Motor imagery rehabilitation system and method based on brain-computer interface

Abstract

The invention discloses a motor imagery rehabilitation system and a motor imagery rehabilitation method based on a brain-computer interface. The system comprises an electroencephalogram acquisition module, wherein the electroencephalogram acquisition module is configured to acquire motor imagery state electroencephalogram data and resting state electroencephalogram data of a patient; the electroencephalogram analysis module is configured to perform motor imagery classification and recognition based on the motor imagery state electroencephalogram data to obtain a classification and recognition result; obtaining average imagination task energy and average resting state energy; determining a movement area change difference index according to the left-right movement brain area center lead electroencephalogram data energy and the left-right movement brain area center lead electroencephalogram data energy; determining whether the motion brain area of the patient is activated or not according to the motion area change difference index; the limb movement control execution module is configured to control the end executor to execute corresponding rehabilitation training actions according to a preset training mode requirement, the classification recognition result and / or whether the conveying brain area is activated or not. According to the method, the activity state of the exercise brain area can be accurately monitored, scientific quantitative indexes are provided for rehabilitation training, and the rehabilitation effect is improved.

Description

Technical Field

[0001] This application belongs to the technical field of brain-computer interfaces, and particularly relates to a motor imagery rehabilitation system and method based on a brain-computer interface. Background Art

[0002] Injuries such as strokes and car accidents can cause damage to the patient's cranial nerve tissue, and their typical sequela is motor function impairment. For example, in a patient whose upper limb cannot move due to damage to the brain motor function area caused by a stroke, the functions of the peripheral nerve tissue and muscle tissue are normal, but the brain motor area is damaged. Neurofeedback therapy can promote the remodeling of the brain motor area and restore the upper limb motor function.

[0003] Conventional treatment methods are usually passive therapies, which are beneficial to helping patients maintain muscle function activity and do not act repeatedly on the brain motor area. Neurofeedback therapy can effectively help patients repair the brain motor area. The patient imagines the movement of the affected limb, and then observes or feels the corresponding movement, forming neurofeedback. This method can accelerate the neural remodeling of the damaged brain area of the patient and promote rehabilitation. Some rehabilitation systems based on brain-computer interface technology use electroencephalogram (EEG) to identify the patient's movement intention and control mechanical devices such as exoskeletons to move the corresponding limb of the patient, realizing neurofeedback therapy.

[0004] However, the above-mentioned rehabilitation system devices based on brain-computer interface technology can only identify the EEG differences of different movement patterns and cannot clearly identify whether this difference comes from the motor brain area, which may lead to ineffective stimulation of non-motor brain areas, while the motor brain area that really needs to be repaired does not receive enough training, thus affecting the rehabilitation effect and delaying the recovery process of the patient's motor function. Summary of the Invention

[0005] The technical objective of this application is to provide a motor imagery rehabilitation system and method based on a brain-computer interface for the technical problem that the existing rehabilitation system devices based on brain-computer interface technology cannot clearly identify whether this difference comes from the motor brain area, thus affecting the rehabilitation effect.

[0006] To achieve the above technical objective, this application adopts the following technical solutions.

[0007] In a first aspect, an embodiment of this application provides a motor imagery rehabilitation system based on a brain-computer interface, including:

[0008] An electroencephalogram (EEG) acquisition module, configured to acquire the EEG data of the patient in the motor imagery state and the resting state;

[0009] The electroencephalogram analysis module is configured to perform motor imagery classification and recognition based on the motor imagery electroencephalogram data to obtain a classification and recognition result; obtain the average imagery task energy and the average resting state energy based on the motor imagery electroencephalogram data and the resting state electroencephalogram data; determine the motor area change difference index based on the average imagery task energy, the average resting state energy, and the electroencephalogram data energy of the central leads in the left and right motor areas; and determine whether the motor area of the patient is activated according to the motor area change difference index.

[0010] The limb movement control execution module is configured to control the end effector to perform corresponding rehabilitation training actions according to the classification and recognition result and whether the / transportation brain area is activated according to the requirements of a preset training modality.

