Cerebral stroke rehabilitation method based on fusion of motor imagery brain-computer interface and transcranial direct current stimulation

Abstract

The invention belongs to the related technical field of neural rehabilitation engineering, and particularly relates to a cerebral apoplexy rehabilitation method based on fusion of a motor imagery brain-computer interface and transcranial direct current stimulation, which comprises the following steps: acquiring a corresponding artifact template based on transcranial direct current stimulation parameters; carrying out source separation on the collected electroencephalogram test times by adopting a joint blind source separation algorithm to obtain clean electroencephalogram signal components; wherein the construction mode of the artifact template library comprises the following steps: extracting electroencephalogram trial artifact samples of multiple stimulation periods, and generating a transcranial direct current stimulation artifact template through an average analysis or principal component analysis mode to obtain the artifact template library; performing motion intention decoding on the clean electroencephalogram signal component to obtain a motion intention of the patient; inputting the motion intention, pre-collected patient body state evaluation information and patient feedback comfort into a fuzzy system algorithm to obtain stimulation parameters of to-be-applied transcranial direct current stimulation, and realizing active rehabilitation through the transcranial direct current stimulation. According to the invention, algorithm-driven closed-loop rehabilitation regulation and control can be realized.

Description

Technical Field

[0001] The present invention belongs to the technical field related to neurorehabilitation engineering, and more specifically, relates to a stroke rehabilitation method based on the fusion of motor imagery brain-computer interface and transcranial direct current stimulation. Background Art

[0002] Stroke (also known as stroke) is a common cerebrovascular disease characterized by sudden disruption of cerebral blood flow, leading to brain tissue damage and consequently limb motor dysfunction, severely impacting patients' ability to perform daily activities and quality of life. Currently, stroke rehabilitation relies primarily on traditional approaches such as physical therapy, drug intervention, and neuromodulation. However, these approaches still suffer from long rehabilitation cycles, poor adaptability among individual patients, poor compliance, and limited rehabilitation outcomes, making them difficult to meet the actual needs for efficient and personalized rehabilitation.

[0003] In recent years, the development of brain-computer interface (BCI) technology has provided new solutions for stroke rehabilitation. Among them, motor imagery (MI), a psychological process that replays motor behavior in the brain through subjective intention without the need for actual limb movement, does not involve actual muscle movement output, but can activate brain areas related to actual movement (such as the primary motor cortex and supplementary motor area), and has significant potential in inducing brain neuroplasticity. MI-based BCI systems can be used to identify user intentions and link neurorehabilitation equipment to achieve functional neural reconstruction, and have good application prospects.

[0004] As a non-invasive neuromodulation technology, transcranial direct current stimulation (tDCS) can effectively regulate the excitability of the cerebral cortex and promote motor function recovery by applying a weak constant current to specific parts of the scalp, compared to traditional physical therapy or single MI rehabilitation methods. In addition, tDCS equipment is portable and low-cost, suitable for home rehabilitation training, and can improve patients' long-term compliance and rehabilitation efficiency. Common tDCS stimulation modes include:

[0005] 1) Anodal tDCS: Enhances cortical excitability in damaged brain areas;

[0006] 2) Cathodal tDCS: inhibits the activity of the contralateral cortex and reduces its inhibition on the affected side;

[0007] 3) Bilateral tDCS: Synchronously regulates both cortices to achieve functional balance.

[0008] According to an analysis of 67 randomized controlled trials, tDCS may be effective in improving daily living abilities, but has weaker effects on upper limb function, lower limb function, and cognitive ability. tDCS may be effective in specific patient groups (such as chronic patients or mild to moderate injuries), but its efficacy in overall stroke patients remains controversial. Summary of the Invention

[0009] In response to the above-mentioned defects or improvement needs of the existing technology, the present invention provides a stroke rehabilitation method based on the fusion of motor imagery brain-computer interface and transcranial direct current stimulation. Its purpose is to propose an algorithm-driven closed-loop rehabilitation control based on personalized tDCS stimulation parameter regulation to achieve an effective patient-active rehabilitation method.

