OpenViBE brain-computer interface decoding control method and system driven by efficient channel selection algorithm

Abstract

The invention relates to the technical field of brain-computer interfaces, and discloses an OpenViBE brain-computer interface decoding control method and system driven by an efficient channel selection algorithm. The method specifically comprises the steps that based on pre-collected electroencephalogram data, the comprehensive importance score of each channel is calculated for each task, a channel-task importance matrix is constructed, then an optimal channel subset is selected from the channel-task importance matrix, an electroencephalogram signal processing model is optimized according to the optimal channel subset, and the electroencephalogram signal processing accuracy is improved. And deploying the optimized electroencephalogram signal processing model to an OpenViBE platform, and carrying out electroencephalogram real-time analysis and processing. The calculation amount can be remarkably reduced while the accuracy rate is guaranteed, the requirement for hardware resources is reduced, and the real-time performance of the model on portable equipment is improved. In addition, a corresponding channel selection strategy can be flexibly selected according to a specific application scene, and an optimal channel subset is selected from the channel-task importance matrix according to the selected channel selection strategy, so that the universality of the model is remarkably improved.

Description

Technical Field

[0001] This application relates to the technical field of brain-computer interfaces, and particularly to an OpenViBE brain-computer interface decoding control method and system driven by an efficient channel selection algorithm. Background Art

[0002] Brain-computer interface (BMI) technology provides a new approach for the movement rehabilitation of disabled people, especially for controlling devices such as prosthetics or wheelchairs using motor imagery (MI). Compared with non-invasive electroencephalogram (EEG), implantable brain-computer interfaces are more suitable for long-term stable rehabilitation training due to their higher signal quality and more convenient usage. Convolutional neural networks (CNNs) have shown great potential in MI analysis, and their ability to automatically learn features avoids cumbersome manual feature engineering. However, high-performance CNN models usually need to run on powerful remote servers with high computing capabilities, which can lead to data transmission delays, affect system availability, and pose data privacy and security risks.

[0003] To solve these problems, a feasible solution is to perform local data processing on a portable computing terminal (such as a laptop). This solution takes into account both computing power and portability, facilitating doctors to monitor and adjust parameters in real time, and is also easier to maintain and upgrade than an embedded system. However, this solution also faces huge challenges: how to run a high-performance CNN model on the limited computing resources of a portable computing terminal and ensure real-time performance and decoding accuracy? Existing brain-computer interface systems based on platforms such as OpenViBE often rely on complex algorithms and a large amount of channel data, which are difficult to achieve real-time processing on portable devices. The key contribution of lightweight models such as EEGNet in solving the conflict between real-time performance and accuracy lies in the improvement of their parameters and computational efficiency. By adopting techniques such as depthwise separable convolution and designing a compact network architecture, the number of model parameters and the amount of computation are significantly reduced, enabling it to achieve fast inference on resource-constrained hardware platforms and meet the requirements of real-time BCI applications. However, models such as EEGNet do not explicitly consider the problem of channel redundancy and directly process all input channels. This results in low computational efficiency of the model, being prone to overfitting to noise and redundant information, thereby interfering with feature learning, reducing generalization ability, and making the model results difficult to interpret. Summary of the Invention

[0004] This application provides an OpenViBE brain-computer interface decoding control method driven by an efficient channel selection algorithm to solve the problem in the prior art that when lightweight models run on resource-constrained hardware platforms, they do not explicitly consider the problem of channel redundancy, resulting in low computational efficiency, being prone to overfitting to noise and redundant information, thereby interfering with feature learning, reducing generalization ability, and making the model results difficult to interpret.

[0005] Correspondingly, the present application also provides an OpenViBE brain-computer interface decoding control system driven by an efficient channel selection algorithm to ensure the implementation and application of the above method.

[0006] To solve the above technical problems, the present application discloses an OpenViBE brain-computer interface decoding control method driven by an efficient channel selection algorithm, and the method includes: Based on pre-collected electroencephalogram (EEG) data, calculate the comprehensive importance scores of each channel for each task respectively, and construct a channel-task importance matrix; Determine a channel selection strategy, and select an optimal channel subset from the channel-task importance matrix according to the selected channel selection strategy; Optimize the EEG signal processing model according to the optimal channel subset; Deploy the optimized EEG signal processing model to the OpenViBE platform for real-time EEG analysis and processing.

[0007] Preferably, based on pre-collected EEG data, calculating the comprehensive importance scores of each channel for each task respectively, and constructing a channel-task importance matrix includes: Use all-channel data of the EEG data to train an EEG signal processing model to obtain the feature weights of each layer; For each task, calculate the weight gradient information according to the feature weights of each layer, obtain the importance score of each channel, and construct a channel-task importance matrix.

