Brain injury function training device based on brain-computer interface

Abstract

The invention is applicable to the field of brain-computer interfaces and rehabilitation medicine, and provides a brain injury function training device based on a brain-computer interface, which comprises a support frame, an electroencephalogram acquisition module, a signal processing module, a training control module, a feedback module, an evaluation module and a user interaction module, in the first stage, accurate collection and preprocessing of electroencephalogram signals are achieved through a flexible electrode array and a low-noise amplification and self-adaptive noise reduction technology; secondly, through linkage of an evaluation module and a training control module, dynamic optimization of a personalized training scheme is achieved; thirdly, constructing a visual, auditory and tactile multi-mode feedback and quantitative evaluation closed loop; according to the brain injury function training device, the problems that an existing device is low in electroencephalogram signal collection precision, fixed in training scheme and lack of a training-evaluation-adjustment closed loop are solved, the accuracy, adaptability and user participation degree of brain injury function training are improved, and the brain injury function training device is suitable for function recovery training of movement, cognition, language and the like of brain injury patients with cerebral apoplexy, cerebral trauma and the like.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of brain-computer interface and rehabilitation medicine, and particularly relates to a brain injury function training device based on a brain-computer interface. BACKGROUND

[0002] The existing brain-computer interface brain injury training device uses a hard electrode for the brain electrical signal acquisition component, which has poor adhesion to the scalp and lacks effective signal amplification and noise reduction mechanisms, resulting in that the collected brain electrical signals are seriously interfered by electromyography, electrocardiography, environmental electromagnetic and the like, the signal precision is low, and then the training instruction generated by the subsequent training control module deviates from the actual brain electrical intention of the user, affecting the training accuracy and failing to meet the fine training needs of brain injury patients.

[0003] The training scheme of the existing device is mostly a preset fixed template, which can only select a single scheme according to the initial brain injury condition of the user, cannot analyze the brain electrical signal changes and training completion condition in the training process of the user in real time, and cannot dynamically adjust the training difficulty, task type and training duration, resulting in that the training scheme does not match the real-time recovery state of the user, easy to appear insufficient training or overtraining, reduces the training efficiency and prolongs the rehabilitation cycle.

[0004] The feedback mode of the existing device is mostly single visual feedback, lacks multi-modal feedback mechanisms such as hearing and touch, the immersion and participation of the user in the training process are low, and a closed loop of "training-evaluation-adjustment" is not formed, the recovery evaluation report cannot be generated in real time, the user and medical staff are difficult to master the training effect and recovery progress, it is not conducive to timely adjusting the training strategy, and affects the overall rehabilitation effect.

[0005] Therefore, a brain injury function training device based on a brain-computer interface is needed to solve the above problems. SUMMARY

[0006] The purpose of the embodiment of the application is to provide a brain injury function training device based on a brain-computer interface to solve the problems proposed in the background.

[0007] To achieve the above purpose, the application provides the following technical scheme:

[0008] The application discloses a brain injury function training device based on a brain-computer interface, which comprises a support frame, an electroencephalogram acquisition module, a signal processing module, a training control module, a feedback module, an evaluation module and a user interaction module; the electroencephalogram acquisition module is detachably arranged on the support frame and used for acquiring electroencephalogram signals of a user; an input end of the signal processing module is electrically connected with an output end of the electroencephalogram acquisition module and used for pre-processing the electroencephalogram signals; an input end of the training control module is electrically connected with an output end of the signal processing module and used for generating a training scheme and outputting training instructions according to the pre-processed electroencephalogram signals; an input end of the feedback module is electrically connected with an output end of the training control module and used for receiving the training instructions and providing multi-modal training feedback to the user; input ends of the evaluation module are respectively electrically connected with the output end of the signal processing module and the output end of the training control module, an output end of the evaluation module is electrically connected with the input end of the training control module, the evaluation module is used for evaluating a function recovery state of the user according to the pre-processed electroencephalogram signals and a training instruction execution condition and feeding back evaluation results to the training control module to adjust the training scheme.

[0009] Through the orderly electrical connection of the modules, a complete function system from electroencephalogram signal acquisition to training adjustment is constructed, the modules work cooperatively, the device can realize accurate acquisition, individualized training and closed-loop evaluation, the device provides systematic function training support for brain injury patients and avoids the problem of fragmentation of functions of existing devices.