[0010] To achieve the above objectives, according to one aspect of the present invention, a stroke rehabilitation method based on the fusion of motor imagery brain-computer interface and transcranial direct current stimulation is provided, comprising:

[0011] Collecting EEG trials from the patient; obtaining corresponding artifact templates from an artifact template library based on known transcranial direct current stimulation parameters; performing source separation on the EEG trials using a joint blind source separation algorithm to obtain clean EEG signal components; wherein the artifact template library is constructed by extracting artifact samples from EEG trials over multiple stimulation cycles, generating transcranial direct current stimulation artifact templates through averaging analysis or principal component analysis, and obtaining the artifact template library;

[0012] The clean EEG signal components are decoded for movement intention to obtain the patient's movement intention, including the movement start time, target brain area, and intention confidence; the movement intention, as well as pre-collected patient physical state assessment information and patient feedback comfort, are input into a fuzzy system algorithm to obtain the stimulation parameters of the transcranial direct current stimulation to be applied, wherein the rule base of the fuzzy system algorithm is constructed in the following manner: based on the rehabilitation treatment stage reflected in the patient's physical state assessment information and the confidence level of the patient's active rehabilitation intention obtained from the movement intention, the stimulation intensity parameters are set accordingly; based on the patient's motor function, the stimulation polarity parameters are set accordingly; and based on the patient's feedback comfort, the stimulation intensity parameters are set accordingly;

[0013] Based on the patient's movement intention, transcranial direct current stimulation with specific intensity and polarity parameters determined by a fuzzy system algorithm is applied to the target brain area at the start time of the movement to achieve active rehabilitation.

[0014] Furthermore, the average analysis method is:

[0015] The artifact portions of the same polarity and current among multiple EEG trial artifact samples of each specific preset transcranial direct current stimulation parameter condition are time-aligned and averaged, which is used as the artifact template of the specific preset transcranial direct current stimulation parameter condition.

[0016] Furthermore, the principal component analysis method is:

[0017] Principal component analysis was performed on multiple EEG artifact samples for each specific preset transcranial direct current stimulation parameter condition, and the first n principal components were retained. The dimensionality was reduced to a subspace to obtain an artifact template for the specific preset transcranial direct current stimulation parameter condition, where the value of n satisfied that the data after dimensionality reduction could retain 95% of the variance.

[0018] Furthermore, the method of performing source separation on the EEG trials is:

[0019] Time-aligning the acquired artifact template with the current EEG trial to identify the artifact portion of the current EEG trial;

[0020] Template subtraction, artifact subspace projection or linear regression is used to subtract the artifact part from the current EEG trial, and the EEG trial after artifact elimination is retained as the clean EEG signal component of the current EEG trial. If the average analysis method is used to construct the artifact template, template subtraction or linear regression is used for artifact elimination. If the principal component analysis method is used to construct the artifact template, artifact subspace projection is used for artifact elimination.

[0021] Furthermore, before decoding, the method further includes: applying an adaptive filter to the clean EEG signal component to obtain a final cleaned EEG signal for subsequent decoding.

[0022] Furthermore, the ERD strength is used to characterize the intent confidence.

[0023] According to another aspect of the present invention, an electronic device is provided, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0024] According to another aspect of the present invention, a computer-readable storage medium is provided, which includes a stored computer program, wherein when the computer program is executed by a processor, the device where the storage medium is located is controlled to execute the steps of the above-mentioned method.

[0025] According to another aspect of the present invention, a computer program product is provided, comprising a computer program or instructions, which implement the steps of the method described above when executed by a processor.