[0008] Preferably, for each task, calculating the weight gradient information according to the feature weights of each layer, obtaining the importance score of each channel, and constructing a channel-task importance matrix includes: For the current task, use the validation set data to calculate the loss function, and obtain the gradient of the feature weights of each layer with respect to the loss function through backpropagation; Based on the gradients and feature weights of each layer, calculate the channel importance score of each channel in each layer; Calculate the comprehensive importance score of each channel according to the channel importance scores of each channel in each layer; Construct a channel-task importance matrix according to the comprehensive importance scores of each channel obtained under each task.

[0009] Preferably, the channel selection strategy includes: A global strategy, select one or more channels in all tasks in the order of decreasing comprehensive importance scores; A task-specific strategy, select one or more channels for a specific task in the order of decreasing comprehensive importance scores; The mixed strategy is to select one or more channels shared by multiple tasks and one or more channels for a specific task in descending order of the comprehensive importance score.

[0010] Preferably, a channel selection strategy is determined, and an optimal channel subset is selected from the channel-task importance matrix according to the selected channel selection strategy, including: Determine the channel selection strategy from the global strategy, task-specific strategy, and mixed strategy; Based on the selected channel selection strategy, determine the channel selection scheme; Select and obtain the optimal channel subset from the channel-task importance matrix according to the channel selection scheme.

[0011] Preferably, the channel selection scheme is any one of the following: Scheme 1: Predetermine the target number of channels K, and select the K channels with the highest importance scores according to the selected channel selection strategy to generate the optimal channel subset; Scheme 2: Set an importance threshold, and select all channels with importance scores higher than the importance threshold according to the selected channel selection strategy to generate the optimal channel subset; Scheme 3: Determine the constraint conditions, and select the optimal channel combination according to the selected channel selection strategy on the premise of meeting the constraint conditions to generate the optimal channel subset.

[0012] Preferably, optimizing the electroencephalogram (EEG) signal processing model according to the optimal channel subset includes: In the EEG data, select the data related to the optimal channel subset to train the EEG signal processing model.

[0013] Preferably, before calculating the comprehensive importance score of each channel for each task based on the pre-collected EEG data and constructing the channel-task importance matrix, the method further includes: Filter the EEG data and use a preset method to remove eye movement and electromyogram artifacts in the EEG data.

[0014] Preferably, before deploying the optimized EEG signal processing model to the OpenViBE platform for real-time EEG analysis and processing, the method further includes: Configure the relevant parameters of the OpenViBE platform, including the sampling rate, channel data, and data format.

[0015] This application also discloses an OpenViBE brain-computer interface decoding control system driven by an efficient channel selection algorithm, and the system includes: A channel importance calculation module, configured to calculate the comprehensive importance score of each channel for each task based on the pre-collected EEG data and construct a channel-task importance matrix; A channel selection module, which is used to determine a channel selection strategy and select an optimal channel subset from a channel-task importance matrix according to the selected channel selection strategy; A model optimization module, which is used to optimize an electroencephalogram (EEG) signal processing model according to the optimal channel subset; A model deployment module, which is used to deploy the optimized EEG signal processing model to the OpenViBE platform for real-time EEG analysis and processing.

[0016] In this application, based on pre-collected EEG data, the comprehensive importance scores of each channel are calculated for each task respectively to construct a channel-task importance matrix. Then, an optimal channel subset is selected from the channel-task importance matrix, and the EEG signal processing model is optimized according to the optimal channel subset. The optimized EEG signal processing model is deployed to the OpenViBE platform for real-time EEG analysis and processing. It can significantly reduce the computational load while ensuring the accuracy, reduce the demand for hardware resources, and improve the real-time performance of the model on portable devices. In addition, the corresponding channel selection strategy can be flexibly selected according to the specific application scenario, and an optimal channel subset is selected from the channel-task importance matrix according to the selected channel selection strategy, which significantly improves the versatility of the model.

[0017] Additional aspects and advantages of the present application will be given in the following description section, which will become apparent from the following description or be understood through the practice of the present application. Description of the Drawings

[0018] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description of the embodiments in conjunction with the drawings, where: Figure 1 is a flowchart of an OpenViBE brain-computer interface decoding control method driven by an efficient channel selection algorithm provided by an embodiment of the present application; Figure 2 is an overall flowchart of decoding control using an OpenViBE brain-computer interface provided by an embodiment of the present application; Figure 3 is a flowchart of real-time model optimization and deployment provided by an embodiment of the present application; Figure 4 is a schematic structural diagram of an OpenViBE brain-computer interface decoding control system driven by an efficient channel selection algorithm provided by an embodiment of the present application; Figure 5 is a schematic structural diagram of a complete system for decoding control of an OpenViBE brain-computer interface provided by an embodiment of the present application. Detailed Embodiments

[0019] Embodiments of the present application will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, where like or similar reference numerals denote like or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and are only used to explain the present application and should not be construed as a limitation of the present application.