[0010] In a further technical scheme, the electroencephalogram acquisition module comprises a flexible electrode array and a signal amplification unit; the flexible electrode array is electrically connected with an input end of the signal amplification unit, and an output end of the signal amplification unit is electrically connected with an input end of the signal processing module; the flexible electrode array is used for directly contacting a scalp of the user to acquire electroencephalogram signals, and the signal amplification unit is composed of a low-noise operational amplifier and is used for amplifying the weak electroencephalogram signals acquired.

[0011] The flexible electrode array is made of a medical flexible material, can deform along a contour of the scalp, improves a bonding area with the scalp, reduces contact impedance and reduces signal distortion caused by poor contact; the signal amplification unit adopts a low-noise amplification circuit, can amplify the weak electroencephalogram signals while suppressing self noise, ensures that a high-quality original signal can be acquired by the subsequent signal processing module, solves the problem of low acquisition precision of existing devices and provides a basis for the accuracy of subsequent training instructions.

[0012] The further technical scheme is characterized in that the signal processing module comprises a filtering unit, a noise reduction unit and a feature extraction unit; the input end of the filtering unit is electrically connected with the output end of the electroencephalogram acquisition module, the output end of the filtering unit is electrically connected with the input end of the noise reduction unit, the output end of the noise reduction unit is electrically connected with the input end of the feature extraction unit, and the output end of the feature extraction unit is electrically connected with the input end of the training control module; the filtering unit is an RC band-pass filter circuit, which is used for filtering interference signals (such as power frequency interference and low frequency drift) of non-target frequency; the noise reduction unit adopts an adaptive noise reduction circuit based on a least mean square error (LMS) algorithm, and eliminates noise signals such as electromyography and electrocardiography by adjusting filter coefficients in real time, and the core iterative formula is:

[0013] w(n+1)=w(n)+2μe(n)x(n),

[0014] wherein w(n) is a filter coefficient vector at n time, w(n+1) is an updated filter coefficient vector at n+1 time, μ is a step factor (the value range is 0<μ<1 / λ_max, λ_max is the maximum eigenvalue of the input signal autocorrelation matrix, which is used for ensuring the convergence of the algorithm), e(n) is an error signal at n time (i.e. the difference between the expected signal and the filter output signal), and x(n) is an input signal vector at n time; the feature extraction unit is an embedded microprocessor, which extracts features by calculating the power spectral density (PSD) of the electroencephalogram signal by the Welch method, and the core formula is:

[0015] PSD(f)=(1 / (N×Fs))×FFT(x(n)×w(n))²,

[0016] wherein PSD(f) is the power spectral density at frequency f, N is the data window length, Fs is the signal sampling frequency, x(n) is the electroencephalogram sampling signal at n time, w(n) is a window function (Hanning window is used, which is used for reducing spectral leakage), FFT(·) is a fast Fourier transform operation, and ² is a modulus square operation; the feature extraction unit extracts waveband features related to brain function, such as μ waveband (8-13Hz, related to motor imagery), β waveband (14-30Hz, related to motor execution), θ waveband (4-7Hz, related to cognitive attention) and γ waveband (30-50Hz, related to language function);

[0017] The filtering unit retains the target frequency band signal, the LMS adaptive noise reduction algorithm realizes real-time noise suppression by dynamically adjusting the coefficient, the power spectral density calculation can accurately quantify the energy change of each function-related waveband, and the three cooperate to ensure that the output electroencephalogram features can truly reflect the brain function state of the user, thereby providing reliable decision basis for the training control module and avoiding training deviation caused by invalid signals.

[0018] The further technical scheme is characterized in that the training control module comprises a personalized scheme storage unit and a scheme calling unit; the personalized scheme storage unit is electrically connected with the scheme calling unit; the input end of the scheme calling unit is electrically connected with the output end of the signal processing module and the output end of the evaluation module respectively; the output end of the scheme calling unit is electrically connected with the input end of the feedback module; the personalized scheme storage unit is used for storing training schemes corresponding to different brain injury types and different recovery stages; and the scheme calling unit is used for calling or adjusting the training scheme according to the electroencephalogram signal features and the evaluation result.

[0019] The personalized scheme storage unit pre-stores training schemes for different injury types and different recovery stages; the scheme calling unit automatically calls or adjusts the scheme by analyzing the electroencephalogram signal features and the evaluation result, realizes dynamic personalized adjustment of the training scheme, solves the problem of fixed scheme of the existing device, and improves the training efficiency.