[0026] In general, the above technical solutions conceived by the present invention have the following beneficial effects compared with the prior art:

[0027] 1. The present invention provides a stroke rehabilitation method based on the fusion of motor imagery brain-computer interface and transcranial direct current stimulation. First, an artifact template is constructed based on transcranial electrical stimulation with controllable parameters, which can accurately reflect the temporal and spatial variation characteristics of the artifact under different stimulation conditions, thereby achieving effective modeling and accurate stripping of interference signals. Secondly, based on the decoding results of motor imagery EEG signals, the decoding results of movement intentions and pre-collected patient physical state assessment information and patient feedback comfort are input into the fuzzy system algorithm to obtain the stimulation parameters of the transcranial direct current stimulation to be applied, and personalized and automated transcranial direct current stimulation parameter selection is achieved, thereby improving the safety and effectiveness of the stimulation effect, and cyclic stimulation is performed based on the stimulation parameters output by the fuzzy system algorithm to construct a closed-loop intelligent rehabilitation model. This method integrates brain-computer interface, artificial intelligence and neural regulation technology, provides a new solution for stroke rehabilitation medicine, realizes intelligent, safe, non-invasive and low-cost stroke rehabilitation solutions, and provides patients with more accurate and personalized intelligent rehabilitation means.

[0028] 2. The present invention also proposes applying an adaptive filter to the clean EEG signal component before decoding to obtain a final cleaned EEG signal for subsequent decoding, thereby improving the effectiveness of EEG signal decoding. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 A flowchart of a stroke rehabilitation method based on the fusion of motor imagery brain-computer interface and transcranial direct current stimulation provided by the method of this embodiment;

[0030] Figure 2 This is a schematic diagram of the complete process of the stroke rehabilitation system based on motor imagery brain-computer interface and transcranial electrical stimulation provided by the method of this embodiment. DETAILED DESCRIPTION

[0031] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0032] Example 1

[0033] A stroke rehabilitation method based on the fusion of motor imagery brain-computer interface and transcranial direct current stimulation, such as Figure 1 Shown, including:

[0034] Collecting the patient's EEG trials; obtaining corresponding artifact templates from an artifact template library based on known transcranial direct current stimulation parameters; performing source separation on the EEG trials using a joint blind source separation algorithm to obtain clean EEG signal components; wherein the artifact template library is constructed by extracting artifact samples from EEG trials over multiple stimulation cycles, generating transcranial direct current stimulation artifact templates through averaging analysis or principal component analysis, and obtaining the artifact template library;

[0035] The clean EEG signal components are decoded for movement intention (for example, using a feature extraction and classification model based on common spatial patterns and support vector machines) to obtain the patient's movement intention, including movement start time, target brain area, and intention confidence; the movement intention and pre-collected patient physical state assessment information and feedback comfort are input into a fuzzy system algorithm to obtain stimulation parameters for the transcranial direct current stimulation to be applied, wherein the rule base of the fuzzy system algorithm is constructed in the following manner: based on the rehabilitation treatment stage reflected in the patient's physical state assessment information and the confidence of the patient's active rehabilitation intention obtained from the movement intention, the stimulation intensity parameters are set accordingly; based on the patient's motor function, the stimulation polarity parameters are set accordingly; based on the patient's feedback comfort, the stimulation intensity parameters are set accordingly;

[0036] Based on the movement intention, transcranial direct current stimulation with specific intensity and polarity parameters determined by the fuzzy system algorithm is applied to the target brain area at the start time of the movement to achieve active rehabilitation.

[0037] Considering the need to optimize personalized tDCS stimulation parameters in the future, automated intelligent systems will be used as a supplement, while also incorporating traditional treatment methods, such as electromyography stimulation using neurorehabilitation diagnostic and therapeutic devices or arm movement driven by a robotic arm, to provide multiple auxiliary treatment methods. Domestic rehabilitation equipment is still primarily based on single-function devices such as pure tDCS or exoskeletons. Mature products with algorithm-driven closed-loop control have yet to be developed, leaving a significant unmet clinical need.

