Data optimization method, device and equipment for electroencephalogram signal classification
By filtering and segmenting EEG signals and calculating approximate entropy, and by optimizing low-quality datasets using channel electrode location maps, the problem of insufficient data utilization in EEG signal classification in existing technologies is solved, and high-precision EEG signal classification is achieved.
Patent Information
- Application Number
- CN202510949126.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-10
- Publication Date
- 2025-10-28
AI Technical Summary
Existing technologies for EEG signal classification suffer from problems such as relying on heavy prior knowledge or ignoring individual differences in general methods, making it difficult to effectively utilize some low-quality EEG datasets.
By filtering and segmenting the initial EEG signal, calculating approximate entropy values to filter high- and low-quality datasets, optimizing low-quality datasets using channel electrode location maps, and merging high-quality datasets to improve dataset usability.
This improves the usability of low-quality EEG datasets, providing a reliable foundation for subsequent EEG classification and enabling robust and high-precision classification results.
Smart Images
Figure CN120850003A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of computer technology, and more particularly to the field of intelligent information processing technology, specifically to a data optimization method, apparatus, and device for classifying electroencephalogram (EEG) signals. Background Technology
[0002] In the field of brain-computer interfaces based on motor imagery, high-quality preprocessing of raw EEG signals is a crucial first step in ensuring the accuracy and reliability of the system.
[0003] Existing technologies mainly revolve around two core tasks: first, suppressing noise and artifacts to improve signal quality, such as the universal average reference method for reducing common-mode noise between electrodes; and second, personalized signal processing to address significant individual differences, such as the attention-based adaptive channel weighting method that optimizes interpolation processing of bad leads by learning channel correlations within individual data. Although existing technologies have made some progress in EEG signal processing, a significant gap exists: one type of method focuses on general signal cleansing but relies heavily on prior knowledge; another type attempts to address individual differences but introduces new prior knowledge dependencies; and yet another type reduces prior dependencies but ignores individual differences. These problems make it difficult to use some low-quality EEG datasets for EEG signal classification. Summary of the Invention
[0004] This application provides a data optimization method, apparatus, and device for classifying electroencephalogram (EEG) signals to improve the usability of low-quality EEG datasets.
[0005] According to one aspect of this application, a data optimization method for electroencephalogram (EEG) signal classification is provided, the method comprising:
[0006] The initial EEG signal of the target user is filtered and segmented to obtain an initial EEG dataset; wherein, the initial EEG signal is obtained by detecting the user's EEG signal through at least one channel electrode; the initial EEG dataset includes at least one EEG signal matrix;
[0007] Approximate entropy calculation is performed on the at least one EEG signal matrix to obtain the approximate entropy value of the at least one EEG signal matrix;
[0008] Based on the approximate entropy value of the at least one EEG signal matrix, high-quality EEG datasets and low-quality EEG datasets are selected from the initial EEG dataset.
[0009] Based on the approximate entropy value of the at least one EEG signal matrix and the channel electrode location map of the target user, the low-quality EEG dataset is optimized to obtain a candidate EEG dataset; wherein, the channel electrode location map is used to characterize the location distribution of the at least one channel electrode;
[0010] The high-quality EEG dataset is merged with the candidate EEG dataset to obtain the target EEG dataset for EEG signal classification.
[0011] According to another aspect of this application, a data optimization device for electroencephalogram (EEG) signal classification is provided, the device comprising:
[0012] The signal preprocessing module is used to filter and segment the initial EEG signal of the target user to obtain an initial EEG dataset; wherein, the initial EEG signal is obtained by detecting the user's EEG signal through at least one channel electrode; the initial EEG dataset includes at least one EEG signal matrix;
[0013] An approximate entropy calculation module is used to perform approximate entropy calculation on the at least one EEG signal matrix to obtain the approximate entropy value of the at least one EEG signal matrix;
[0014] The dataset partitioning module is used to filter out high-quality EEG datasets and low-quality EEG datasets from the initial EEG dataset based on the approximate entropy value of the at least one EEG signal matrix.
[0015] The dataset optimization module is used to optimize the low-quality EEG dataset based on the approximate entropy value of the at least one EEG signal matrix and the channel electrode location map of the target user to obtain a candidate EEG dataset; wherein, the channel electrode location map is used to characterize the location distribution of the at least one channel electrode;
[0016] The dataset merging module is used to merge the high-quality EEG dataset with the candidate EEG dataset to obtain the target EEG dataset for EEG signal classification.