[0011] In a second aspect, an embodiment of the present application provides a motor imagery rehabilitation method based on a brain-computer interface, including:

[0012] Collect the motor imagery electroencephalogram data and the resting state electroencephalogram data of a patient.

[0013] Perform motor imagery classification and recognition based on the motor imagery electroencephalogram data to obtain a classification and recognition result; obtain the average imagery task energy and the average resting state energy based on the motor imagery electroencephalogram data and the resting state electroencephalogram data; determine the motor area change difference index based on the average imagery task energy, the average resting state energy, and the electroencephalogram data energy of the central leads in the left and right motor areas; and determine whether the motor area of the patient is activated according to the motor area change difference index.

[0014] Control the end effector to perform corresponding rehabilitation training actions according to the classification and recognition result and whether the / transportation brain area is activated according to the requirements of a preset training modality.

[0015] Compared with the prior art, the motor imagery rehabilitation system and method provided by the embodiment of the present application have the following beneficial technical effects: By calculating the average imagery task energy, the average resting state energy, and the electroencephalogram data energy of the central leads in the left and right motor areas to determine the motor area change difference index, and judging whether the motor area of the patient is activated accordingly. This method can accurately monitor the activity state of the motor area, provide a scientific quantitative index for rehabilitation training, help adjust the training plan in time, and improve the rehabilitation effect. The rehabilitation system provided by the present application can customize rehabilitation training actions for each patient according to the specific electroencephalogram data characteristics and the activation situation of the motor area of the patient, realize personalized rehabilitation treatment, and improve the effectiveness and adaptability of training. Specific embodiments

[0016] To enable those skilled in the art to better understand the technical solutions in this application, the following will clearly and completely describe the technical solutions in the embodiments of this application in conjunction with the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this application without creative efforts shall fall within the scope of protection of this application.

[0017] In the description of this application, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more features.

[0018] This embodiment provides a motor imagery rehabilitation system based on a brain-computer interface, including an electroencephalogram (EEG) acquisition module, an EEG analysis module, and a limb movement control execution module.

[0019] Among them, the EEG acquisition module is configured to acquire the motor imagery EEG data and resting-state EEG data of the patient.

[0020] The EEG analysis module is configured to perform motor imagery classification and recognition based on the motor imagery EEG data to obtain a classification and recognition result; obtain the average imagery task energy and average resting-state energy based on the motor imagery EEG data and the resting-state EEG data; determine the movement area change difference index based on the average imagery task energy, the average resting-state energy, and the EEG data energy of the central leads in the left and right motor areas; and determine whether the patient's motor area is activated according to the movement area change difference index.

[0021] The limb movement control execution module is configured to control the end effector to perform corresponding rehabilitation training actions according to the classification and recognition result and whether the / transportation brain area is activated according to the requirements of the preset training modality.

[0022] In the embodiment, the EEG acquisition module acquires EEG data, which can be used to evaluate the patient's motor intention, and further evaluate the motor initiative to encourage the patient to increase initiative.

[0023] In some embodiments, the EEG acquisition module is further configured to complete the preprocessing of EEG data. In the preprocessing stage, artifacts are identified based on the standard deviation of the signal values (such as amplitude) of each lead and the deviation value calculated according to the standard deviation. The influence of abnormal data on the system is reduced.

[0024] As an example, according to the actual EEG data situation, data preprocessing is performed, including:

[0025] (1) Use scalp electroencephalogram (EEG) as EEG data. The EEG acquisition electrodes must comply with the international 10-20 standard and must include C3, C4, Cz and surrounding electrodes. Ensure that the EEG activity in the brain motor area is captured to ensure smooth subsequent processing.

[0026] (2) Apply real-time bandpass filtering (e.g. 5Hz-35Hz) and 50Hz notch filtering to the EEG data, retaining motion-related potentials and removing power frequency and other artifacts.