[0038] During the stroke rehabilitation process, in order to achieve closed-loop control of motor imagery EEG decoding and transcranial electrical stimulation, EEG signal acquisition and tDCS stimulation control need to be performed simultaneously. However, the weak DC current applied by tDCS can easily introduce artifact interference, affecting the quality of EEG signals and the accuracy of MI decoding. Existing systems mostly use time-sharing operation to avoid interference, which cannot meet high real-time requirements. The method of this embodiment constructs an artifact template based on transcranial electrical stimulation with controllable parameters, which can accurately reflect the temporal and spatial variation characteristics of artifacts under different stimulation conditions, thereby achieving effective modeling and accurate stripping of interference signals. Compared with traditional bandpass filtering, the artifact template can retain low-frequency neural signals and more effectively remove structured artifacts. Compared with adaptive filtering, it does not rely on the setting of the reference channel and has stronger individual applicability.

[0039] Existing tDCS stimulation mostly uses fixed parameters and lacks the ability to dynamically adjust according to the individual's brain activity state, making it difficult to achieve precise and personalized cortical regulation. At the same time, how to implement a closed-loop stimulation control mechanism without interrupting or interfering with EEG acquisition and decoding is a major technical bottleneck affecting the practical application of rehabilitation systems. The method of this embodiment proposes inputting movement intentions and pre-collected patient physical state assessment information and feedback comfort into a fuzzy system algorithm to obtain the stimulation parameters for the transcranial direct current stimulation to be applied. The rule base of the fuzzy system algorithm is constructed in the following way: based on the rehabilitation treatment stage reflected in the patient's physical state assessment information and the confidence level of the patient's active rehabilitation intention obtained from the movement intention, the stimulation intensity parameters are set accordingly; based on the patient's motor function, the stimulation polarity parameters are set accordingly; and based on the patient's feedback comfort, the stimulation intensity parameters are set accordingly. In other words, this method uses a fuzzy reasoning module to perform personalized adaptive control of parameters such as stimulation intensity and stimulation polarity, thereby improving the safety and effectiveness of the stimulation effect. The stimulation time and stimulation position are determined by the EEG decoding of the movement intention.

[0040] Therefore, this embodiment addresses the key bottleneck in the application of motor imagery and transcranial electrical stimulation in stroke rehabilitation, and proposes a closed-loop rehabilitation method that integrates precise MI decoding, adaptive tDCS regulation, and anti-interference EEG acquisition. It aims to improve the intelligence level and real-time response capability in the rehabilitation process, break through the limitations of traditional rehabilitation programs in decoding accuracy, stimulation safety, and system integration, and thus realize a new model of efficient, personalized, closed-loop controlled stroke rehabilitation treatment.

[0041] In addition, the intelligent stroke rehabilitation BCI system that combines MI and tDCS can combine MI precise decoding and tDCS precise parameter specification treatment strategies to build a complete intelligent stroke rehabilitation BCI system. Furthermore, after detecting MI, in addition to implementing tDCS, traditional muscle electrical stimulation or machine-driven hand movement can also be applied to realize a closed-loop stroke rehabilitation treatment intelligent system. This tDCS technology based on the MI precise decoding algorithm can fill the technical gap in the precise regulation of tDCS-based stroke rehabilitation, realize a new application paradigm of BCI in the field of medical rehabilitation, and provide stroke patients with more efficient and safe intelligent rehabilitation solutions, thereby improving patients' long-term prognosis and quality of life.

[0042] Give an example, the system composition and complete process are as follows Figure 2As shown. The patient actively performs motor imagery training, and the system uses an EEG cap to collect EEG signals in real time. It then accurately decodes the EEG patterns based on algorithms such as transfer learning to identify motor imagery states, including key patterns such as the start time of movement, the brain area corresponding to the target limb, the confidence level of task execution, and the intensity of event-related desynchronization (ERD). The decoded information is then transmitted to the transcranial direct current stimulator, which is further combined with information such as the treatment stage, doctor experience, and patient physical condition assessment to intelligently and comprehensively decide on stimulation parameters, including stimulation intensity and polarity, using algorithms such as fuzzy systems. The system backend uses a neurorehabilitation diagnosis and treatment platform to manage data and update models, and integrates other auxiliary treatment modules, such as the currently more commonly used myoelectric stimulation or robotic arm-assisted movement, to achieve multimodal collaborative intervention, thereby building a complete closed-loop intelligent stroke rehabilitation system.