[0020] Those skilled in the art of the present technology can understand that, unless specifically stated otherwise, the singular forms "a", "an", "the", and "said" used herein may also include the plural forms. It should be further understood that the term "comprising" used in the specification of the present application means the presence of features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or their combinations. It should be understood that when we say that an element is "connected" or "coupled" to another element, it can be directly connected or coupled to other elements, or there may also be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The phrase "and / or" used herein includes all or any unit and all combinations of one or more of the associated listed items.

[0021] Those skilled in the art can understand that, unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as the general understanding of those of ordinary skill in the art to which the present invention belongs. It should also be understood that terms such as those defined in a general dictionary should be understood to have a meaning consistent with the meaning in the context of the prior art, and will not be interpreted with an idealized or overly formal meaning unless specifically defined as here.

[0022] The solution provided by the embodiments of the present application can be executed by any electronic device, such as a terminal device or a server. Among them, the server can be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. The terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, etc., but is not limited thereto. The terminal and the server can be directly or indirectly connected through wired or wireless communication methods, and the present application does not limit this. Regarding the technical problems existing in the prior art, the efficient channel selection algorithm-driven OpenViBE brain-computer interface decoding control method and system provided by the present application aim to solve at least one of the technical problems in the prior art. The following uses specific embodiments to elaborate in detail on the technical solution of the present application and how the technical solution of the present application solves the above technical problems. These several specific embodiments below can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0023] The embodiment of the present application provides an OpenViBE brain-computer interface decoding control method driven by an efficient channel selection algorithm. As Figure 1 shown, this method may include the following steps: Step 101, based on the pre-collected electroencephalogram (EEG) data, calculate the comprehensive importance scores of each channel for each task respectively, and construct a channel-task importance matrix.

[0024] In the embodiment of the present application, an EEG acquisition device (for example, an EEG cap) can be used to collect EEG data from the scalp of a user. The selection of the acquisition device depends on the specific requirements of the research, such as the number of channels required, the sampling rate, and the anti-interference ability, etc.

[0025] Based on the collected EEG device, by constructing a channel-task importance matrix to evaluate the comprehensive importance of each channel under each task, key channels can be effectively identified and data redundancy can be reduced.

[0026] Step 102, determine a channel selection strategy, and select an optimal channel subset from the channel-task importance matrix according to the selected channel selection strategy.

[0027] Step 103, optimize the EEG signal processing model according to the optimal channel subset.

[0028] In the embodiment of the present application, different channel selection strategies can be selected for different application scenarios, and then an optimal channel subset adapted to the application scenario can be selected. By optimizing the EEG signal processing model with the optimal channel subset obtained under a specific strategy, different task requirements can be met and the generality of the model can be improved.

[0029] Step 104, deploy the optimized EEG signal processing model to the OpenViBE platform for real-time EEG analysis and processing.

[0030] In the embodiments of the present application, based on the pre-collected EEG data, the comprehensive importance scores of each channel are calculated for each task respectively, a channel-task importance matrix is constructed, and then an optimal channel subset is selected from the channel-task importance matrix. The EEG signal processing model is optimized according to the optimal channel subset, and the optimized EEG signal processing model is deployed to the OpenViBE platform for real-time EEG analysis and processing. This can significantly reduce the computational amount while ensuring the accuracy, reduce the demand for hardware resources, and improve the real-time performance of the model on portable devices. In addition, the corresponding channel selection strategy can be flexibly selected according to the specific application scenario, and the optimal channel subset is selected from the channel-task importance matrix according to the selected channel selection strategy, which significantly improves the generality of the model.

[0031] In an alternative embodiment, before calculating the comprehensive importance scores of each channel for each task respectively based on the pre-collected EEG data and constructing a channel-task importance matrix, the method further includes: Filter the EEG data, and use a preset method to remove eye movement and electromyogram artifacts in the EEG data.

[0032] In the embodiments of the present application, preprocessing is performed on the EEG data to obtain more accurate data that is convenient for analysis and processing. By configuring and connecting to the built-in modules provided by the OpenViBE platform, filtering can be performed according to the research objectives, and independent component analysis (ICA) or other methods can be used to remove artifacts such as eye movement and electromyogram. Finally, the data is segmented into epochs for subsequent real-time EEG analysis and processing.

[0033] In an alternative embodiment, calculating the comprehensive importance scores of each channel for each task respectively based on the pre-collected EEG data and constructing a channel-task importance matrix includes: Train an EEG signal processing model using all channel data of the EEG data to obtain the feature weights of each layer; For each task, calculate the weight gradient information according to the feature weights of each layer, obtain the importance score of each channel, and construct a channel-task importance matrix.

[0034] In the embodiments of the present application, the EEG signal processing model can be any CNN network with a spatial filtering layer, such as EEGNet, or typical EEG classification network architectures such as Shallow ConvNet.