[0027] (3) At time interval t g Cut length t s If the data segment is seriously interfered by artifacts, it will be discarded.

[0028] As an example, the artifact detection method is as follows:

[0029] ① Calculate the standard deviation of the signal value of each lead in the data segment ( represents the standard deviation of the signal value of lead i). If the standard deviation of the signal value of any lead exceeds 100 μV, the data has obvious artifact interference;

[0030] ②Calculate the standard deviation of all lead signal values The standard deviation of c And the standard deviation of all lead signal values The mean value m c , calculate the deviation value of each lead signal value

[0031]

[0032] Among them, abs is the absolute value calculation, is the signal value deviation value of lead i.

[0033] If the deviation value is greater than the preset value (such as 5), it means that there is serious artifact interference in a certain lead and the data in this segment should be discarded.

[0034] In some embodiments, the EEG analysis module classifies and identifies motor imagery in EEG data. There are two typical identification methods, one is a deep learning method based on a neural network, and the other is a machine learning method such as a linear SVM identification method. The following assumption is that the recognition model has been trained (the specific training method is a prior art and is not described in detail in this application), and the current motor intention can be identified based on the EEG data (for example, identifying whether the patient imagines left or right hand movement). For the above two types of recognition models, the recognition method of motor intention is described:

[0035] ① For deep learning recognition methods: for N c For classification tasks, it is generally stipulated that the network output layer generates Nc output items, respectively representing N c motor imagery classifications. The output items are processed by the Softmax function, making the sum of all output items equal to 1, and the range of each output item is 0 - 1. Find the largest output item, which is the predicted intent classification for the current data, and the percentage of the output item is the current intent score.

[0036] ② Classification and recognition methods of machine learning, such as the linear SVM recognition method: Before using the classifier, the model parameters need to be adjusted using training data, and the training data needs to include all recognition classifications.

[0037] In some embodiments, the EEG analysis module is further configured to determine the intent score of the current sample based on the distances between all samples of each category in the EEG data and the corresponding classification hyperplane, expressed as:

[0038]

[0039] where is the intent score of the j-th sample of category c i , is the distance between the j-th sample of category c i after classification and the corresponding classification hyperplane, μ i is the average value of the distances between all samples in category c i and the corresponding classification hyperplane, σ is the standard deviation of the distances between all samples in category c i and the corresponding classification hyperplane; N i is the number of classifications. ; N c is the number of classifications.

[0040] The above expression quantifies a certain "intent" degree of the sample in classification through information such as the distance between the sample and the classification hyperplane. Dividing by 0.5 is to scale and adjust the integral result. Since the integral value of the normal distribution from negative infinity to positive infinity is 1, and the integral value from negative infinity to the mean is 0.5, here it is to adjust the calculated probability value to a suitable range. Dividing by N c , N c is the number of classifications. This step normalizes the above results among all classifications, making the intent scores calculated under different classifications within a comparable range.

[0041] The intent score of the j-th sample of category c i ranges from 1 / N to 1; c ~1;

[0042] For example, a motor imagery rehabilitation system based on a brain-computer interface identifies whether a patient is attempting to move their right hand. The identification task is a binary classification task of rest state and right hand movement. The intention classification is either rest state or right hand movement, and the intention score represents the intensity of the current intention. The higher the score, the stronger the intention, with a range of 0 - 100. Identifying the intention classification and intention score can assist in subsequent threshold processing.

[0043] In the embodiment, the electroencephalogram (EEG) analysis module is configured to evaluate the activation of the patient's motor functional area of the brain: calculate the activation of each region of the brain, evaluate whether there is an obvious activation phenomenon in the motor area, and determine whether the patient has correctly performed the motor imagery task to ensure that the treatment focuses on motor nerve rehabilitation.

[0044] Specifically, before a person performs an action, the motor area of the brain (near positions such as C3 and C4) generates a series of EEG responses prior to the action. Typical responses include a decrease or increase in the energy of the μ rhythm (8 - 12 Hz) and β rhythm (12 - 30 Hz) in the above regions, which is defined as motor activation here. Detecting whether the EEG response conforms to the typical motor activation characteristics can determine whether the patient has correctly performed the motor imagery task to ensure the effectiveness of stimulating the motor area of the brain.