[0020] The further technical scheme is characterized in that the feedback module comprises a visual feedback unit, an auditory feedback unit and a tactile feedback unit; the input end of each of the visual feedback unit, the auditory feedback unit and the tactile feedback unit is electrically connected with the output end of the training control module; the visual feedback unit is used for displaying training tasks, progress and electroencephalogram signal related information; the auditory feedback unit is used for playing voice prompts and training result feedback; and the tactile feedback unit is used for providing tactile prompts through vibration.

[0021] The multi-modal feedback provides real-time feedback for the user from the visual, auditory and tactile dimensions, can effectively improve the immersion and participation of the user in the training process, avoids user fatigue caused by single feedback, and solves the problem of single feedback mode of the existing device.

[0022] The further technical scheme is characterized in that the evaluation module comprises a recovery index analysis unit and an evaluation report generation unit; the input end of the recovery index analysis unit is electrically connected with the output end of the signal processing module and the output end of the training control module respectively; the output end of the recovery index analysis unit is electrically connected with the input end of the training control module and the input end of the evaluation report generation unit respectively; the recovery index analysis unit obtains a recovery index (RI) by weighted calculation of a training accuracy (Acc) and a change rate of electroencephalogram signal features (ΔFE), and the core formula is:

[0023] RI = α × Acc + (1-α) × ΔFE,

[0024] Wherein, RI is a recovery index (value range 0-1, the greater the value, the better the recovery state), a is a weight coefficient (value range 0< a <1, set by medical staff according to training goals, a = 0.6 during motor function training, a = 0.5 during cognitive / language function training), Acc is the training accuracy (i.e. the ratio of the number of successful training tasks to the total number of tasks), and Delta FE is the brain electrical signal feature change rate (i.e. the difference between the current training cycle feature value and the last cycle feature value divided by the last cycle feature value); the evaluation report generation unit is configured to generate a recovery evaluation report according to the recovery index, the report including the RI value, the RI trend graph, the waveband feature comparison table, and the training suggestion;

[0025] The subjective training performance (accuracy) is combined with the objective brain electrical signal feature (change rate) through the weighting formula, avoiding the one-sidedness of single index evaluation; the recovery index calculation logic is transparent, medical staff can adjust the weight to adapt to the training needs of different patients, and the evaluation report intuitively displays the recovery progress, solving the problem of single evaluation dimension and non-quantitative results of existing devices.

[0026] The further technical solution further comprises a user interaction module; an output end of the user interaction module is electrically connected with an input end of the training control module, and an input end of the user interaction module is electrically connected with an output end of the evaluation module; the user interaction module comprises a touch panel and a data interface, which are used for the user to input personal information, training preferences and view evaluation reports, and are also used for medical staff to input training parameters and adjust evaluation standards;

[0027] The touch panel facilitates the user to complete information input and report viewing through simple touch operation, and the data interface supports medical staff to input personalized training parameters and adjust evaluation standards, improving the clinical applicability and flexibility of the device.

[0028] The further technical solution adopts a head-mounted structure, comprising a main frame and an elastic adjusting belt; the main frame is used for mounting the brain electrical signal acquisition module, and the elastic adjusting belt is arranged on both sides of the main frame and is used for adapting to the head circumference of different users and ensuring stable contact between the brain electrical signal acquisition module and the scalp;

[0029] The head-mounted structure is convenient for the user to wear, the elastic adjusting belt can be self-adaptively adjusted according to the head circumference of different users, ensuring stable contact between the brain electrical signal acquisition module and the scalp during the training process, avoiding interruption or distortion of signal acquisition caused by loosening of the device, and improving the stability and comfort of the device.

[0030] The further technical solution is that the flexible electrode array is made of medical silicone material, and an electrically conductive gel layer is arranged on the surface of the electrode array; the electrically conductive gel layer is used for reducing the contact impedance between the electrode and the scalp and improving the stability of brain electrical signal acquisition;

[0031] Medical-grade silicone material has good biocompatibility and flexibility, which can reduce scalp irritation from prolonged wear; the conductive gel layer can effectively reduce the contact impedance between the electrodes and the scalp, reduce signal transmission loss, ensure that the acquired EEG signals have a higher signal-to-noise ratio, and improve the stability and reliability of signal acquisition.