[0017] According to another aspect of this application, an electronic device is provided, the electronic device comprising:
[0018] One or more processors;
[0019] Memory, used to store one or more programs;
[0020] When the one or more programs are executed by the one or more processors, the one or more processors implement any of the data optimization methods for EEG signal classification provided in the embodiments of this application.
[0021] According to another aspect of this application, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements any of the data optimization methods for electroencephalogram (EEG) signal classification provided in the embodiments of this application.
[0022] According to another aspect of this application, a computer program product is provided, including a computer program that, when executed by a processor, implements any of the data optimization methods for EEG signal classification provided in the embodiments of this application.
[0023] This application obtains an initial EEG dataset by filtering and segmenting the initial EEG signals of a target user. The initial EEG signals are obtained by detecting the user's EEG signals through at least one channel electrode. The initial EEG dataset includes at least one EEG signal matrix. Approximate entropy is calculated on the at least one EEG signal matrix to obtain its approximate entropy value. Based on the approximate entropy value of the at least one EEG signal matrix, high-quality and low-quality EEG datasets are selected from the initial EEG dataset. The low-quality EEG dataset is optimized based on the approximate entropy value of the at least one EEG signal matrix and the channel electrode location map of the target user to obtain a candidate EEG dataset. The channel electrode location map is used to characterize the positional distribution of at least one channel electrode. The high-quality EEG dataset and the candidate EEG dataset are merged to obtain the target EEG dataset for EEG signal classification. This technical solution improves the usability of the low-quality dataset through the synergistic effect of adaptive quality assessment of the EEG dataset and self-supervised repair of the low-quality dataset, laying a reliable foundation for subsequent EEG classification and enabling robust and high-precision classification results from a poor-quality data foundation. Attached Figure Description
[0024] Figure 1 This is a flowchart of a data optimization method for electroencephalogram (EEG) signal classification according to Embodiment 1 of this application;
[0025] Figure 2 This is a flowchart of a data optimization method for electroencephalogram (EEG) signal classification according to Embodiment 2 of this application;
[0026] Figure 3 This is a schematic diagram of a data optimization device for electroencephalogram (EEG) signal classification according to Embodiment 3 of this application;
[0027] Figure 4 This is a schematic diagram of the structure of an electronic device that implements the data optimization method for EEG signal classification according to Embodiment 4 of this application. Detailed Implementation
[0028] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.
[0029] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0030] Furthermore, it should be noted that the collection, storage, use, processing, transmission, provision, and disclosure of initial EEG signals and EEG signal matrices and other related data involved in the technical solution of this application all comply with the provisions of relevant laws and regulations and do not violate public order and good morals.
[0031] Example 1
[0032] Figure 1 This is a flowchart of a data optimization method for EEG signal classification according to Embodiment 1 of this application. This embodiment is applicable to optimizing low-quality EEG datasets used for EEG signal classification. It can be executed by a data optimization device for EEG signal classification, which can be implemented in hardware and / or software and can be configured in a computer device, such as a server. Figure 1 As shown, the method includes:
[0033] S110. The initial EEG signal of the target user is filtered and segmented to obtain an initial EEG dataset; wherein, the initial EEG signal is obtained by detecting the user's EEG signal through at least one channel electrode; the initial EEG dataset includes at least one EEG signal matrix.
[0034] In this embodiment, the target user refers to the user currently undergoing EEG signal detection. The initial EEG signal refers to the signal detected from the target user's EEG activity through at least one channel electrode. Channel electrodes are part of the EEG instrument and are responsible for receiving EEG signals from the target user's scalp; one electrode records the signal from one channel, and multiple electrodes are used to record EEG activity at different locations. The EEG signal matrix refers to the result of arranging the EEG signals acquired by multiple channel electrodes in a time sequence; each row of the matrix typically represents data from multiple electrodes at a certain point in time, and each column represents data within a specific time period.
[0035] Optionally, the initial EEG signal of the target user can be filtered by a bandpass filter and a notch filter to obtain candidate EEG signals; based on a preset time window, the candidate EEG signals can be divided into segments of fixed time length to obtain an initial EEG dataset.