[0035] In an alternative embodiment, for each task, calculating the weight gradient information according to the feature weights of each layer, obtaining the importance score of each channel, and constructing a channel-task importance matrix includes: For the current task, the validation set data is used to calculate the loss function, and the gradients of the feature weights of each layer with respect to the loss function are obtained through backpropagation; Based on the gradients and feature weights of each layer, calculate the channel importance scores of each channel in each layer; According to the channel importance scores of each channel in each layer, calculate the comprehensive importance scores of each channel; According to the comprehensive importance scores of each channel obtained under each task, construct a channel-task importance matrix.

[0036] In the embodiments of the present application, the preprocessed full-channel data can be used to train the electroencephalogram signal processing model. After training is completed, the weight parameters of the model are saved to prepare for subsequent channel selection. When necessary, the learning rate or the number of training epochs needs to be adjusted to ensure that the network can fully learn the spatial feature information in the full-channel data. Extract the feature weights of each layer from the trained model, including the weight parameters of multiple layers such as the spatial convolutional layer and the temporal convolutional layer. Based on the feature weights of each layer, calculate the weight gradient information through the backpropagation of the loss function to construct a channel-task importance matrix.

[0037] In the embodiments of the present application, first, for each task category t, the validation set data is used to calculate the loss function L. Then, the gradients of each layer of weights are obtained through backpropagation W with respect to the loss function ∇W . For a channel i in layer j , the gradient importance score can be expressed as:

[0038] After that, based on the weight gradient information obtained previously, calculate the comprehensive importance score of each channel:

[0039] This step comprehensively considers the contributions of multi-layer features and balances the importance of different layers through the weight coefficient α. The higher the score, the greater the impact of the channel on the model output.

[0040] In the embodiments of the present application, channel importance analysis is performed separately for each task. Construct a channel-task importance matrix M , where M i , t represents the importance score of channel i for task t . In this way, the channels that are most critical for a specific task and the channels that play important roles in multiple tasks can be identified.

[0041] In an alternative embodiment, the channel selection strategy includes:​ Global strategy, select one or more channels from all tasks in descending order of the comprehensive importance score; Task-specific strategy, select one or more channels for a specific task in descending order of the comprehensive importance score; Hybrid strategy, select one or more channels shared by multiple tasks and one or more channels for a specific task in descending order of the comprehensive importance score.

[0042] In the embodiments of the present application, any one of the above channel selection strategies can be flexibly selected according to the actual application scenario to implement an optimization solution for different application scenarios.

[0043] In an optional embodiment, determine the channel selection strategy, and select and obtain the optimal channel subset from the channel-task importance matrix according to the selected channel selection strategy, including: Determine the channel selection strategy from the global strategy, task-specific strategy, and hybrid strategy; Based on the selected channel selection strategy, determine the channel selection scheme; Select and obtain the optimal channel subset from the channel-task importance matrix according to the channel selection scheme.

[0044] In an optional embodiment, the channel selection scheme is any one of the following: Scheme 1: Predetermine the target number of channels K, and select the K channels with the highest importance scores according to the selected channel selection strategy to generate the optimal channel subset; this method is intuitive and easy to control the model scale; Scheme 2: Set an importance threshold, and select all channels with importance scores higher than the importance threshold according to the selected channel selection strategy to generate the optimal channel subset; this method can adaptively retain truly important channels; Scheme 3: Determine the constraint conditions, and on the premise of meeting the constraint conditions, select the optimal channel combination according to the selected channel selection strategy to generate the optimal channel subset; this method considers actual constraint conditions such as computing resources and device limitations, transforms the channel selection into an optimization problem, and selects the optimal channel combination on the premise of meeting the constraint conditions. The specific implementation method of this scheme is as follows: First, construct the constraint conditions. From the perspective of computing resource constraints, factors such as the memory of the hardware platform and the computing power of the CPU need to be considered. For example, if the memory of the hardware platform is limited, the data storage volume involved in the channel selection process can be used as a constraint condition. Assume that the memory size occupied by each channel's data is m , and the upper limit of the memory provided by the hardware platform for storing channel data is M max , then the constraint condition can be expressed as , where the S represents the set of candidate channels, m i is the channel i data memory occupancy. From the perspective of device limit constraints, it needs to be constructed based on the characteristics of the brain-computer interface device used. For example, the sampling frequency limit of the device. If the highest sampling frequency of the device is f max , and different channel combinations may affect the overall data sampling frequency f , then the constraint can be set as f ≤ f max .

[0045] Secondly, optimize and select the optimal channel combination. First, define the objective function, aiming at the balance between the channel importance score and meeting the constraint conditions. Let the importance score of channel i be M i , t (t is the task), and the objective function F ( S ) can be defined as , where is the penalty term for violating the constraint conditions, λ is the balance coefficient, used to adjust the weight between the importance score and constraint satisfaction.

[0046] In the selection of the optimization algorithm, the greedy algorithm can be used. Starting from the channel with the highest importance score, add channels to the candidate set S one by one. Each time a channel is added, check whether all constraint conditions are met. If not, abandon the channel and continue to add the next channel with a high score until no more channels can be added. At this time, the candidate set S is the optimal channel combination.