[0032] Compared with the prior art, the beneficial effects of the present invention are:

[0033] This invention achieves precise acquisition and preprocessing of EEG signals through the coordinated operation of a flexible electrode array and signal amplification unit in the EEG acquisition module, and the collaborative work of a filtering unit and a noise reduction unit in the signal processing module. The flexible electrode array improves scalp fit and reduces contact interference, the signal amplification unit enhances weak signals, the filtering unit filters out irrelevant frequency interference, and the noise reduction unit eliminates physiological noise. This quadruple protection ensures that the EEG signals output to the training control module are accurate and reliable, significantly improving the accuracy of training commands and solving the problem of low acquisition accuracy in existing devices. It provides a foundation for refined functional training for patients with brain injuries.

[0034] This invention establishes a dynamic adjustment mechanism for personalized training programs by electrically connecting the personalized program storage unit of the training control module with the recovery index analysis unit of the evaluation module. The personalized program storage unit covers programs for various types of brain injury and recovery stages. The recovery index analysis unit evaluates the user's training effect in real time and feeds it back to the training control module. The program invocation unit automatically invokes or adjusts the program based on the evaluation results, achieving precise matching between the training program and the user's real-time recovery status. This avoids overtraining or undertraining, solves the problem of fixed programs in existing devices, effectively improves training efficiency, and shortens the rehabilitation cycle.

[0035] This invention, through the linkage of the visual, auditory, and tactile multimodal feedback units of the feedback module with the assessment report generation unit and training control module of the assessment module, forms a complete closed loop of "collection-processing-training-feedback-assessment-adjustment". Multimodal feedback enhances user training participation and immersion from multiple sensory dimensions, reduces training fatigue, and the assessment report generation unit intuitively displays the recovery progress. At the same time, the assessment results provide a basis for adjusting the training plan, achieving seamless connection between training and assessment. This solves the problem of the lack of a closed loop in existing devices, helping users and medical staff to grasp the recovery status in a timely manner, optimize training strategies, and improve the overall rehabilitation effect.

[0036] To more clearly illustrate the structural features and effects of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. Attached Figure Description

[0037] Figure 1 This is a block diagram of the overall structure of the present invention;

[0038] Figure 2It is a structural schematic diagram of the EEG acquisition module of the application;

[0039] Figure 3 It is an internal structure block diagram of the signal processing module of the application;

[0040] Figure 4 It is a connection relationship diagram of the training control module of the application;

[0041] Figure 5 It is a composition schematic diagram of the feedback module of the application;

[0042] Figure 6 It is a personalized training scheme adjustment flowchart of the application.

[0043] In the figure: 1, EEG acquisition module; 11, flexible electrode array; 111, medical silica gel electrode sheet; 12, signal amplification unit; 121, low noise amplifier; 2, signal processing module; 21, filter unit; 211, band-pass filter circuit; 22, noise reduction unit; 221, adaptive noise reduction circuit; 23, feature extraction unit; 3, training control module; 31, personalized scheme storage unit; 311, flash memory chip; 32, scheme calling unit; 321, microcontroller; 4, feedback module; 41, visual feedback unit; 411, TFT color display screen; 42, auditory feedback unit; 421, micro speaker; 43, tactile feedback unit; 5, evaluation module; 51, recovery index analysis unit; 511, digital signal processor; 52, evaluation report generation unit; 521, evaluation report software module; 6, user interaction module; 61, touch panel; 62, data interface. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical scheme and advantages of the application more clear and obvious, the application is further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the application, and are not used to limit the application.

[0045] The specific implementation of the application is described in detail below in combination with specific examples.

[0046] Example 1

[0047] This example is aimed at the upper limb motor imagination training of stroke patients with motor dysfunction;

[0048] As Figures 1-6As shown, the embodiment of the present application provides a brain injury function training device based on a brain-computer interface. The support frame adopts a head-mounted body, is equipped with an elastic adjusting belt, and can be adjusted in tightness according to the head circumference of a patient. The flexible electrode array 11 of the electroencephalogram acquisition module 1 is eight medical silica gel electrode pieces 111, which are respectively arranged at positions corresponding to the scalp C3, C4 (motor area), P3, P4 (sensory area) and the like on the inner side of the head-mounted body, and the surface of the electrode piece is covered with a conductive gel layer. The signal amplification unit 12 adopts a low-noise amplifier 121, and the amplification multiple can be adaptively adjusted. The filter unit 21 of the signal processing module 2 is a band-pass filter circuit 211 (filtering frequency band 0.5-30 Hz), the noise reduction unit 22 is an adaptive noise reduction circuit 221 (based on the least mean square error algorithm), and the feature extraction unit 23 is an embedded microprocessor, which mainly extracts the energy features of the mu band and beta band related to motor imagination.