[0036] In this embodiment, the bandpass filter is an electronic filter that allows signals within a certain frequency range to pass through while suppressing signals outside that range; it is typically used to remove noise and retain signals in the frequency band of interest. The notch filter is a filter specifically designed to eliminate noise within a specific frequency range. The candidate EEG signal refers to the filtered EEG signal, which retains key information and removes most noise and interference. The preset time window is manually set based on actual conditions or experience; this embodiment does not specifically limit its usage.
[0037] For example, noise and artifacts in the initial EEG signal of the target user are removed by using a bandpass filter and a notch filter to obtain candidate EEG signals; according to a preset time window, the continuous candidate EEG signals are cut into multiple segments according to a fixed time period, and each segment is used for independent analysis or processing to obtain an initial EEG dataset composed of multiple segments.
[0038] S120. Perform approximate entropy calculation on at least one EEG signal matrix to obtain the approximate entropy value of at least one EEG signal matrix.
[0039] In this embodiment, the approximate entropy value is an indicator used to measure the complexity and irregularity of a signal. It evaluates the predictability and irregularity of a signal by performing local complexity analysis on the data sequence. A higher approximate entropy value indicates that the signal is more complex or more random, while a lower approximate entropy value indicates that the signal is more regular or periodic.
[0040] For example, by selecting an embedding dimension of 2 and a tolerance factor of 0.2, an approximate entropy calculation is performed on at least one EEG signal matrix to obtain an approximate entropy value for at least one EEG signal matrix.
[0041] S130. Based on the approximate entropy value of at least one EEG signal matrix, select high-quality EEG datasets and low-quality EEG datasets from the initial EEG dataset.
[0042] In this embodiment, a high-quality EEG dataset refers to a collection of EEG data with clear signals, low noise, and the ability to effectively represent brain activity. A low-quality EEG dataset refers to a collection of data whose signals are significantly affected by noise or artifacts.
[0043] For example, a competitive learning network is used to cluster all approximate entropy values. Based on the distribution of approximate entropy values, the boundary between high quality and low quality is automatically determined, thereby filtering high-quality and low-quality EEG datasets from the initial EEG dataset according to this boundary.
[0044] Optionally, an approximate entropy threshold for at least one EEG signal matrix is determined based on the approximate entropy value of at least one EEG signal matrix; high-quality EEG datasets and low-quality EEG datasets are selected from the initial EEG dataset by comparing the approximate entropy value of at least one EEG signal matrix with the approximate entropy threshold.
[0045] In this embodiment, the approximate entropy threshold refers to the standard or boundary value used to distinguish different signal states in EEG signal data processing.
[0046] S140. Based on the approximate entropy value of at least one EEG signal matrix and the channel electrode location map of the target user, optimize the low-quality EEG dataset to obtain a candidate EEG dataset; wherein, the channel electrode location map is used to characterize the location distribution of at least one channel electrode.
[0047] In this embodiment, the channel electrode location map refers to a schematic diagram or data table describing the positional distribution of various electrodes on the scalp in the EEG instrument; this location map helps in understanding the activity of different brain regions. The candidate EEG dataset refers to a collection of data with better signal quality, optimized from a low-quality dataset.
[0048] Optionally, weighted data of the low-quality EEG dataset is determined based on the approximate entropy value of at least one EEG signal matrix and the channel electrode location map of the target user, and the low-quality EEG dataset is weighted using the weighted data to obtain a candidate EEG dataset.
[0049] In this embodiment, weighted data refers to data used to enhance the signal of a low-quality EEG dataset.
[0050] S150. Merge the high-quality EEG dataset with the candidate EEG dataset to obtain the target EEG dataset for EEG signal classification.
[0051] In this embodiment, the target EEG dataset is the complete dataset ultimately used for EEG signal classification.
[0052] For example, a high-quality EEG dataset is merged with an enhanced low-quality EEG dataset to obtain a final merged target EEG dataset, which is then fed into a classification model for classification of motor imagery tasks.