[0045] In the embodiment, the detection method for whether the motor area of the brain is activated is as follows:

[0046] ① The EEG acquisition module obtains the EEG data in the imagined task state and the EEG data in the rest state generated by the patient's corresponding motor imagery task, and performs preprocessing.

[0047] Specifically, the EEG data in the rest state is the most recent data when performing this motor imagery task to ensure a closer fit to the actual EEG situation. The lengths of the EEG data in one motor imagery state and the EEG data in the rest state are empirical values, usually taken as 3 seconds and 1 second respectively. Preprocessing steps are performed separately, and if artifacts are found, the data is discarded.

[0048] ② The EEG analysis module takes the absolute value of the preprocessed EEG data in the imagined task state and the EEG data in the rest state, and then takes the average. The average motor imagery energy and the average rest state energy are obtained.

[0049] Specifically, when the EEG acquisition module performs preprocessing, the band-pass filtering range is (5 Hz - 35 Hz), which completely covers the rhythm ranges of the μ rhythm (8 - 12 Hz) and β rhythm (12 - 30 Hz). After taking the absolute value, the energy situation of the EEG can be more clearly reflected.

[0050] ③ For a certain channel ch, subtract the average rest state energy from the average motor imagery energy of this channel, and then take the absolute value to obtain the average motor imagery energy difference Pd of the EEG signal of this channel. ch。It can be more significantly observed that the energy increases or decreases compared to the resting-state data.

[0051] ④ Calculate the difference index of the change in the movement area, including the first difference index I motor1 and the second difference index I motor2 ;

[0052]

[0053] where Pd chl is the average motor imagery energy difference of the electroencephalogram signal of the central lead in the left motor brain area, Pd chr is the average motor imagery energy difference of the electroencephalogram signal of the central lead in the right motor brain area, Σ surl Pd ch is the sum of the average motor imagery energy differences of the electroencephalogram signals of the peripheral leads in the left brain area, ∑ surr Pd ch is the sum of the peripheral leads in the left and right motor brain areas, N surl is the number of peripheral leads in the left motor brain area, N surr is the number of peripheral leads in the right motor brain area. I motor1 reflects the increase in energy in the motor brain area, I motor2 reflects the decrease in energy in the motor brain area.

[0054] ⑤ Determine whether the motor brain area is activated according to the difference index of the change in the movement area.

[0055] As an example, if I motor1 > the first threshold (such as 1.2), or I motor2 < the second threshold (such as 0.83), then the motor brain area is activated, otherwise the motor brain area is not activated.

[0056] Considering that the motor brain area of the patient may be damaged and the relevant motor activation imagination is not obvious, the determination conditions for the activation phenomenon need to be compatible with various situations. Whether there is an increase or decrease in the corresponding energy in the left or right brain area can be regarded as effective activation. The relevant thresholds are empirical values obtained based on clinical data statistics.

[0057] In the embodiment, the preset training modalities include the first training modality, and in some embodiments, the second training module and / or the third training module are also included, which can respectively correspond to the assistance mode, the active mode, and the resistance mode of the rehabilitation task. The assistance mode is relatively easy, and the resistance mode is the most difficult. Provide a certain degree of challenge for patients at different stages and enrich the rehabilitation training content, so as to prompt patients to maintain attention and initiative, which is beneficial to rehabilitation.

[0058] Specifically, the working mode of the periodic detection of the electroencephalogram acquisition module is as follows: Before the patient starts the rehabilitation task, maintain the resting state for several seconds, and then start to complete the rehabilitation task. At the beginning of t motorAfter a certain time, the EEG analysis module processes the EEG data of the most recent t motor time, and the limb movement control execution module returns a control instruction to control the end effector to perform the corresponding rehabilitation training action. Subsequently, the EEG acquisition module intercepts the most recent t gap time, intercepts the EEG data of the most recent t motor time and processes it until the rehabilitation task is completed. t motor is the length of the motor imagery data, and the empirical value is generally taken as 3 seconds. t gap is the control instruction output interval, and the empirical value is taken as 1 second.