[0049] In this embodiment, after the patient wears the support frame, the elastic adjusting belt ensures that the medical silica gel electrode piece 111 is in contact with the scalp without gap, the conductive gel layer controls the contact impedance at a low level, and reduces the signal distortion caused by poor contact. The low-noise amplifier 121 amplifies the micro-volt level motor imagination electroencephalogram signal, and avoids signal attenuation. The band-pass filter circuit 211 filters the power frequency interference and low frequency drift, and the adaptive noise reduction circuit 221 iteratively updates the filter coefficients through the LMS algorithm: the initial moment w(0)=[0,0,...,0] (16-order coefficient vector), the input signal x(n) is the filtered electroencephalogram signal, the expected signal is a preset “right hand clenched fist” standard mu band signal, the error signal e(n) is calculated in real time and used to adjust w(n), and finally the myoelectric noise suppression rate is significantly improved. The feature extraction unit 23 calculates the mu band energy value through the PSD formula, and if the energy value is higher than the threshold value (the threshold value is 1.5 times the mu band energy of the patient in the resting state before training), it is determined as “effective motor imagination”. Through the system, the training control module 3 can generate training instructions based on the accurate signal and the brain electrical intention of the patient, for example, when the mu band energy of C4 lead meets the standard, the visual feedback unit 41 displays “right hand clenched fist success”, and the tactile feedback unit 43 vibrates once, helping the patient to establish the neural connection of “brain electrical intention-motor feedback”, solving the problem of low collection accuracy of the existing device, and improving the accuracy of motor function training. Figure 1

[0050] Embodiment 2

[0051] This embodiment is aimed at the attention training of patients with cognitive attention disorder after brain trauma.

[0052] The difference between this embodiment and embodiment 1 is that:

[0053] ​The individualized scheme storage unit 31 of the training control module 3 is a flash memory chip 311, which pre-stores three attention training schemes (mild: digital recognition; moderate: digital sequence memory; severe: digital and pattern matching); the scheme calling unit 32 is a microcontroller 321; the recovery index analysis unit 51 of the evaluation module 5 is a digital signal processor 511, which analyzes the changes of the attention-related θ band energy and the training completion condition (accuracy rate, reaction time length) to determine the recovery index; and the evaluation report generation unit 52 is an evaluation report software module 521.

[0054] In this embodiment, when the patient initially trains, the medical staff inputs that the patient is “moderate cognitive impairment” through the touch panel 61, and the scheme calling unit 321 calls the “digital sequence memory” scheme; during the training, the signal processing module 2 regularly sends the θ band features to the recovery index analysis unit 511, and the training control module 3 synchronously sends the training completion condition; if the analysis finds that the patient's θ band energy decreases (attention improves) and the accuracy rate significantly improves, the recovery index analysis unit 511 generates a “difficulty increase” result, and the scheme calling unit 321 calls a higher difficulty digital sequence memory scheme; if the accuracy rate decreases, the difficulty is reduced; the recovery index calculation adopts an RI weighted formula, and the weight coefficient α=0.5 (cognitive training focuses on the balanced evaluation of the electroencephalogram features and the accuracy rate); the training accuracy rate Acc is “the number of correct times of digital sequence memory / the total number of times”, and the electroencephalogram feature change rate ΔFE is the change rate of the θ band power spectrum density (a decrease in the θ band energy indicates an improvement in attention, and the absolute value is taken when ΔFE is negative to participate in the calculation); in this process, the training scheme is dynamically adjusted according to the evaluation result, avoiding the insufficient training or overtraining caused by the fixed scheme of the existing device, realizing the precise matching of the scheme and the real-time recovery state of the patient, and improving the cognitive function training efficiency.

[0055] Embodiment 3

[0056] This embodiment is aimed at the language imagination training of patients with language dysfunction caused by cerebral infarction;

[0057] The difference between this embodiment and embodiment 2 is that:

[0058] The visual feedback unit 41 of the feedback module 4 is a TFT color display screen 411 (displaying tasks and waveforms), the auditory feedback unit 42 is a micro speaker 421 (playing guide / result voice), and the tactile feedback unit 43 is a micro vibration motor 431 (1 Hz vibration for task success and 5 Hz vibration for task failure); the recovery index analysis unit 51 of the evaluation module 5 analyzes the γ band energy related to language imagination and the number of task successes, and the evaluation report generation unit 52 generates a trend chart report, which is displayed through the touch panel 61.