[0053] This application embodiment obtains an initial EEG dataset by filtering and segmenting the initial EEG signal of the target user. The initial EEG signal is obtained by detecting the user's EEG signal through at least one channel electrode. The initial EEG dataset includes at least one EEG signal matrix. Approximate entropy is calculated on the at least one EEG signal matrix to obtain its approximate entropy value. Based on the approximate entropy value of the at least one EEG signal matrix, high-quality and low-quality EEG datasets are selected from the initial EEG dataset. The low-quality EEG dataset is optimized based on the approximate entropy value of the at least one EEG signal matrix and the channel electrode location map of the target user to obtain a candidate EEG dataset. The channel electrode location map is used to characterize the location distribution of at least one channel electrode. The high-quality EEG dataset and the candidate EEG dataset are merged to obtain the target EEG dataset for EEG signal classification. This technical solution improves the usability of the low-quality dataset through the synergistic effect of adaptive quality assessment of the EEG dataset and self-supervised repair of the low-quality dataset, laying a reliable foundation for subsequent EEG classification and enabling robust and high-precision classification results from a poor-quality data foundation.
[0054] Example 2
[0055] Figure 2 This is a flowchart of a data optimization method for EEG signal classification according to Embodiment 2 of this application. Based on the technical solutions of the above embodiments, this embodiment refines the process of "optimizing a low-quality EEG dataset based on the approximate entropy value of at least one EEG signal matrix and the channel electrode position map of the target user to obtain a candidate EEG dataset" into "calculating the relative great circle distance of the current position data of at least one channel electrode in the channel electrode position map of the target user to obtain a spatial weight matrix; using the spatial weight matrix, weighting the low-quality EEG dataset to obtain a weighted EEG dataset; optimizing the low-quality EEG dataset based on the spatial weight matrix, the weighted EEG dataset, and the approximate entropy value of at least one EEG signal matrix to obtain a candidate EEG dataset." It should be noted that for parts not detailed in this embodiment, please refer to the relevant descriptions in other embodiments. Figure 2 As shown, the method includes:
[0056] S210. Filter and segment the initial EEG signal of the target user to obtain the initial EEG dataset.
[0057] S220. Perform approximate entropy calculation on at least one EEG signal matrix to obtain an approximate entropy value for at least one EEG signal matrix.
[0058] S230. Based on the approximate entropy value of at least one EEG signal matrix, select high-quality EEG datasets and low-quality EEG datasets from the initial EEG dataset.
[0059] Optionally, based on a competitive learning network, cluster analysis is performed on the approximate entropy values of at least one EEG signal matrix to obtain an approximate entropy threshold for at least one EEG signal matrix; based on the approximate entropy threshold and the approximate entropy values of at least one EEG signal matrix, high-quality EEG datasets and low-quality EEG datasets are selected from the initial EEG dataset.
[0060] In this embodiment, the competitive learning network is an unsupervised learning neural network structure, mainly used to cluster or classify data through a competitive mechanism. Clustering analysis is an unsupervised learning technique that aims to divide data in a dataset into several clusters, such that data within the same cluster has high similarity, while data between different clusters has low similarity. The approximate entropy threshold is a standard value used to distinguish different types of signals; the approximate entropy value obtained through clustering analysis can be used to set the threshold to distinguish between high-quality and low-quality signals; a lower approximate entropy value may indicate a more regular signal (such as normal brain electrical activity), while a higher approximate entropy value may indicate a complex or noisy signal (such as abnormal signals or noise).
[0061] Furthermore, a numerical comparison is made between the approximate entropy threshold and the approximate entropy value of at least one EEG signal matrix to obtain the comparison result of the approximate entropy value of at least one EEG signal matrix. Based on the comparison result of the approximate entropy value, at least one EEG signal matrix in the initial EEG dataset that meets the low-quality condition of the dataset is determined as a low-quality EEG dataset. Based on the comparison result of the approximate entropy value, at least one EEG signal matrix in the initial EEG dataset that does not meet the low-quality condition of the dataset is determined as a high-quality EEG dataset.
[0062] In this embodiment, the approximate entropy comparison result refers to the numerical comparison result between the approximate entropy value of at least one EEG signal matrix and the approximate entropy threshold. The low-quality condition of the dataset is artificially preset based on actual conditions or empirical values, and this embodiment does not specifically limit it; for example, the low-quality condition of the dataset can be that the approximate entropy value is greater than or equal to the approximate entropy threshold.
[0063] S240. Calculate the relative great circle distance for the current position data of at least one channel electrode in the channel electrode position map of the target user to obtain the spatial weight matrix.
[0064] In this embodiment, relative great circle distance is a metric used to calculate the distance between two points on the Earth's surface, typically applied in spherical coordinate systems. In electroencephalogram (EEG) signal analysis, the relative positions of electrodes can be described using this distance metric, thereby revealing the spatial relationships between electrodes. The spatial weight matrix is a matrix used in EEG signal analysis that describes the spatial relationships between individual electrodes; the elements in the matrix represent the spatial distances or relationships between different electrodes.
