A wireless electroencephalogram data processing method, system, terminal and storage medium

Abstract

The application discloses a wireless electroencephalogram data processing method and system, a terminal and a storage medium, wherein the method comprises the following steps: obtaining a target electroencephalogram signal, a first difference component and a second difference component generated in advance; inputting the target electroencephalogram signal after preprocessing into a deep learning electroencephalogram signal model to output the latent features of the target electroencephalogram signal; obtaining a target component according to the first difference component and the second difference component, wherein the target component is one of the first difference component and the second difference component; obtaining a tinnitus parameter; fitting the latent features, the target component and the tinnitus parameter through a linear mixing model to obtain a prediction parameter and output the prediction parameter. Through the method, the wireless electroencephalogram data and the corresponding parameters of tinnitus can be associated, so that the tinnitus parameter information can be more accurately obtained.

Description

Technical Field

[0001] This invention relates to the field of data processing, and particularly to a wireless EEG data processing method, system, terminal, and storage medium. Background Technology

[0002] Currently, information related to subjective tinnitus can be obtained from unlimited EEG data. In exploring the attention mechanism of tinnitus, the concept of attention diffusion is usually adopted. Attention can diffuse between different channels in multisensory objects, such as between vision and hearing. This is called "cross-channel attention diffusion." However, cross-channel attention diffusion becomes unbalanced in tinnitus, making it difficult to separate attention from the EEG signals corresponding to tinnitus.

[0003] However, in the current process of combining tinnitus-related data with wireless EEG data, behavioral paradigms or deep learning methods are usually used, which cannot accurately link tinnitus-related parameters, such as pure tone hearing threshold range, tinnitus frequency, and tinnitus loudness, with wireless EEG data. Therefore, it is impossible to obtain tinnitus parameter information based on wireless EEG data.

[0004] Therefore, existing technologies still need improvement and development. Summary of the Invention

[0005] The main objective of this invention is to provide a wireless EEG data processing method, system, terminal, and storage medium, aiming to solve the problem in the prior art that, in the process of combining tinnitus-related data with wireless EEG data, only behavioral paradigms or deep learning methods are typically used, which cannot accurately link the tinnitus-related parameters with the wireless EEG data, thus making it impossible to obtain tinnitus parameter information based on the wireless EEG data.

[0006] To achieve the aforementioned objective, a first aspect of the present invention provides a wireless EEG data processing method, wherein the wireless EEG data processing method includes:

[0007] Acquire a pre-generated target EEG signal, a first differential component, and a second differential component, wherein the first differential component includes the reaction rate and accuracy obtained using a visual diffusion to auditory attention paradigm, and the second differential component includes the reaction rate and accuracy obtained using an auditory diffusion to visual attention paradigm.

[0008] After preprocessing the target EEG signal, it is input into a deep learning model for extracting EEG signals, and the potential features of the target EEG signal are output.

[0009] A target component is obtained based on the first difference component and the second difference component, wherein the target component is one of the first difference component and the second difference component;

[0010] The tinnitus parameters are obtained, and the latent features are fitted with the target components and the tinnitus parameters through a linear mixture model to obtain the predicted parameters, which are then output.

[0011] Optionally, the step of preprocessing the target EEG signal and then inputting it into a deep learning model for extracting EEG signals to output the latent features of the target EEG signal includes:

[0012] The target EEG signal is preprocessed to obtain a preprocessed EEG signal;

[0013] The energy density of the preprocessed EEG signal is obtained by acquiring the energy corresponding to a preset frequency band in the preprocessed EEG signal.

[0014] The energy corresponding to the preset frequency band in the preprocessed EEG signal is input into the deep learning-based EEG signal extraction model, and the potential features of the target EEG signal are output.

[0015] Optionally, the step of preprocessing the target EEG signal to obtain a preprocessed EEG signal includes:

[0016] The target EEG signal is acquired, and the target EEG signal is processed according to the outlier removal method to obtain a first preprocessed EEG signal.

[0017] The first preprocessed EEG signal is downsampled, and the downsampled first preprocessed EEG signal is input into a band-stop filter to obtain the second preprocessed EEG signal.

[0018] The second preprocessed EEG signal is subjected to mean-degradation processing to obtain the preprocessed EEG signal.

[0019] Optionally, the step of obtaining the target component based on the first difference component and the second difference component includes:

[0020] Obtain the first data of the first difference component and the second data of the second difference component;

[0021] Compare the first data and the second data to obtain the comparison result;

[0022] Based on the comparison results, the target component is selected from the first data of the component of attention spreading from vision to hearing and the component of attention spreading from hearing to vision.