[0059] In this embodiment, after the training starts, the display screen 411 displays the "apple" picture and the text, and the speaker 421 plays "imagine saying 'apple'"; during the patient's training, the display screen 411 displays the real-time brain wave form, and if the task is successful, the progress bar increases, "successful recognition" is played, and 1Hz vibration is performed; if the task fails, the progress bar remains unchanged, "try again" is played, and 5Hz vibration is performed; after the training is completed, the recovery index analysis unit 511 compares the gamma band energy and the number of successful times this time and last time to generate a report; the medical staff inputs a new vocabulary list through the data interface 62 to update the scheme of the flash memory chip 311; when the evaluation module calculates the RI, the weight coefficient a = 0.5 (the accuracy rate and the brain electrical feature are equally important in language training); the training accuracy rate Acc is the number of successful recognition times / total times in the language imagination task, and the brain electrical feature change rate DeltaFE is the change rate of the gamma band power spectrum density (the increase of the gamma band energy indicates the improvement of the language function); the multi-modal feedback improves the immersion of the patient, and the "training-evaluation-adjustment" closed loop ensures that the training continuously adapts to the recovery situation, solves the problem that the existing device lacks a closed loop, helps the patient to directly see the progress, enhances the rehabilitation confidence, and improves the language function training effect.

[0060] The working principle and use process of the application are as follows:

[0061] The complete working process of the brain injury function training device based on the brain-computer interface follows the closed loop logic of "signal acquisition-preprocessing-training control-multi-modal feedback-quantitative evaluation-scheme optimization", and the specific steps are as follows:

[0062] Device initialization and user preparation stage: the medical staff inputs the patient information (brain injury type, injury site, recovery stage) and training target (motor / cognitive / language function) through the touch panel 61 of the user interaction module 6, and adjusts the recovery index weight coefficient a of the evaluation module 5; at the same time, the patient wears the support frame, and ensures that the flexible electrode array 11 of the electroencephalogram acquisition module 1 is tightly attached to the scalp through the elastic adjusting belt, the signal amplification unit 12 starts self-checking, and confirms that the amplification multiple is adapted to the scalp impedance of the patient;

[0063] Electroencephalogram signal acquisition stage: the flexible electrode array 11 of the electroencephalogram acquisition module 1 acquires the electroencephalogram signal of the scalp of the patient in real time, the microvolt-level signal collected is transmitted to the signal amplification unit 12, and after being amplified by the low-noise operational amplifier, the millivolt-level signal is output to the signal processing module 2, so that the signal is not submerged by noise in the transmission process;

[0064] Electroencephalogram signal preprocessing stage:

[0065] (1) Filtering processing: the band-pass filter unit 21 (filtering frequency band 0.5-30Hz) of the signal processing module 2 filters the power frequency interference, low-frequency drift and high-frequency noise in the signal, and retains the target frequency band signal related to brain function;

[0066] (2) Adaptive noise reduction: The noise reduction unit 22 adopts the LMS algorithm, takes the "filtered electroencephalogram signal" as the input x(n), and takes the "preset function-related standard electroencephalogram signal" as the expected signal. The filter coefficient is adjusted in real time through the iterative formula w(n+1)=w(n)+2μe(n)x(n) to eliminate physiological noise such as electromyogram and electrocardiogram. The error signal e(n) converges below the set threshold, and the coefficient adjustment is stopped;

[0067] (3) Feature extraction: The feature extraction unit 23 adopts the Welch method, calculates the power spectral density of the specific waveband (μ / θ / γ waveband) of the target lead (C3 / C4 for motor training, P3 / P4 for cognitive training, and T3 / T4 for language training) according to the formula PSD(f)=(1 / (N×Fs))×FFT(x(n)×w(n))², extracts the average energy value of the waveband as the electroencephalogram feature, and transmits it to the training control module 3 and the evaluation module 5;

[0068] Personalized training control phase: After the training control module 3 receives the electroencephalogram feature, the scheme calling unit 32 first judges whether the feature meets the starting threshold of the current training scheme: if it meets, it calls the current scheme and outputs the training instruction; if it does not meet, it outputs the "adjust attention" prompt instruction to the feedback module 4; at the same time, the scheme calling unit 32 receives the recovery index RI fed back by the evaluation module 5, and if RI reaches the set high threshold, the next stage scheme with higher difficulty is called; if RI is lower than the set low threshold, the difficulty of the current scheme is reduced or the task type is adjusted;