[0065] For example, by using the spatial physical location information of the electrodes on the target user's EEG cap, the relative great circle distance between the electrodes is calculated, thereby constructing a spatial weight matrix.
[0066] S250. Using a spatial weight matrix, the low-quality EEG dataset is weighted to obtain a weighted EEG dataset.
[0067] In this embodiment, the weighted EEG dataset refers to the dataset obtained by weighting the data using a spatial weight matrix based on the low-quality EEG dataset. The purpose of weighting is to adjust the contribution of the signal according to the spatial relationship between the electrodes in order to enhance the signal quality.
[0068] S260. Based on the spatial weight matrix, the weighted EEG dataset, and the approximate entropy value of at least one EEG signal matrix, optimize the low-quality EEG dataset to obtain candidate EEG datasets.
[0069] Optionally, based on an improved autoencoder, latent features are extracted from the weighted EEG dataset to obtain a latent feature set of the weighted EEG dataset; wherein, the improved autoencoder is trained on the weighted EEG dataset; the loss function of the improved autoencoder is determined based on the approximate entropy value of at least one EEG signal matrix; a linear regression algorithm is used to perform linear regression mapping on the latent feature set to obtain the scaling factor of the latent feature set; and the low-quality EEG dataset is weighted according to the spatial weight matrix and the scaling factor to obtain a candidate EEG dataset.
[0070] In this embodiment, the improved autoencoder refers to an autoencoder trained by learning a compressed latent representation of a weighted EEG dataset. The latent feature set refers to the latent features extracted from the weighted EEG dataset through processing with the improved autoencoder; these features are a compressed representation of the input data, retaining important information while reducing redundancy. The loss function is used to measure the difference between the autoencoder output and the target output; it can be understood that by incorporating approximate entropy values, the improved autoencoder can better understand the complexity of the signal and optimize the model's performance during training. Linear regression is a basic statistical learning method used to establish a linear relationship between input and output variables; in this process, linear regression is used to map the latent feature set to derive a scaling factor for the latent feature set. The scaling factor, derived through the linear regression algorithm, represents how the weights of the latent features are adjusted.
[0071] For example, an autoencoder is trained to learn a compressed latent representation of weighted low-quality data; the loss function for training the autoencoder is determined based on approximate entropy, aiming to ensure that the reconstructed signal has similar complexity to the original signal; latent features of each low-quality signal are extracted using the encoder part of the improved autoencoder, and then a linear regression algorithm is used to map the latent features to an optimal personalized scaling factor, the goal of which is to make the approximate entropy level of the augmented low-quality data match the average level of the approximate entropy of the high-quality dataset; the low-quality EEG dataset is augmented using the scaling factor and spatial weight matrix to obtain a candidate EEG dataset.
[0072] S270. Merge the high-quality EEG dataset with the candidate EEG dataset to obtain the target EEG dataset for EEG signal classification.
[0073] This application embodiment obtains an initial EEG dataset by filtering and segmenting the initial EEG signal of the target user; calculates the approximate entropy of at least one EEG signal matrix to obtain an approximate entropy value of at least one EEG signal matrix; selects high-quality and low-quality EEG datasets from the initial EEG dataset based on the approximate entropy value of at least one EEG signal matrix; calculates the relative great circle distance of the current position data of at least one channel electrode in the channel electrode location map of the target user to obtain a spatial weight matrix; uses the spatial weight matrix to weight the low-quality EEG dataset to obtain a weighted EEG dataset; optimizes the low-quality EEG dataset based on the spatial weight matrix, the weighted EEG dataset, and the approximate entropy value of at least one EEG signal matrix to obtain a candidate EEG dataset; and merges the high-quality EEG dataset with the candidate EEG dataset to obtain the target EEG dataset for EEG signal classification. The above technical solution improves the usability of low-quality datasets through the synergistic effect of adaptive quality assessment of EEG datasets and self-supervised repair of low-quality datasets, laying a reliable foundation for subsequent EEG classification and enabling subsequent EEG classification to obtain robust and high-precision classification results from poor-quality data.
[0074] Example 3
[0075] Figure 3 This is a schematic diagram of a data optimization device for EEG signal classification according to Embodiment 3 of this application. It can be applied to the optimization of low-quality EEG datasets for EEG signal classification. The data optimization device for EEG signal classification can be implemented in hardware and / or software and can be configured in a computer device, such as a server.