[0043] To address the artifactual interference caused by transcranial direct current stimulation (DCS) during motor imagery EEG acquisition, this embodiment proposes a multi-level decoupled anti-interference signal processing method. Compared to existing solutions that rely solely on time staggering (i.e., collecting EEG signals only during the intervals between DCS stimulations), this embodiment's method achieves real-time suppression of DCS discharge artifacts and steady-state EEG signal quality assurance by building specific technical mechanisms at the algorithm processing and hardware system levels, ensuring the continuity and reliability of the brain-computer interface system.

[0044] Preferably, the above-mentioned average analysis method is: time-align the same polarity and same current artifact parts of multiple EEG trial artifact samples for each specific preset transcranial direct current stimulation parameter condition and then take the average as the artifact template for the specific preset transcranial direct current stimulation parameter condition.

[0045] Preferably, the principal component analysis method is: perform principal component analysis on multiple EEG trial artifact samples for each specific preset transcranial direct current stimulation parameter condition, retain the first n principal components, reduce the dimension to the subspace, and obtain the artifact template of the specific preset transcranial direct current stimulation parameter condition, wherein the value of n satisfies: the data after dimensionality reduction can retain 95% of the variance.

[0046] As a preferred method, the source separation of EEG trials is performed as follows:

[0047] Time-aligning the acquired artifact template with the current EEG trial to identify the artifact portion of the current EEG trial;

[0048] Template subtraction, artifact subspace projection or linear regression is used to subtract the artifact part from the current EEG trial, and the EEG trial after artifact elimination is retained as the clean EEG signal component of the current EEG trial. If the average analysis method is used to construct the artifact template, template subtraction or linear regression is used for artifact elimination. If the principal component analysis method is used to construct the artifact template, artifact subspace projection is used for artifact elimination.

[0049] Preferably, before decoding, the method further comprises: applying an adaptive filter to the clean EEG signal component to obtain a final cleaned EEG signal for subsequent decoding.

[0050] This embodiment proposes an interference suppression algorithm that integrates artifact modeling, dynamic template matching, and adaptive filtering. As an implementation example, the algorithm specifically includes the following steps:

[0051] Artifact template pre-construction: During the system initialization phase, the EEG responses during several rounds of tDCS stimulation are recorded. After aligning the stimulation trigger events, the spatial distribution (electrode channel contribution pattern), spectral characteristics (including power frequency coupling and low-frequency drift), and timing waveform (typical time window of stimulation response) related to the artifact are extracted to construct a multi-channel tDCS artifact feature library. The principal components of the artifacts are extracted through principal component analysis to form an artifact feature vector group.

[0052] Intensity mapping model fitting: After collecting multiple rounds of stimulation data, nonlinear regression (Gaussian kernel regression) is used to fit the functional relationship between stimulation intensity and artifact amplitude, so that the artifact interference amplitude can be subsequently predicted based on the real-time stimulation current.

[0053] Real-time template matching and interference extraction: During system operation, the corresponding artifact template is called according to the current tDCS parameters, and the interference component most similar to the artifact template is extracted from the currently collected EEG data using the least mean square matching or correlation coefficient matching method.

[0054] Joint blind source separation processing: A joint blind source separation module based on independent vector analysis is introduced, and the artifact component is used as the constraint channel to guide the separation process to explicitly remove the interference source and realize clean signal reconstruction in multi-channel EEG.

[0055] Adaptive filter correction: A time-varying filter is applied to the reconstructed signal, and the filter coefficient is dynamically adjusted based on the real-time feedback error to adapt to the nonlinear disturbance caused by different current intensities and scalp conductivity characteristics, thereby achieving fine signal repair.