[0023] Optionally, the step of obtaining tinnitus parameters, fitting the latent features with the target components and the tinnitus parameters using a linear mixture model to obtain predicted parameters, and outputting the predicted parameters includes:

[0024] Obtain tinnitus parameters;

[0025] The latent features, target components, and tinnitus parameters are fitted using the linear mixture model to obtain the fitting parameters corresponding to each tinnitus parameter;

[0026] Compare the correlation coefficients of the fitted parameters corresponding to each tinnitus parameter, and select the fitted parameter with the largest correlation coefficient as the current prediction parameter;

[0027] A preset number of prediction parameters are obtained based on the current prediction parameters corresponding to each tinnitus parameter, and then output.

[0028] A second aspect of the present invention provides a wireless EEG data processing system, wherein the wireless EEG data processing system comprises:

[0029] The data acquisition module is used to acquire a pre-generated target EEG signal, a first differential component, and a second differential component, wherein the first differential component includes the reaction rate and accuracy obtained using the visual diffusion to auditory attention paradigm, and the second differential component includes the reaction rate and accuracy obtained using the auditory diffusion to visual attention paradigm.

[0030] The feature acquisition module is used to preprocess the target EEG signal, input it into a deep learning model for extracting EEG signals, and output the potential features of the target EEG signal.

[0031] A component acquisition module is used to acquire a target component based on the first differential component and the second differential component, wherein the target component is one of the first differential component and the second differential component;

[0032] The result output module is used to obtain tinnitus parameters, fit the latent features with the target components and the tinnitus parameters through a linear mixture model to obtain predicted parameters, and output them.

[0033] The feature acquisition module includes:

[0034] The preprocessing unit is used to preprocess the target EEG signal to obtain a preprocessed EEG signal;

[0035] An energy acquisition unit is used to acquire the energy density of the preprocessed EEG signal to obtain the energy corresponding to a preset frequency band in the preprocessed EEG signal.

[0036] The latent feature generation unit is used to input the energy corresponding to a preset frequency band in the preprocessed EEG signal into the deep learning-based EEG signal extraction model and output the latent features of the target EEG signal.

[0037] The component acquisition module includes:

[0038] A component data acquisition unit is used to acquire first data of the first differential component and second data of the second differential component;

[0039] A comparison unit is used to compare the first data and the second data to obtain a comparison result;

[0040] A target component generation unit is configured to select the target component from first data of the components in which attention spreads from vision to hearing and the components in which attention spreads from hearing to vision, based on the comparison result.

[0041] The result output module includes:

[0042] The parameter acquisition unit is used to acquire tinnitus parameters;

[0043] Multiple fitting parameter generation units are used to fit the latent features with the target components and the tinnitus parameters through the linear mixture model to obtain fitting parameters corresponding to each tinnitus parameter;

[0044] The current prediction parameter acquisition unit is used to compare the correlation coefficients of the fitted parameters corresponding to each tinnitus parameter and select the fitted parameter with the largest correlation coefficient as the current prediction parameter.

[0045] The prediction parameter acquisition unit is used to obtain a preset number of prediction parameters based on the current prediction parameters corresponding to each tinnitus parameter, and then output them.

[0046] A third aspect of the present invention provides a terminal, the terminal including a memory, a processor, and a wireless EEG data processing program stored in the memory and executable on the processor, wherein the wireless EEG data processing program, when executed by the processor, implements the steps of any one of the wireless EEG data processing methods.

[0047] As can be seen from the above, in the present invention, a pre-generated target EEG signal, a first differential component, and a second differential component are obtained. The first differential component includes the reaction rate and accuracy obtained using a visual-to-auditory attention paradigm, and the second differential component includes the reaction rate and accuracy obtained using an auditory-to-visual attention paradigm. The target EEG signal is preprocessed and then input into a deep learning model for extracting EEG signals, outputting the latent features of the target EEG signal. A target component is obtained based on the first and second differential components, wherein the target component is one of the first and second differential components. Tinnitus parameters are obtained, and the latent features are fitted with the target component and the tinnitus parameters using a linear mixture model to obtain prediction parameters, which are then output.