[0076] like Figure 3 As shown, the device includes:
[0077] The signal preprocessing module 310 is used to filter and segment the initial EEG signal of the target user to obtain an initial EEG dataset; wherein, the initial EEG signal is obtained by detecting the user's EEG signal through at least one channel electrode; the initial EEG dataset includes at least one EEG signal matrix;
[0078] The approximate entropy calculation module 320 is used to perform approximate entropy calculation on at least one EEG signal matrix to obtain the approximate entropy value of at least one EEG signal matrix;
[0079] The dataset partitioning module 330 is used to filter high-quality EEG datasets and low-quality EEG datasets from the initial EEG dataset based on the approximate entropy value of at least one EEG signal matrix.
[0080] The dataset optimization module 340 is used to optimize the low-quality EEG dataset based on the approximate entropy value of at least one EEG signal matrix and the channel electrode location map of the target user to obtain a candidate EEG dataset; wherein, the channel electrode location map is used to characterize the location distribution of at least one channel electrode.
[0081] The dataset merging module 350 is used to merge high-quality EEG datasets with candidate EEG datasets to obtain a target EEG dataset for EEG signal classification.
[0082] This application embodiment obtains an initial EEG dataset by filtering and segmenting the initial EEG signal of the target user. The initial EEG signal is obtained by detecting the user's EEG signal through at least one channel electrode. The initial EEG dataset includes at least one EEG signal matrix. Approximate entropy is calculated on the at least one EEG signal matrix to obtain its approximate entropy value. Based on the approximate entropy value of the at least one EEG signal matrix, high-quality and low-quality EEG datasets are selected from the initial EEG dataset. The low-quality EEG dataset is optimized based on the approximate entropy value of the at least one EEG signal matrix and the channel electrode location map of the target user to obtain a candidate EEG dataset. The channel electrode location map is used to characterize the location distribution of at least one channel electrode. The high-quality EEG dataset and the candidate EEG dataset are merged to obtain the target EEG dataset for EEG signal classification. This technical solution improves the usability of the low-quality dataset through the synergistic effect of adaptive quality assessment of the EEG dataset and self-supervised repair of the low-quality dataset, laying a reliable foundation for subsequent EEG classification and enabling robust and high-precision classification results from a poor-quality data foundation.
[0083] Optionally, the dataset optimization module 340 includes:
[0084] The weight matrix determination unit is used to calculate the relative great circle distance of the current position data of at least one channel electrode in the channel electrode position map of the target user, and obtain the spatial weight matrix.
[0085] The dataset weighting unit is used to weight low-quality EEG datasets using a spatial weight matrix to obtain weighted EEG datasets;
[0086] The dataset optimization unit is used to optimize low-quality EEG datasets based on the spatial weight matrix, the weighted EEG dataset, and the approximate entropy value of at least one EEG signal matrix to obtain candidate EEG datasets.
[0087] Optional, dataset optimization unit, specifically used for:
[0088] Based on the improved autoencoder, latent features are extracted from the weighted EEG dataset to obtain the latent feature set of the weighted EEG dataset; wherein, the improved autoencoder is trained on the weighted EEG dataset; the loss function of the improved autoencoder is determined based on the approximate entropy value of at least one EEG signal matrix;
[0089] A linear regression algorithm is used to perform linear regression mapping on the latent feature set to obtain the scaling factor of the latent feature set;
[0090] Based on the spatial weight matrix and scaling factor, the low-quality EEG dataset is weighted to obtain candidate EEG datasets.
[0091] Optionally, the dataset partitioning module 330 includes:
[0092] An approximate entropy clustering unit is used to perform cluster analysis on the approximate entropy values of at least one EEG signal matrix based on a competitive learning network, and to obtain the approximate entropy threshold of at least one EEG signal matrix.
[0093] The dataset partitioning unit is used to filter high-quality and low-quality EEG datasets from the initial EEG dataset based on an approximate entropy threshold and the approximate entropy value of at least one EEG signal matrix.
[0094] Optional, dataset partitioning units, specifically used for:
[0095] Numerical comparison is performed between the approximate entropy threshold and the approximate entropy value of at least one EEG signal matrix to obtain the comparison result of the approximate entropy value of at least one EEG signal matrix;
[0096] Based on the comparison results of approximate entropy values, at least one EEG signal matrix in the initial EEG dataset that satisfies the low-quality condition of the dataset is identified as a low-quality EEG dataset.