[0056] Compared to directly using bandpass filtering to remove interference, this solution can further process non-stationary interference, retain low-frequency and high-frequency information, and account for spatial crosstalk. It combines real-time and adaptive capabilities, preserves characteristic structures, and improves decoding robustness. Through this design, the present invention can effectively suppress electrical stimulation artifacts when tDCS stimulation and EEG acquisition are performed simultaneously, achieving continuous, stable, and high-quality acquisition of EEG signals, providing a signal quality assurance foundation for the implementation of closed-loop EEG stimulation systems.

[0057] Preferably, ERD strength is used to characterize the intention confidence.

[0058] Regarding personalized tDCS stimulation parameter control. To address the issues of tDCS parameter setting being highly dependent on physician experience, exhibiting significant individual variability, and lacking intelligent control, this embodiment constructs a personalized stimulation parameter decision module based on a fuzzy system. This module integrates EEG decoding pattern information, patient clinical scores, and treatment stage status to achieve dynamic adjustment and safety control of the stimulation regimen.

[0059] As an example, the fuzzy rule base is constructed based on neurorehabilitation knowledge, clinical experience, and information reflected by EEG signals, including:

[0060] If the ERD intensity is “high” and the patient is in the “mid-to-late recovery stage”, the stimulation intensity is set to “low” to avoid excessive cortical activation causing neural fatigue or discomfort;

[0061] If the ERD intensity is “low” and the patient is in the “early recovery stage”, the stimulation intensity is set to “medium-high” to enhance cortical excitability and promote neuroplasticity;

[0062] If the muscle tension on the affected side is "high", choose "cathode stimulation" of the motor cortex on the healthy side to reduce the inhibitory effect of the healthy cortex on the affected side;

[0063] If the Fugl-Meyer score shows that the motor function is "low" and the patient is in the "chronic stage", "bilateral stimulation" is used to balance the excitation state of both hemispheres;

[0064] If the patient's subjective feedback is "discomfort" or the skin resistance is abnormally increased, the stimulation intensity is immediately reduced to "low" and the stimulation position is adjusted to the suboptimal area.

[0065] Fuzzy inference mechanism implementation: Input features (including ERD intensity, Fugl-Meyer score, and muscle tension status) are fuzzified (Gaussian membership function) and then fed into the Mamdani fuzzy inference system. This system then activates rules, calculates weights, and integrates outputs. Finally, the center of gravity method is used to defuzzify the system to obtain specific stimulation parameters, including intensity, start and end times, and location. This module supports knowledge-data collaboration, balancing system adaptability and interpretability.

[0066] In summary, this paper proposes a closed-loop stroke rehabilitation method based on the fusion of a motor imagery brain-computer interface and transcranial direct current stimulation (tDCS). By integrating EEG signal decoding, adaptive stimulation parameter control, and the ability to combine with multimodal intervention devices into a unified platform, it achieves an intelligent rehabilitation closed-loop control process driven by active intention. The system uses the patient's motor imagery as the active input signal source and utilizes multi-channel EEG acquisition equipment to acquire movement-related EEG signals in real time. Transfer learning and open-set recognition mechanisms improve adaptability to individual differences and class heterogeneity, accurately identifying the target limb, task status, and movement initiation intention. To address artifact interference during tDCS stimulation, the method introduces a multi-level signal decoupling mechanism, including artifact template modeling, blind source separation, and adaptive filtering, to ensure high-quality EEG signal acquisition and real-time availability. Based on the decoding results and information such as the patient's rehabilitation stage and clinical assessment, the system uses a fuzzy inference module to perform personalized adaptive control of parameters such as stimulation intensity and polarity, improving the safety and effectiveness of the stimulation effect. Ultimately, the system integrates external auxiliary treatment modules, such as electromyography (EMT) stimulation or robotic arm actuators, to achieve coordinated action feedback driven by brain intent, establishing a closed-loop rehabilitation pathway consisting of neural activation, motor feedback, and stimulation regulation. This invention provides a highly efficient, intelligent, and scalable rehabilitation system integration solution for clinical practice, demonstrating significant technological advancement and widespread application value.