[0047] In an alternative embodiment, optimize the electroencephalogram (EEG) signal processing model according to the optimal channel subset, including: In the EEG data, select the data related to the optimal channel subset to train the EEG signal processing model.

[0048] The specific implementation method can be: freeze the parameters of the main layers of the network, and only adjust the weights related to the selected channels. During the fine-tuning process, use a smaller learning rate (usually 0.1 or 0.01 of the initial learning rate) to avoid destroying the feature representation learned by the model. At the same time, use the data of the original training set, but only retain the information of the selected channels for training, so that the model can adapt to the reduced channel configuration.

[0049] ​In the embodiments of the present application, through dual means of channel selection preprocessing and model optimization, while ensuring the accuracy rate, the computational amount is significantly reduced, the demand for hardware resources is decreased, and the real-time performance of the system on portable devices is improved.

[0050] In an optional embodiment, after optimizing the electroencephalogram (EEG) signal processing model according to the optimal channel subset, the method further includes: Evaluating the performance of the optimized EEG signal processing model on the validation set.

[0051] The evaluation metrics include: classification accuracy rate to ensure that channel optimization does not significantly reduce the model performance; channel compression rate, which calculates the ratio of the number of remaining channels to the number of original channels; inference time to verify the real-time performance of the optimized model; and resource occupancy to evaluate the reduction in memory usage and computational load.

[0052] In an optional embodiment, deploying the optimized EEG signal processing model to the OpenViBE platform for real-time EEG analysis and processing includes: Deploying the optimized model to the OpenViBE platform for real-time EEG analysis and processing.

[0053] Deploying the optimized model to the OpenViBE platform requires using its Python script integration function, which is mainly achieved by creating a Python script that inherits from the OVBox class. In the script, appropriate input and output interfaces need to be defined to handle signal input and classification result output, and necessary initialization, processing, and termination functions need to be implemented. In the initialization phase, the optimized model needs to be loaded, signal preprocessing parameters need to be set, and the channel selection strategy corresponding to the optimized channel subset needs to be configured. In the signal processing phase, the script will receive real-time EEG signals, perform necessary data format conversions, call the model for prediction, and output classification results. After completing the script writing, add a Python Scripting Box in the OpenViBE Designer, configure the script path, and connect the corresponding signal input and output streams. Finally, conduct system debugging, verify the data flow path, confirm the real-time processing performance, optimize the processing delay, and monitor the resource usage to ensure that the model can run stably and reliably on the OpenViBE platform.

[0054] In an optional embodiment, before deploying the optimized EEG signal processing model to the OpenViBE platform, the method further includes: Configuring the relevant parameters of the OpenViBE platform, including the sampling rate, channel data, and data format.

[0055] Configure the above - related parameters to ensure that the OpenViBE platform can correctly receive and process EEG data. After the configuration is completed, testing is required to ensure that the EEG data can be correctly transmitted from the EEG acquisition device to the OpenViBE platform.

[0056] In an optional embodiment, after deploying the optimized EEG signal processing model to the OpenViBE platform, the method further includes: Convert the classification results obtained from real - time EEG analysis and processing on the OpenViBE platform into control instructions, and output the control instructions to a specified device to instruct the specified device to perform corresponding operations.

[0057] There are mainly two solutions to implement peripheral device control on the OpenViBE platform: One is the network protocol (TCP / IP or UDP) solution. After converting the classification results into control instructions in a Python script, they are sent through a network socket to a control program running on a local or remotely specified device. After the control program receives and parses these instructions, it performs corresponding device operations. The other is the LSL (Lab Streaming Layer) interface solution. The control instructions are sent in the form of a data stream through the LSL output interface of OpenViBE. An LSL client program is written at the receiving end to receive these instructions and control the specified device. To ensure the real - time performance and reliability of control, a stable communication mechanism needs to be established, including error handling and status feedback. At the same time, an appropriate control instruction buffering strategy is set to avoid misoperations caused by signal jitter, and corresponding control logics and security protection mechanisms are designed according to the specific requirements of different types of peripheral devices.

[0058] Exemplarily, the specific implementation steps of applying the OpenViBE platform for real - time EEG analysis and processing and peripheral device control in the embodiments of the present application are as Figure 2 follows: Step 1: Use an EEG acquisition device to collect EEG data; Step 2: Drive the OpenViBE platform to receive EEG data; In the embodiments of the present application, the OpenViBE platform is configured to receive and process data from the EEG acquisition device; Step 3: Pre - process the EEG data to remove eye movement and electromyogram artifacts in the EEG data; Step 4: Real - time model optimization and deployment, that is, optimize the EEG signal processing model through the above - mentioned steps 101 - 103. In step 104, deploy the optimized EEG signal processing model on the OpenViBE platform to achieve real - time EEG analysis and processing on the OpenViBE platform; Step 5, Peripheral control: Convert the classification results obtained through real-time EEG analysis and processing on the OpenViBE platform into control instructions, and output the control instructions to the specified device to instruct the specified device to perform corresponding operations.