[0097] Based on the comparison of approximate entropy values, at least one EEG signal matrix in the initial EEG dataset that does not meet the low-quality condition of the dataset is identified as a high-quality EEG dataset.
[0098] Optional, the signal preprocessing module 310 is specifically used for:
[0099] The initial EEG signal of the target user is filtered by a bandpass filter and a notch filter to obtain candidate EEG signals.
[0100] Based on a preset time window, candidate EEG signals are divided into segments of fixed time length to obtain an initial EEG dataset.
[0101] The data optimization device for EEG signal classification provided in this application can execute the data optimization method for EEG signal classification provided in any embodiment of this application, and has the corresponding functional modules and beneficial effects for executing each data optimization method for EEG signal classification.
[0102] According to embodiments of this application, this application also provides an electronic device, a readable storage medium, and a computer program product.
[0103] Example 4
[0104] Figure 4 This is a schematic diagram of the structure of an electronic device 410 implementing the data optimization method for EEG signal classification according to embodiments of this application. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present application described and / or claimed herein.
[0105] like Figure 4 As shown, the electronic device 410 includes at least one processor 411 and a memory, such as a read-only memory (ROM) 412 or a random access memory (RAM) 413, communicatively connected to the at least one processor 411. The memory stores computer programs executable by the at least one processor. The processor 411 can perform various appropriate actions and processes based on the computer program stored in the ROM 412 or loaded from storage unit 418 into the RAM 413. The RAM 413 may also store various programs and data required for the operation of the electronic device 410. The processor 411, ROM 412, and RAM 413 are interconnected via a bus 414. An input / output (I / O) interface 415 is also connected to the bus 414.
[0106] Multiple components in electronic device 410 are connected to I / O interface 415, including: input unit 416, such as keyboard, mouse, etc.; output unit 417, such as various types of displays, speakers, etc.; storage unit 418, such as disk, optical disk, etc.; and communication unit 419, such as network card, modem, wireless transceiver, etc. Communication unit 419 allows electronic device 410 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.
[0107] Processor 411 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 411 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 411 performs the various methods and processes described above, such as data optimization methods for electroencephalogram (EEG) signal classification.
[0108] In some embodiments, the data optimization method for EEG signal classification may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 418. In some embodiments, part or all of the computer program may be loaded and / or mounted on electronic device 410 via ROM 412 and / or communication unit 419. When the computer program is loaded into RAM 413 and executed by processor 411, one or more steps of the data optimization method for EEG signal classification described above may be performed. Alternatively, in other embodiments, processor 411 may be configured for the data optimization method for EEG signal classification by any other suitable means (e.g., by means of firmware).
[0109] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.
[0110] Computer programs used to implement the methods of this application may be written in any combination of one or more programming languages. These computer programs may be provided to the processor of a general-purpose computer, a special-purpose computer, or other programmable data optimization device for EEG signal classification, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The computer programs may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.
[0111] In the context of this application, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.
[0112] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).
[0113] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.
[0114] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.
[0115] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this application can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this application can be achieved, and this is not limited herein.
[0116] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.
Claims
1. A data optimization method for classifying electroencephalogram (EEG) signals, characterized in that, include: The initial EEG signal of the target user is filtered and segmented to obtain an initial EEG dataset; wherein, the initial EEG signal is obtained by detecting the user's EEG signal through at least one channel electrode; the initial EEG dataset includes at least one EEG signal matrix; Approximate entropy calculation is performed on the at least one EEG signal matrix to obtain the approximate entropy value of the at least one EEG signal matrix; Based on the approximate entropy value of the at least one EEG signal matrix, high-quality EEG datasets and low-quality EEG datasets are selected from the initial EEG dataset. Based on the approximate entropy value of the at least one EEG signal matrix and the channel electrode location map of the target user, the low-quality EEG dataset is optimized to obtain a candidate EEG dataset; wherein, the channel electrode location map is used to characterize the location distribution of the at least one channel electrode; The high-quality EEG dataset is merged with the candidate EEG dataset to obtain the target EEG dataset for EEG signal classification.