[0048] Compared to existing technologies, this invention addresses the problem that current methods for combining tinnitus-related data with wireless EEG data typically rely solely on behavioral paradigms or deep learning, failing to accurately link tinnitus-related parameters with wireless EEG data and thus hindering the acquisition of tinnitus parameter information from the wireless EEG data. This invention utilizes a deep learning-based EEG signal extraction model to process preprocessed target EEG signals, combines spatially selective visual attention and spatially selective auditory attention paradigms to obtain target components, and finally combines tinnitus parameters to output predicted parameters. This approach leverages behavioral paradigms and employs deep learning to process EEG signals, linking wireless EEG data with tinnitus-related parameters. Furthermore, the predicted parameters are the result of fitting wireless EEG data and tinnitus-related parameters, allowing for more accurate acquisition of tinnitus-related parameter information. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 This is a flowchart illustrating a wireless EEG data processing method provided in an embodiment of the present invention;

[0051] Figure 2 This is a schematic diagram of the overall model framework of the wireless EEG data processing method provided in this embodiment of the invention;

[0052] Figure 3 This is a schematic diagram illustrating the acquisition of components of attention diffusing from vision to hearing, provided by an embodiment of the present invention;

[0053] Figure 4 This is a schematic diagram illustrating the component acquisition of attention diffusion from auditory to visual perception provided in an embodiment of the present invention;

[0054] Figure 5 This is a schematic diagram of the constituent modules of a wireless EEG data processing system provided in an embodiment of the present invention;

[0055] Figure 6 This is a block diagram illustrating the internal functional principle of the feature acquisition module of the wireless EEG data processing system provided in this embodiment of the invention.

[0056] Figure 7 This is a block diagram illustrating the internal functional principle of the component acquisition module of the wireless EEG data processing system provided in this embodiment of the invention.

[0057] Figure 8This is a block diagram illustrating the internal functional principle of the result output module of the wireless EEG data processing system provided in this embodiment of the invention.

[0058] Figure 9 This is a block diagram illustrating the internal structure of a terminal provided in an embodiment of the present invention. Detailed Implementation

[0059] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will understand that the invention can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0060] It should be understood that, when used in this specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0061] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0062] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0063] As used in this specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to classification." Similarly, the phrases "if determined" or "if classified to [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once classified to [the described condition or event]," or "in response to classification to [the described condition or event]."

[0064] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0065] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0066] Currently, information related to subjective tinnitus can be obtained from unlimited EEG data. In exploring the attentional mechanisms of tinnitus, the concept of attentional spread is often used. Attention is the cognitive process by which an individual selects specific information from numerous external stimuli for processing, ignoring other information. Early research on visual attention based on objects found that visual attention can spread between different features within a visual object, a phenomenon known as "attentional spread" within the visual object. Attention can also spread between different channels within multisensory objects, such as between vision and hearing; this is called "cross-channel attentional spread." However, cross-channel attentional spread becomes unbalanced in tinnitus, making it difficult to separate attention from the EEG signals corresponding to tinnitus.

[0067] However, in the current process of combining tinnitus-related data with wireless EEG data, behavioral paradigms or deep learning methods are usually used, which cannot accurately link tinnitus-related parameters, such as pure tone hearing threshold range, tinnitus frequency, and tinnitus loudness, with wireless EEG data. Therefore, it is impossible to obtain tinnitus parameter information based on wireless EEG data.

[0068] To address at least one of the aforementioned problems, the present invention provides a wireless EEG data processing method, system, terminal, and storage medium. Specifically, the method involves acquiring a pre-generated target EEG signal, a first differential component, and a second differential component. The first differential component includes reaction rate and accuracy obtained using a visual-to-auditory attention paradigm, and the second differential component includes reaction rate and accuracy obtained using an auditory-to-visual attention paradigm. The target EEG signal is preprocessed and then input into a deep learning model for extracting EEG signals, outputting latent features of the target EEG signal. A target component is obtained based on the first and second differential components, wherein the target component is one of the first and second differential components. Tinnitus parameters are acquired, and the latent features are fitted with the target component and the tinnitus parameters using a linear mixture model to obtain predicted parameters, which are then output.

[0069] This invention extracts and processes preprocessed target EEG signals using a deep learning model, combines visual-to-auditory attention paradigms and auditory-to-visual attention paradigms to obtain target components, and finally combines tinnitus parameters to output predicted parameters. This utilizes behavioral paradigms and employs deep learning to process EEG signals, linking wireless EEG data with tinnitus-related parameters. The predicted parameters are the result of fitting wireless EEG data and tinnitus-related parameters, allowing for more accurate acquisition of tinnitus parameter information.