[0059] More specifically, as Figure 3 shown, the real-time optimization and deployment of the EEG signal processing model in Step 4 can include a data preparation stage, a feature analysis stage, and a model optimization stage.

[0060] In the data preparation stage: a. Train the model using all-channel data; after training is completed, save the weight parameters of the model to prepare for subsequent channel selection.

[0061] b. Multilayer feature weight extraction: Extract the feature weights of each layer from the trained model.

[0062] c. Calculate the weight gradient information: For each task t, use the validation set data to calculate the loss function L, and then obtain the gradient ∇W of each layer's feature weight W with respect to the loss function through backpropagation, and determine the gradient importance score of channel i in layer j.

[0063] In the feature analysis stage: d. Calculate the gradient-weighted channel importance: Based on the weight gradient information obtained in step c, calculate the comprehensive importance score of each channel.

[0064] e. Task-specific channel analysis: Conduct channel importance analysis for each task separately. Construct a channel-task importance matrix M.

[0065] f. Determine the channel selection strategy: Select the corresponding strategy from the global strategy, task-specific strategy, and hybrid strategy; g. Select the optimal channel subset: According to the selected strategy, determine the final channel selection plan, and select the optimal channel subset according to the final channel selection plan; In the model optimization stage: h. Retrain the model using specific channels: Freeze the parameters of the main layers of the network and only adjust the weights related to the selected channels.

[0066] i. Evaluate the model performance: Comprehensively evaluate the performance of the optimized model on the validation set.

[0067] j. Final model deployment: Deploy the optimized model to the OpenViBE platform.

[0068] Based on the above method, the embodiments of the present application can effectively solve the channel redundancy problem, significantly improve the decoding accuracy, generalization ability and computational efficiency. Applying this method can construct a more stable, reliable and easy-to-apply brain-computer interface system.

[0069] Based on the same principle as the method provided by the embodiments of the present application, the embodiments of the present application also provide an OpenViBE brain-computer interface decoding control system driven by an efficient channel selection algorithm, as Figure 4 shown, the system includes: A channel importance calculation module 401, configured to calculate the comprehensive importance scores of each channel for each task respectively based on the pre-collected electroencephalogram data, and construct a channel-task importance matrix.

[0070] In the embodiments of the present application, an electroencephalogram acquisition device (for example, an EEG cap) can be used to collect electroencephalogram data from the scalp of a user. The selection of the acquisition device depends on the specific requirements of the research, such as the number of channels required, the sampling rate, and the anti-interference ability.

[0071] Based on the collected electroencephalogram device, by constructing a channel-task importance matrix to evaluate the comprehensive importance of each channel under each task, key channels can be effectively identified and data redundancy can be reduced.

[0072] A channel selection module 402, configured to determine a channel selection strategy, and select an optimal channel subset from the channel-task importance matrix according to the selected channel selection strategy.

[0073] A model optimization module 403, configured to optimize the electroencephalogram signal processing model according to the optimal channel subset.

[0074] In the embodiments of the present application, different channel selection strategies can be selected for different application scenarios, and then an optimal channel subset adapted to the application scenario can be selected. By optimizing the electroencephalogram signal processing model with the optimal channel subset obtained under a specific strategy, different task requirements can be met and the generality of the model can be improved.

[0075] A model deployment module 404, configured to deploy the optimized electroencephalogram signal processing model to the OpenViBE platform for real-time electroencephalogram analysis and processing.

[0076] In the embodiments of the present application, based on the pre-collected EEG data, the comprehensive importance scores of each channel are calculated for each task respectively to construct a channel-task importance matrix. Then, an optimal channel subset is selected from the channel-task importance matrix, and the EEG signal processing model is optimized according to the optimal channel subset. The optimized EEG signal processing model is deployed to the OpenViBE platform for real-time EEG analysis and processing. This can significantly reduce the computational amount while ensuring the accuracy, reduce the demand for hardware resources, and improve the real-time performance of the model on portable devices. In addition, the corresponding channel selection strategy can be flexibly selected according to the specific application scenario, and the optimal channel subset is selected from the channel-task importance matrix according to the selected channel selection strategy, which significantly improves the versatility of the system.

[0077] In an alternative embodiment, the channel importance calculation module 401 includes: The first channel importance calculation sub-module is used to train the EEG signal processing model using all channel data of the EEG data to obtain the feature weights of each layer; The second channel importance calculation sub-module is used to, for each task, calculate the weight gradient information according to the feature weights of each layer, obtain the importance score of each channel, and construct a channel-task importance matrix.

