received EEG signals and EMG signals respectively; The preprocessed EEG signals are transmitted to the online testing module, and the preprocessed EMG signals are transmitted to the evaluation module; The online testing module uses the pre-trained motor imagery classification model to classify the pre-processed EEG signals and obtains the motor imagery classification results for judging the patient's intention; The patient's motor imagery classification results are broadcasted through the UDP protocol and conveyed to the motor control module and the experimental paradigm module in the same network frequency band; The motion control module is a rehabilitation training system for upper and lower limbs that operates independently, including active mode control logic and passive mode control logic; it generates control instructions based on the received motor imagery classification results, and executes the motor assistance function under the control of the passive mode guided patient training game experimental paradigm or the motor assistance function under the control of the active mode guided patient training game experimental paradigm according to the control instructions. The motor assistance function under the passive mode guided patient training game experimental paradigm is: using the passive mode guided patient training game experimental paradigm to drive the patient to perform synchronous movements of the upper and lower limbs to complete motor rehabilitation training; the motor assistance function under the active mode control logic is: based on the patient's motor imagery, the patient is allowed to autonomously perform upper and lower limb rehabilitation training synchronized with the active mode guided patient training experimental paradigm, and record the patient's electromyographic signals of the movement, which are transmitted to the evaluation module after passing through the acquisition module and the preprocessing module in turn; The evaluation module is used to evaluate the patient's muscle strength and motor imagery after the patient completes the staged training, combining the received electromyographic signals and expert experience, and adjust the parameter settings of the experimental paradigm and the parameter settings of the motor assistance function in the corresponding mode according to the evaluation results.

7. The active and passive rehabilitation training system for upper and lower limbs based on motor imagery brain-computer interface according to claim 6, characterized in that: The active and passive rehabilitation training system for upper and lower limbs based on motor imagery brain-computer interface is also provided with a model training module, which uses a common space mode to extract features from the received EEG signal sample data to obtain motor imagery category features and label information of the EEG signal sample data at each end; and trains a motor imagery classification model based on the motor imagery category features and label information.

Citation Information

Patent Citations

  • Training and mode switching method of pure idea control rehabilitation robot

    CN111544846A

  • Brain-computer interface multi-mode rehabilitation training system

    CN114237387A

  • Rehabilitation bicycle training method and system based on motor imagery

    CN117493801A

  • Brain-controlled body movement assistance devices and methods

    US20140277582A1

  • Game-based rehabilitation system having brain-computer interface (BCI) technology combined with functional electrical stimulation for maximizing activation of mirror nerve cell, and control method therefor

    WO2019212149A1

Cited By

  • Mechanical exoskeleton rehabilitation training system and method based on brain-computer interface

    CN120514396A

  • A brain-computer interface-based mechanical exoskeleton rehabilitation training system and method

    CN120514396B

  • Rehabilitation system for adjusting training normal form based on intracranial electroencephalogram signals

    CN121641343A

  • Method for non-invasive determination of parameters of rhythmic activity of autonomic nervous system in perirenal dorsal lumbar regions of person

    RU2855823C1

  • Closed-loop brain computer interface-based rehabilitation training method and apparatus, device, and storage medium

    WO2026112803A1