[0059] The steps for the motor imagery rehabilitation system based on the brain-computer interface to process EEG data are as follows:

[0060] 1. The EEG acquisition module intercepts the EEG data and completes the preprocessing steps. The preprocessing method is the same as described before.

[0061] 2. The EEG acquisition module sends the preprocessed data to the EEG analysis module to obtain the motor intention (intention classification and / or intention score).

[0062] 3. The EEG analysis module processes the EEG data to obtain whether the motor brain area is activated.

[0063] Assistive mode: The motor imagery rehabilitation system periodically detects. If the user's motor function area of the brain is correctly activated or the system recognizes the correct motor intention, it controls the end effector (such as a grip or other mechanical structures such as exoskeletons) to move a fixed distance along a fixed route, controls the user to complete the corresponding limb movement, and completes the neurofeedback. If the user cannot trigger the movement of the grip for several consecutive cycles, the grip is triggered to move a fixed distance. It is suitable for patients who are not familiar with this therapy to help them gradually master the skills.

[0064] Before the start of the assistive mode training task, the system or the rehabilitation therapist will inform the patient how to move the grip (here it is to move the right hand to the right, horizontally), and the patient performs the motor imagery task. The system periodically processes the EEG data to obtain the intention classification, intention score, and activation status of the motor brain area in a single cycle.

[0065] As an example, the limb movement control execution module is specifically configured to: if any of the following conditions is met, it is determined that the requirements of the first training modality are met, and the end effector is controlled to perform the corresponding rehabilitation training action of the first training modality (assistive mode);

[0066] Condition 1: The motor imagery classification result is correct;

[0067] Condition 2: The patient's motor brain area has been activated;

[0068] Condition 3: The patient has been unable to move the end effector by himself for a continuous set number of cycles (N f cycles);

[0069] The rehabilitation training actions corresponding to the first training modality include: the end effector moves a fixed distance.

[0070] In the assistance mode, the third condition is compensatory to prevent the patient from being unable to initiate the grip movement due to poor EEG quality or other reasons. N f is an empirical value, usually taken as 3, that is, the system triggers the grip to move a fixed distance every 4 seconds at most, helping the patient better master the motor imagery technique.

[0071] Active mode: The system periodically detects. If the user's motor brain area is correctly activated and the intention score is greater than or equal to the intention score threshold (i.e., the degree of motor intention reaches the threshold), it controls the grip to move a fixed distance along a fixed route.

[0072] Before the start of the active mode training task, the system or the rehabilitation therapist will inform the patient how to move the grip (here it is to move the right hand to the right, horizontally), and the patient performs the motor imagery task. The system periodically processes the EEG data to obtain the intention classification, intention score, and activation status of the motor brain area in a single cycle.

[0073] As an example, the limb movement control execution module is specifically configured as follows: the preset training modality is the second training modality (active mode). When the following conditions are met simultaneously, it meets the requirements of the active mode and performs the rehabilitation training actions corresponding to the second training modality;

[0074] Condition 1: The motor imagery classification result is correct;

[0075] Condition 2: The patient's motor brain area has been activated;

[0076] Condition 4: The intention score of the patient's current intention classification is greater than or equal to the intention score threshold gc riaht ;

[0077] The rehabilitation training actions corresponding to the second training modality include: using the end effector to execute a fixed distance. Among them, in condition 3, gc right is the intention score threshold of the current intention classification. The current intention classification here is to move the right hand to the right. The threshold is initially

[0078] Resistance mode: The system periodically detects. If the user's motor brain area is correctly activated and the degree of motor intention reaches the threshold. The system controls the grip to move a fixed distance along a fixed route according to the user. If the user triggers the grip to move in several consecutive cycles, the distance the grip moves in each cycle gradually increases until the maximum value. If the user fails to trigger the grip to move in a certain cycle, the distance the grip moves in each cycle decreases until the minimum value.