[0069] Multi-modal feedback phase: After the feedback module 4 receives the training instruction, it synchronously starts three feedback units:

[0070] (1) Visual feedback unit 41: display training task text, progress bar and real-time electroencephalogram feature waveform, task success progress bar increases by a certain proportion, and failure displays prompt information;

[0071] (2) Auditory feedback unit 42: play voice prompts, task success plays positive feedback, and failure plays guidance information;

[0072] (3) Tactile feedback unit 43: task success with set low frequency vibration, failure with set high frequency vibration, and the association between brain electrical intention and feedback is strengthened through tactile stimulation.

[0073] Quantitative evaluation and scheme optimization phase:

[0074] (1) Recovery index calculation: The recovery index analysis unit 51 of the evaluation module 5 receives the "brain electrical characteristics" and "training completion status", and calculates the recovery index according to the formula RI = a x Acc + (1-a) x AF, wherein Acc = success times / total times, AF = (current characteristics - last period characteristics) / last period characteristics;

[0075] (2) Evaluation report generation: The evaluation report generation unit 52 generates an evaluation report containing "recovery level", "training suggestion" and "next period target RI" according to the RI value, RI change trend and characteristic comparison, and displays it through the touch panel 61 of the user interaction module 6;

[0076] (3) Scheme optimization closed loop: After receiving the RI, the training control module 3 automatically adjusts the training parameters if the RI does not reach the target. Medical staff can also import external evaluation data through the data interface 62 to update the scheme library of the individualized scheme storage unit 31, forming a continuous optimization closed loop of "evaluation-adjustment-training".

[0077] Training end and data storage phase:

[0078] After a single training, the device automatically stores the brain electrical characteristics, RI value, training record and evaluation report of this training. The data can be exported to the hospital system through the data interface 62 of the user interaction module 6 for medical staff to analyze the long-term recovery trend of the patient, providing data support for the development of rehabilitation plan;

[0079] In the whole working process, each module realizes real-time signal transmission through electrical connection, and the algorithm formula ensures the accuracy of signal processing and evaluation. The multi-modal feedback and closed loop adjustment ensure that the training adapts to the real-time state of the patient, solving the core problems of existing devices such as "inaccurate collection, fixed scheme and fuzzy evaluation", and realizing the precision, individualization and intelligence of brain injury function training.

[0080] The above only describes the preferred embodiments of the present application and does not limit the present application. Any modification, equivalent replacement and improvement within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A brain damage function training device based on a brain-computer interface, comprising a support frame, an electroencephalogram acquisition module (1), a signal processing module (2), a training control module (3), a feedback module (4) and an evaluation module (5), characterized in that: The electroencephalogram acquisition module (1) is detachably arranged on the support frame and is used for acquiring the electroencephalogram signal of the user. The input end of the signal processing module (2) is electrically connected with the output end of the electroencephalogram acquisition module (1), and is used for pre-processing the electroencephalogram signal. The input end of the training control module (3) is electrically connected with the output end of the signal processing module (2), and is used for generating a training scheme according to the pre-processed electroencephalogram signal and outputting a training instruction. The input end of the feedback module (4) is electrically connected with the output end of the training control module (3), and is used for receiving the training instruction and providing a multi-modal training feedback to the user. The input end of the evaluation module (5) is electrically connected with the output end of the signal processing module (2) and the output end of the training control module (3), the output end of the evaluation module (5) is electrically connected with the input end of the training control module (3), and the evaluation module (5) is used for evaluating the functional recovery state of the user according to the pre-processed electroencephalogram signal and the execution condition of the training instruction, and feeding back the evaluation result to the training control module (3) to adjust the training scheme.

2. The brain injury function training device based on brain-computer interface according to claim 1, characterized in that: The electroencephalogram acquisition module (1) comprises a flexible electrode array (11) and a signal amplification unit (12), the flexible electrode array (11) is electrically connected with the input end of the signal amplification unit (12), the output end of the signal amplification unit (12) is electrically connected with the input end of the signal processing module (2), the flexible electrode array (11) is used for directly contacting the scalp of the user to acquire the electroencephalogram signal, and the signal amplification unit (12) is composed of a low-noise operational amplifier and is used for amplifying the weak electroencephalogram signal acquired.