[0078] In an alternative embodiment, the second channel importance calculation sub-module includes: The first channel importance calculation unit is used to, for the current task, calculate the loss function using the validation set data and obtain the gradient of the feature weights of each layer with respect to the loss function through backpropagation; The second channel importance calculation unit is used to calculate the channel importance score of each channel in each layer based on the gradients and feature weights of each layer; The third channel importance calculation unit is used to calculate the comprehensive importance score of each channel according to the channel importance score of each channel in each layer; The fourth channel importance calculation unit is used to construct a channel-task importance matrix according to the comprehensive importance score of each channel obtained under each task.

[0079] In an alternative embodiment, the channel selection strategy includes: The global strategy is to select one or more channels in all tasks in descending order of the comprehensive importance score; The task-specific strategy is to select one or more channels for a specific task in descending order of the comprehensive importance score; The hybrid strategy is to select one or more channels shared by multiple tasks and one or more channels for a specific task in descending order of the comprehensive importance score.

[0080] In an alternative embodiment, the channel selection module 402 includes: A first channel selection sub-module for determining a channel selection policy from a global policy, a task-specific policy, and a hybrid policy; A second channel selection sub-module for determining a channel selection scheme based on the selected channel selection policy; A third channel selection sub-module for selecting and obtaining an optimal channel subset from the channel-task importance matrix according to the channel selection scheme.

[0081] In an alternative embodiment, the channel selection scheme is any one of the following: Scheme 1: Predetermine the target number of channels K, and select the K channels with the highest importance scores according to the selected channel selection policy to generate an optimal channel subset; Scheme 2: Set an importance threshold, and select all channels with importance scores higher than the importance threshold according to the selected channel selection policy to generate an optimal channel subset; Scheme 3: Determine a constraint condition, and select an optimal channel combination according to the selected channel selection policy on the premise of satisfying the constraint condition to generate an optimal channel subset.

[0082] In an alternative embodiment, the model optimization module 403 includes: A model optimization sub-module for selecting data related to the optimal channel subset in the EEG data to train an EEG signal processing model.

[0083] In an alternative embodiment, the system further includes: An EEG preprocessing module for filtering the EEG data and removing eye movement and electromyogram artifacts in the EEG data using a preset method.

[0084] In an alternative embodiment, the system further includes: An OpenViBE driver module for configuring relevant parameters of the OpenViBE platform, including the sampling rate, channel data, and data format.

[0085] In an alternative embodiment, the system further includes: A peripheral control module for converting the classification result obtained from the real-time EEG analysis and processing on the OpenViBE platform into a control instruction, and outputting the control instruction to a specified device to instruct the specified device to perform corresponding operations.

[0086] Exemplarily, the specific system for real-time EEG analysis and processing using the OpenViBE platform and performing peripheral control in the embodiments of the present application is as Figure 5 shown: The electroencephalogram (EEG) acquisition device 501 is used to acquire EEG data; The OpenViBE driver module 502 is used to drive the OpenViBE platform and receive EEG data; in the embodiments of the present application, the OpenViBE platform is configured to receive and process data from the EEG acquisition device; The EEG preprocessing module 503 is used to preprocess the EEG data and remove eye movement and electromyogram artifacts from the EEG data; The real-time model optimization and deployment module 504 is used to optimize the EEG signal processing model through the above-mentioned channel importance calculation module 401, channel selection module 402, and model optimization module, and deploy the optimized EEG signal processing model on the OpenViBE platform through the model deployment module 404 to realize real-time EEG analysis and processing on the OpenViBE platform; The peripheral control module 505 is used to convert the classification result obtained through real-time EEG analysis and processing on the OpenViBE platform into a control instruction and output the control instruction to a specified device to instruct the specified device to perform corresponding operations.

[0087] The OpenViBE brain-computer interface decoding control system driven by the efficient channel selection algorithm provided in the embodiments of the present application can implement Figures 1 to 4 each process implemented in the method embodiments. To avoid repetition, it will not be described here again.

[0088] The OpenViBE brain-computer interface decoding control system driven by the efficient channel selection algorithm in the embodiments of the present application can execute the OpenViBE brain-computer interface decoding control method provided in the embodiments of the present application, and its implementation principle is similar. The actions performed by each module and unit in the OpenViBE brain-computer interface decoding control system driven by the efficient channel selection algorithm in the embodiments of the present application correspond to the steps in the OpenViBE brain-computer interface decoding control method driven by the efficient channel selection algorithm in the embodiments of the present application. For the detailed function descriptions of each module of the OpenViBE brain-computer interface decoding control system driven by the efficient channel selection algorithm, reference can be specifically made to the descriptions in the corresponding OpenViBE brain-computer interface decoding control method shown in the foregoing text, and it will not be described here again.

[0089] The above description is only a preferred embodiment of the present application and an explanation of the applied technical principles. Those skilled in the art should understand that the scope of disclosure involved in the present application is not limited to the technical solutions formed by the specific combination of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosure concept. For example, the technical solutions formed by mutually replacing the above features with the technical features (but not limited to) disclosed in the present application that have similar functions.