[0079] Before the start of the resistance mode training task, the system or the rehabilitation therapist will inform the patient how to move the grip (here, the right hand moves horizontally to the right), and the patient performs the motor imagery task. The system periodically processes the EEG data to obtain the intention classification, intention score, and activation of the motor brain area in a single cycle.

[0080] Specifically, this mode simulates the movement of a handle with a certain resistance. When the patient continuously triggers the handle to move, the handle initially moves only a small distance, and the subsequent amplitude will gradually increase. If the patient cannot trigger the handle to move, the movement amplitude of the handle will quickly decrease to 0, giving the patient more feedback and greater difficulty to promote their concentration.

[0081] As an example, the limb movement control execution module is specifically configured as follows:

[0082] The preset training modality is the third training modality (impedance mode). When the following conditions are met simultaneously, the requirements of the impedance mode are met, and the first rehabilitation training sub-action corresponding to the third training modality is executed; when the following conditions cannot be met simultaneously, the requirements of the impedance mode are not met, and the second rehabilitation training sub-action corresponding to the third training modality is executed;

[0083] Condition 1: The motor imagery classification result is correct;

[0084] Condition 2: The patient's motor brain area has been activated;

[0085] Condition 4: The intention score of the patient's current intention classification is greater than or equal to the intention score threshold;

[0086] Among them, the first rehabilitation training sub-action includes: determining the first moving distance according to the first method and moving the first distance using the end effector; the second rehabilitation training sub-action includes: determining the second moving distance according to the second method and moving the second distance using the end effector.

[0087] Among them, determining the first moving distance according to the first method, the expression is as follows:

[0088]

[0089] Among them, is the moving distance of the previous cycle, is the fixed moving distance set in the active mode and assist mode. When the requirements of the impedance mode are met, The minimum value of is 0.5d move , and the maximum value is 1.5d move . K1 can take 0.1.

[0090] When the requirements of the impedance mode are not met, the second moving distance is determined according to the second method, and the expression is:

[0091]

[0092] Among them has a minimum value of 0 and a maximum value of 0.25d move , that is, if the trigger requirement cannot be met for two consecutive cycles, the handle stops moving. K1 can take 0.25.

[0093] In some embodiments, the limb movement control execution module is further configured to set an initial threshold for the intention score, and the setting method is as follows: for all samples corresponding to the correctly classified categories in the classification recognition results, arrange the intention scores in descending order, and determine the intention scores corresponding to the top set proportion (such as 70%) as the initial threshold for the intention score.

[0094] Among them, 70% is an empirical value, indicating that more than 70% of the offline samples can meet the requirements, and it is the initial threshold in turn. This value determines the initial difficulty of the system's active mode and resistance mode.

[0095] In some embodiments, the intention score threshold will be updated in real time to adjust the difficulty of the rehabilitation training. It is used to adjust the threshold level and the difficulty range in the active mode and the resistance mode, which is beneficial to controlling the difficulty within a certain range, motivating the user to invest sufficient concentration, and promoting the curative effect.

[0096] Specifically, in the active and resistance modes, after the patient completes N t cycles, the system will adjust the initial threshold of the intention score according to the patient's performance to ensure that the patient is within a certain difficulty range.

[0097] In some embodiments, the limb movement control execution module is configured to, after N t sampling cycles, when the patient's motor imagery classification result is correct, perform the following steps:

[0098] Sort the set of intention scores Score last of the correctly classified category in descending order to obtain the sequence group i ;

[0099] If the number of intention scores in the set of intention scores of the category is less than the set value (for example, the number of data is less than 10), then divide the sequence group i into intervals according to the set proportional interval; determine the interval in the sequence where the intention score threshold of the previous sampling cycle is located as the initial interval, and use the smallest intention score value in the next interval of the initial interval as the updated intention score threshold

[0100] If the number of intention scores in the set of intention scores of the correctly classified category is greater than or equal to the set value, then determine the intention scores corresponding to the top set proportion as the intention score threshold

[0101] The motor imagery rehabilitation system based on a brain-computer interface provided by the embodiments of the present application is equipped with a variety of rehabilitation task modes based on the brain-computer interface, recognizes the activation of the patient's brain regions, adaptively adjusts the difficulty, and is more conducive to stimulating the initiative of the patient and promoting the rehabilitation of the affected limb.