3. The brain injury function training device based on brain-computer interface according to claim 1, characterized in that: The signal processing module (2) comprises a filtering unit (21), a noise reduction unit (22) and a feature extraction unit (23), the input end of the filtering unit (21) is electrically connected with the output end of the electroencephalogram acquisition module (1), the output end of the filtering unit (21) is electrically connected with the input end of the noise reduction unit (22), the output end of the noise reduction unit (22) is electrically connected with the input end of the feature extraction unit (23), and the output end of the feature extraction unit (23) is electrically connected with the input end of the training control module (3), the filtering unit (21) is an RC band-pass filter circuit (211) and is used for filtering the interference signal of a non-target frequency, the noise reduction unit (22) is an adaptive noise reduction circuit (221) based on a least mean square error algorithm and is used for eliminating the noise signal by adjusting the filtering coefficient in real time, and the feature extraction unit (23) is an embedded microprocessor (321) and is used for extracting the electroencephalogram signal feature related to the functional recovery of brain injury by calculating the power spectral density of the electroencephalogram signal.

4. The brain impairment function training device based on brain-computer interface according to claim 1, characterized in that: The training control module (3) comprises a personalized scheme storage unit (31) and a scheme calling unit (32); the personalized scheme storage unit (31) is electrically connected with the scheme calling unit (32), the input end of the scheme calling unit (32) is electrically connected with the output end of the signal processing module (2) and the output end of the evaluation module (5), and the output end of the scheme calling unit (32) is electrically connected with the input end of the feedback module (4); the personalized scheme storage unit (31) is used for storing training schemes corresponding to different brain injury types and different recovery stages, and the scheme calling unit (32) is used for calling or adjusting the training scheme according to the electroencephalogram signal characteristics and the evaluation result.

5. The brain impairment function training device based on brain-computer interface according to claim 1, characterized in that: The feedback module (4) comprises a visual feedback unit (41), an auditory feedback unit (42) and a tactile feedback unit (43); the input end of the visual feedback unit (41), the auditory feedback unit (42) and the tactile feedback unit (43) is electrically connected with the output end of the training control module (3); the visual feedback unit (41) is used for displaying training tasks, progress and electroencephalogram signal related information, the auditory feedback unit (42) is used for playing voice prompts and training result feedback, and the tactile feedback unit (43) is used for providing tactile prompts through vibration.

6. The brain impairment function training device based on brain-computer interface according to claim 1, characterized in that: The evaluation module (5) comprises a recovery index analysis unit (51) and an evaluation report generation unit (52); the input end of the recovery index analysis unit (51) is electrically connected with the output end of the signal processing module (2) and the output end of the training control module (3), and the output end of the recovery index analysis unit (51) is electrically connected with the input end of the training control module (3) and the input end of the evaluation report generation unit (52); the recovery index analysis unit (51) obtains a recovery index by weighted calculation of a training accuracy rate and an electroencephalogram signal characteristic change rate, and is used for analyzing a functional recovery state of a user; and the evaluation report generation unit (52) is used for generating a recovery evaluation report according to the recovery index.

7. The brain impairment function training device based on brain-computer interface according to claim 1, characterized in that: The user interaction module (6) is further comprised; the output end of the user interaction module (6) is electrically connected with the input end of the training control module (3), and the input end of the user interaction module (6) is electrically connected with the output end of the evaluation module (5); the user interaction module (6) comprises a touch panel (61) and a data interface (62), and is used for inputting personal information, training preferences and viewing evaluation reports by a user, and is also used for inputting training parameters and adjusting evaluation standards by medical staff. 8.The brain impairment function training device based on brain-computer interface according to claim 1, characterized in that: The support frame adopts a head-mounted structure, comprises a main frame and an elastic adjusting belt; the main frame is used for mounting the electroencephalogram acquisition module (1), and the elastic adjusting belt is arranged on both sides of the main frame and is used for adapting to head circumferences of different users and ensuring stable contact between the electroencephalogram acquisition module (1) and the scalp.

9. The brain impairment function training device based on brain-computer interface according to claim 2, characterized in that: The flexible electrode array (11) is made of a medical silica gel material, and an electrically conductive gel layer is arranged on the surface of the electrode array; the electrically conductive gel layer is used for reducing contact impedance between the electrode and the scalp and improving stability of electroencephalogram signal acquisition.

Images