[0067] Example 2

[0068] The present application also relates to an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0069] The electronic device may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server. The processor may be a central processing unit (CPU), or other general-purpose processors, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The memory may be used to store computer programs and / or modules, and the processor may perform various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory, and calling the data stored in the memory.

[0070] The relevant technical solutions are the same as above and will not be repeated here.

[0071] Example 3

[0072] The present application also relates to a computer-readable storage medium having a computer program stored thereon, which implements the steps of the above method when the computer program is executed by a processor.

[0073] Specifically, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0074] The relevant technical solutions are the same as above and will not be repeated here.

[0075] Example 4

[0076] The present invention provides a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps of the method of the above-described embodiment of the present invention.

[0077] The relevant technical solutions are the same as above and will not be repeated here.

[0078] It will be easily understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A stroke rehabilitation method based on the fusion of motor imagery brain-computer interface and transcranial direct current stimulation, characterized in that: include: Collect the patient's EEG trials; based on the known transcranial direct current stimulation parameters, obtain the corresponding artifact template from the artifact template library; A joint blind source separation algorithm is used to perform source separation on the EEG trial to obtain clean EEG signal components; wherein the artifact template library is constructed by extracting EEG trial artifact samples from multiple stimulation cycles, generating transcranial direct current stimulation artifact templates through average analysis or principal component analysis, and obtaining the artifact template library; The clean EEG signal components are decoded for movement intention to obtain the patient's movement intention, including the movement start time, target brain area, and intention confidence; the movement intention, as well as pre-collected patient physical state assessment information and patient feedback comfort, are input into a fuzzy system algorithm to obtain stimulation parameters for the transcranial direct current stimulation to be applied, wherein the rule base of the fuzzy system algorithm is obtained in the following manner: based on the rehabilitation treatment stage reflected in the patient's physical state assessment information and the confidence level of the patient's active rehabilitation intention obtained from the movement intention, the stimulation intensity parameters are set accordingly; based on the patient's motor function, the stimulation polarity parameters are set accordingly; and based on the patient's feedback comfort, the stimulation intensity parameters are set accordingly; Based on the patient's movement intention, transcranial direct current stimulation with specific intensity and polarity parameters determined by a fuzzy system algorithm is applied to the target brain area at the start time of the movement to achieve active rehabilitation.

2. The stroke rehabilitation method according to claim 1, wherein: The method of the average analysis is: The artifact portions of the same polarity and current among multiple EEG trial artifact samples of each specific preset transcranial direct current stimulation parameter condition are time-aligned and averaged, which is used as the artifact template of the specific preset transcranial direct current stimulation parameter condition.

3. The stroke rehabilitation method according to claim 1, wherein: The principal component analysis method is: Principal component analysis was performed on multiple EEG artifact samples for each specific preset transcranial direct current stimulation parameter condition, and the first n principal components were retained. The dimensionality was reduced to a subspace to obtain an artifact template for the specific preset transcranial direct current stimulation parameter condition, where the value of n satisfied that the data after dimensionality reduction could retain 95% of the variance.

4. The stroke rehabilitation method according to claim 1, wherein: The method of performing source separation on the EEG trial is: Time-aligning the acquired artifact template with the current EEG trial to identify the artifact portion of the current EEG trial; Template subtraction, artifact subspace projection or linear regression is used to subtract the artifact part from the current EEG trial, and the EEG trial after artifact elimination is retained as the clean EEG signal component of the current EEG trial. If the average analysis method is used to construct the artifact template, template subtraction or linear regression is used for artifact elimination. If the principal component analysis method is used to construct the artifact template, artifact subspace projection is used for artifact elimination.

5. The stroke rehabilitation method according to claim 1, wherein: Before decoding, the method further includes: applying an adaptive filter to the clean EEG signal component to obtain a final cleaned EEG signal for subsequent decoding.

6. The stroke rehabilitation method according to claim 1, wherein: The ERD strength is used to represent the confidence of the intention.

7. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed by a processor, the device where the storage medium is located is controlled to perform the steps of the method according to any one of claims 1 to 6.

9. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images