[0102] Based on the same inventive concept as the motor imagery rehabilitation system based on a brain-computer interface provided by the above embodiments, the embodiments of the present application also provide a motor imagery rehabilitation method based on a brain-computer interface, including:

[0103] Collect the motor imagery EEG data and resting-state EEG data of the patient;

[0104] Perform motor imagery classification and recognition based on the motor imagery EEG data to obtain a classification and recognition result; based on the motor imagery EEG data and the resting-state EEG data, obtain the average imagery task energy and the average resting-state energy; based on the average imagery task energy and the average resting-state energy, as well as the EEG data energy of the central leads of the left and right motor brain regions, determine the motor area change difference index; according to the motor area change difference index, determine whether the patient's motor brain region is activated;

[0105] According to the preset training modality requirements, control the end effector to perform corresponding rehabilitation training actions according to the classification and recognition result and whether the / transport brain region is activated.

[0106] For the specific implementation manners of the steps in this embodiment, reference may be made to the corresponding modules in the above embodiments, and details are not described herein.

[0107] The above has introduced in detail the motor imagery rehabilitation system and method based on a brain-computer interface provided by the present application. Specific examples are used in this article to elaborate on the principle and implementation manner of the present application. The description of the above embodiments is only used to help understand the concept of the present application and should not be construed as a limitation on the protection scope of the present application.

Claims

1. A movement imagination rehabilitation system based on a brain-computer interface, characterized in that, Comprising: An electroencephalogram (EEG) acquisition module configured to acquire the EEG data of the patient in the motor imagery state and the resting state. An EEG analysis module configured to perform motor imagery classification and recognition based on the motor imagery EEG data to obtain a classification and recognition result; obtain the average imagery task energy and the average resting state energy based on the motor imagery EEG data and the resting state EEG data; determine the motor area change difference index based on the average imagery task energy, the average resting state energy, and the EEG data energy of the central leads in the left and right motor areas; and determine whether the patient's motor area is activated according to the motor area change difference index. A limb movement control execution module configured to control the end effector to perform corresponding rehabilitation training actions according to the classification and recognition result and whether the / transportation brain area is activated according to the requirements of a preset training modality.

2. The motor imagery rehabilitation system based on a brain-computer interface according to claim 1, wherein The limb movement control execution module is specifically configured as follows: When the preset training modality is the first training modality, if any of the following conditions is satisfied, the end effector is controlled to perform the rehabilitation training actions corresponding to the first training modality; Condition 1: The motor imagery classification result is correct; Condition 2: The patient's motor area has been activated; Condition 3: The patient has been unable to move the end effector by himself within a continuous set period. The rehabilitation training actions corresponding to the first training modality include: moving the end effector a fixed distance.

3. The motor imagery rehabilitation system based on a brain-computer interface according to claim 1, wherein The limb movement control execution module is specifically configured as follows: When the preset training modality is the second training modality, if the following conditions are satisfied simultaneously, the end effector is controlled to perform the rehabilitation training actions corresponding to the second training modality; Condition 1: The motor imagery classification result is correct; Condition 2: The patient's motor area has been activated; Condition 4: The intention score of the current intention classification of the patient is greater than or equal to the intention score threshold. The rehabilitation training actions corresponding to the second training modality include: moving the end effector a fixed distance.