2. The method according to claim 1, characterized in that, Based on the approximate entropy value of the at least one EEG signal matrix and the channel electrode location map of the target user, the low-quality EEG dataset is optimized to obtain a candidate EEG dataset, including: The spatial weight matrix is obtained by calculating the relative great circle distance of the current position data of at least one channel electrode in the channel electrode position map of the target user. The low-quality EEG dataset is weighted using the spatial weight matrix to obtain a weighted EEG dataset. Based on the spatial weight matrix, the weighted EEG dataset, and the approximate entropy value of the at least one EEG signal matrix, the low-quality EEG dataset is optimized to obtain a candidate EEG dataset.
3. The method according to claim 2, characterized in that, The step of optimizing the low-quality EEG dataset based on the spatial weight matrix, the weighted EEG dataset, and the approximate entropy value of at least one EEG signal matrix to obtain a candidate EEG dataset includes: Based on the improved autoencoder, latent features are extracted from the weighted EEG dataset to obtain the latent feature set of the weighted EEG dataset; wherein, the improved autoencoder is trained based on the weighted EEG dataset; the loss function of the improved autoencoder is determined based on the approximate entropy value of the at least one EEG signal matrix; A linear regression algorithm is used to perform linear regression mapping on the latent feature set to obtain the scaling factor of the latent feature set; The low-quality EEG dataset is weighted according to the spatial weight matrix and the scaling factor to obtain a candidate EEG dataset.
4. The method according to claim 1, characterized in that, The step of filtering high-quality and low-quality EEG datasets from the initial EEG dataset based on the approximate entropy value of the at least one EEG signal matrix includes: Based on a competitive learning network, cluster analysis is performed on the approximate entropy values of the at least one EEG signal matrix to obtain the approximate entropy threshold of the at least one EEG signal matrix. Based on the approximate entropy threshold and the approximate entropy value of the at least one EEG signal matrix, high-quality EEG datasets and low-quality EEG datasets are selected from the initial EEG dataset.
5. The method according to claim 4, characterized in that, The step of filtering high-quality and low-quality EEG datasets from the initial EEG dataset based on the approximate entropy threshold and the approximate entropy value of the at least one EEG signal matrix includes: The approximate entropy threshold and the approximate entropy value of the at least one EEG signal matrix are numerically compared to obtain the comparison result of the approximate entropy value of the at least one EEG signal matrix; Based on the comparison results of the approximate entropy values, at least one EEG signal matrix in the initial EEG dataset that satisfies the low-quality dataset condition is determined as a low-quality EEG dataset. Based on the comparison results of the approximate entropy values, at least one EEG signal matrix in the initial EEG dataset that does not meet the low-quality condition of the dataset is determined as a high-quality EEG dataset.
6. The method according to claim 1, characterized in that, The initial EEG signal of the target user is filtered and segmented to obtain the initial EEG dataset, including: The initial EEG signal of the target user is filtered by a bandpass filter and a notch filter to obtain candidate EEG signals. Based on a preset time window, the candidate EEG signals are divided into segments of fixed time length to obtain an initial EEG dataset.
7. A data optimization device for classifying electroencephalogram (EEG) signals, characterized in that, include: The signal preprocessing module is used to filter and segment the initial EEG signal of the target user to obtain an initial EEG dataset; wherein, the initial EEG signal is obtained by detecting the user's EEG signal through at least one channel electrode; the initial EEG dataset includes at least one EEG signal matrix; An approximate entropy calculation module is used to perform approximate entropy calculation on the at least one EEG signal matrix to obtain the approximate entropy value of the at least one EEG signal matrix; The dataset partitioning module is used to filter out high-quality EEG datasets and low-quality EEG datasets from the initial EEG dataset based on the approximate entropy value of the at least one EEG signal matrix. The dataset optimization module is used to optimize the low-quality EEG dataset based on the approximate entropy value of the at least one EEG signal matrix and the channel electrode location map of the target user to obtain a candidate EEG dataset; wherein, the channel electrode location map is used to characterize the location distribution of the at least one channel electrode; The dataset merging module is used to merge the high-quality EEG dataset with the candidate EEG dataset to obtain the target EEG dataset for EEG signal classification.
8. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the data optimization method for EEG signal classification as described in any one of claims 1-6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the data optimization method for EEG signal classification as described in any one of claims 1-6.
10. A computer program product comprising a computer program that, when executed by a processor, implements the data optimization method for classifying electroencephalogram (EEG) signals according to any one of claims 1-6.