Claims

1. An OpenViBE brain-computer interface decoding control method driven by an efficient channel selection algorithm, characterized in that The method includes: Based on pre - collected EEG data, calculate the comprehensive importance scores of each channel for each task respectively, and construct a channel - task importance matrix; Determine a channel selection strategy, and select an optimal channel subset from the channel - task importance matrix according to the selected channel selection strategy; Optimize the EEG signal processing model according to the optimal channel subset; Deploy the optimized EEG signal processing model to the OpenViBE platform for real - time EEG analysis and processing.

2. The decoding control method of the OpenViBE brain-computer interface driven by the efficient channel selection algorithm according to claim 1, characterized in that The step of "Based on pre - collected EEG data, calculate the comprehensive importance scores of each channel for each task respectively, and construct a channel - task importance matrix" includes: Use all channel data of the EEG data to train an EEG signal processing model to obtain the feature weights of each layer; For each task, calculate the weight gradient information according to the feature weights of each layer, obtain the importance score of each channel, and construct the channel - task importance matrix.

3. The decoding control method of the OpenViBE brain-computer interface driven by the efficient channel selection algorithm according to claim 2, wherein The step of "For each task, calculate the weight gradient information according to the feature weights of each layer, obtain the importance score of each channel, and construct the channel - task importance matrix" includes: For the current task, calculate the loss function using the validation set data, and obtain the gradient of the feature weights of each layer with respect to the loss function through backpropagation; Based on the gradients and feature weights of each layer, calculate the channel importance scores of each channel in each layer; Calculate the comprehensive importance score of each channel according to the channel importance scores of each channel in each layer; Construct the channel - task importance matrix according to the comprehensive importance scores of each channel obtained under each task.

4. The decoding control method of the OpenViBE brain-computer interface driven by the efficient channel selection algorithm according to any one of claims 1-3, characterized in that The channel selection strategy includes: A global strategy, which selects one or more channels in all tasks in descending order of the comprehensive importance scores; A task - specific strategy, which selects one or more channels for a specific task in descending order of the comprehensive importance scores; A hybrid strategy, which selects one or more channels shared by multiple tasks and one or more channels for a specific task in descending order of the comprehensive importance scores.

5. The decoding control method of the OpenViBE brain-computer interface driven by the efficient channel selection algorithm according to claim 4, wherein The step of "Determine a channel selection strategy, and select an optimal channel subset from the channel - task importance matrix according to the selected channel selection strategy" includes: Determine a channel selection strategy from the global strategy, the task - specific strategy, and the hybrid strategy; Based on the selected channel selection strategy, determine a channel selection scheme; Select the optimal channel subset from the channel - task importance matrix according to the channel selection scheme.

6. The decoding control method of the OpenViBE brain-computer interface driven by the efficient channel selection algorithm according to claim 5, wherein The channel selection scheme is any one of the following: Scheme 1: Predetermine the target number of channels K, and select the K channels with the highest importance scores according to the selected channel selection strategy to generate the optimal channel subset; Scheme 2: Set an importance threshold, and select all channels with importance scores higher than the importance threshold according to the selected channel selection strategy to generate the optimal channel subset; Scheme 3: Determine a constraint condition, and select the optimal channel combination according to the selected channel selection strategy under the premise of meeting the constraint condition to generate the optimal channel subset.

7. The decoding control method of the OpenViBE brain-computer interface driven by the high-efficiency channel selection algorithm according to claim 2, wherein Optimizing the EEG signal processing model according to the optimal channel subset includes: In the EEG data, select the data related to the optimal channel subset to train the EEG signal processing model.

8. The decoding control method of the OpenViBE brain-computer interface driven by the efficient channel selection algorithm according to claim 2, characterized in that Before calculating the comprehensive importance score of each channel for each task based on the pre-collected EEG data and constructing the channel-task importance matrix, the method further includes: Filter the EEG data and use a preset method to remove eye movement and electromyogram artifacts in the EEG data.

9. The decoding control method of the OpenViBE brain-computer interface driven by the efficient channel selection algorithm according to claim 1, characterized in that Before deploying the optimized EEG signal processing model to the OpenViBE platform for real-time EEG analysis and processing, the method further includes: Configure the relevant parameters of the OpenViBE platform, including the sampling rate, channel data, and data format.

10. An OpenViBE brain-computer interface decoding control system driven by an efficient channel selection algorithm, characterized in that The system includes: A channel importance calculation module, configured to calculate the comprehensive importance score of each channel for each task based on the pre-collected EEG data and construct a channel-task importance matrix; A channel selection module, configured to determine a channel selection strategy and select an optimal channel subset from the channel-task importance matrix according to the selected channel selection strategy; A model optimization module, configured to optimize the EEG signal processing model according to the optimal channel subset; A model application module, configured to deploy the optimized EEG signal processing model to the OpenViBE platform for real-time EEG analysis and processing.

Images