4. The motor imagery rehabilitation system based on a brain-computer interface according to claim 3, characterized in that, The limb movement control execution module is specifically configured as follows: When the preset training modality is the third training modality, if Conditions 1, 2, and 4 are satisfied simultaneously, the first rehabilitation training sub-action corresponding to the third training modality is executed; if Conditions 1, 2, and 4 cannot be satisfied simultaneously, the second rehabilitation training sub-action corresponding to the third training modality is executed; The first rehabilitation training sub-action includes: moving the end effector a first distance, and the first distance is determined in a first manner; The second rehabilitation training sub-action includes: moving the end effector a second distance, and the second distance is determined in a second manner.

5. The motor imagery rehabilitation system based on a brain-computer interface according to claim 3, wherein The EEG analysis module is further configured to determine the intention score of the current sample based on the distances between all samples of each category in the motor imagery EEG data and the corresponding classification hyperplane, and the expression is: in For category c i The intention score of the jth sample, For category c i The distance between the jth sample after classification and the corresponding classification hyperplane, μ i For category c i The distance between all samples and the corresponding classification hyperplane The average value, σ i For category c i The distance between all samples and the corresponding classification hyperplane Standard deviation of category c i N is the correct category for the classification recognition result. c is the number of categories.

6. The motor imagery rehabilitation system based on a brain-computer interface according to claim 5, characterized in that, Category c i The intention score of the j-th sample has a value range of 1 / N c to 1, and the expression is as follows:

7. The motor imagery rehabilitation system based on a brain-computer interface according to claim 3, wherein The limb movement control execution module is further configured to set an initial intention score threshold, and the setting method is as follows: arrange the intention scores corresponding to all samples in the category with the correct classification recognition result in descending order, and determine the intention scores corresponding to the top set proportion as the initial intention score threshold.

8. The motor imagery rehabilitation system based on a brain-computer interface according to claim 3, characterized in that, The limb movement control execution module is further configured to, t after N sampling periods, when the classification result of the patient's motor imagery is correct, perform the following steps: Sort the set of intention scores corresponding to the correctly classified categories in descending order to obtain a sequence; If the number of intention scores in the intention score set is less than a set value, divide the sequence into intervals at set proportional intervals; determine the interval in the sequence where the intention score threshold was located in the previous sampling period as the initial interval, and use the smallest intention score value in the next interval of the initial interval as the updated intention score threshold; If the number of intention scores in the intention score set is greater than or equal to the set value, determine the intention scores corresponding to the positions in the top set proportion as the intention score threshold.

9. The motor imagery rehabilitation system based on a brain-computer interface according to claim 1, characterized in that, Determine the expression of the motion area change difference index as follows: Among them, Pd chl is the average motor imagery energy difference of the EEG signals of the central leads in the left motor brain area. Pd chr is the average motor imagery energy difference of the EEG signals of the central leads in the right motor brain area. ∑ surl Pd ch is the sum of the average motor imagery energy differences of the EEG signals of the peripheral leads in the left brain area. ∑ surr Pd ch is the sum of the peripheral leads in the left and right motor brain areas. N surl is the number of peripheral leads in the left motor brain area. N surr is the number of peripheral leads in the right motor brain area; I motor1 is the first difference index. I motor2 is the second difference index; If the first difference index is greater than the first threshold, or the second difference index is less than the second threshold, the motor brain area is activated; otherwise, the motor brain area is not activated.

10. A motor imagery rehabilitation method based on a brain-computer interface, characterized in that, Include: Collect the motor imagery EEG data and resting-state EEG data of the patient; Perform motor imagery classification and recognition based on the motor imagery EEG data to obtain a classification and recognition result; Based on the motor imagery EEG data and the resting-state EEG data, obtain the average imagery task energy and the average resting-state energy; based on the average imagery task energy and the average resting-state energy, as well as the energy of the EEG data of the central leads in the left and right motor brain areas, determine the motion area change difference index; according to the motion area change difference index, determine whether the patient's motor brain area is activated; According to the preset training modality requirements, control the end effector to perform corresponding rehabilitation training actions based on the classification and recognition result and whether the / transportation brain area is activated.

Images