[0070] Exemplary methods

[0071] like Figure 1 As shown, this embodiment of the invention provides a wireless EEG data processing method. Specifically, the wireless EEG data processing method includes the following steps:

[0072] Step S100: Acquire the pre-generated target EEG signal, a first differential component, and a second differential component, wherein the first differential component includes the reaction rate and accuracy obtained using the visual diffusion to auditory attention paradigm, and the second differential component includes the reaction rate and accuracy obtained using the auditory diffusion to visual attention paradigm.

[0073] It should be noted that the target EEG signal is a signal pre-acquired by a wireless EEG acquisition device. The first differential component, namely the component of attention spreading from vision to hearing, and the second differential component, namely the component of attention spreading from hearing to vision, are also pre-acquired data. The components of attention spreading from vision to hearing and from hearing to vision both contain corresponding reaction time and accuracy. Furthermore, the components of attention spreading from vision to hearing and from hearing to vision are represented by a 128-dimensional sequence, which contains reaction time information and accuracy information.

[0074] Specifically, the component of attention diffusing from visual to auditory is information acquired from the transchannel attentional diffusion phenomenon, where attentional resources shift from the visual channel to the auditory channel. This includes the reaction rate and accuracy obtained using the visual-to-auditory attention paradigm. Specifically, the spatially selective visual attention paradigm, i.e., the visual-to-auditory attention paradigm, presents visual stimuli from the left and right sides, such as... Figure 2As shown, auditory stimuli are presented simultaneously in the central position. Subjects focus only on visual stimuli from one side (e.g., the right side) and respond by pressing a key on the visual target appearing at that position, ignoring both auditory and unattended visual stimuli. Specifically, in the data analysis, the difference waveforms between the simultaneously presented auditory and visual stimuli at the attended and unattended positions (AV-V) were first calculated to extract the effect of auditory components on the ERP (event-related potential) under multi-sensory stimulation conditions. Then, the auditory components extracted under visual attention and unattended conditions were compared. By obtaining the subjects' reaction time and accuracy to the corresponding stimuli, the component of attention diffusing from vision to hearing can be obtained.

[0075] The component of attention diffusion from auditory to visual is information pre-observed from the transchannel attention diffusion phenomenon, where attentional resources shift from the auditory channel to the visual channel. This includes the reaction rate and accuracy obtained using the auditory-to-visual attention paradigm. Specifically, it employs a spatially selective auditory attention paradigm, i.e., the auditory-to-visual attention paradigm, such as... Figure 3 As shown, a fixation point is displayed in the center of the screen. After a time interval of 950–1050 milliseconds, an auditory target stimulus is presented from the left / right speakers for 33 milliseconds, while a visual stimulus is presented at the bottom of the screen (vertical distance of 6°). Subjects ignore the visual stimulus presented simultaneously with the central auditory stimulus or presented alone, focusing only on the auditory stimulus on one side of the screen and ignoring any stimuli on the other side. They only react by pressing buttons when the auditory and visual / auditory target stimuli appear at the attentional location. Specifically, in the ERP data analysis, the difference waveforms between the visual / auditory stimuli and the auditory stimuli presented simultaneously at the attended and unattended locations, i.e., Attended (AV-V) and Unattended (AV-V), were first calculated to extract the effect of the visual component-induced ERP under multi-sensory stimulus conditions. Then, the visual components extracted under auditory attention and unattended conditions were compared. Obtaining the subject's reaction time and accuracy for the corresponding stimulus reveals the component of attention diffusion from auditory to visual.

[0076] Step S200: After preprocessing the target EEG signal, input it into the deep learning model for extracting EEG signals, and output the potential features of the target EEG signal.

[0077] Specifically, after obtaining the corresponding target EEG signal, the target EEG signal is preprocessed to facilitate signal processing by the deep learning EEG signal extraction model. Then, the preprocessed data is input into the deep learning EEG signal extraction model to output the potential features of the target EEG signal.

[0078] Furthermore, the step of preprocessing the target EEG signal and inputting it into a deep learning model for extracting EEG signals, and then outputting the latent features of the target EEG signal, includes:

[0079] The target EEG signal is preprocessed to obtain a preprocessed EEG signal;

[0080] The energy density of the preprocessed EEG signal is obtained by acquiring the energy corresponding to a preset frequency band in the preprocessed EEG signal.

[0081] The energy corresponding to the preset frequency band in the preprocessed EEG signal is input into the deep learning-based EEG signal extraction model, and the potential features of the target EEG signal are output.

[0082] The step of preprocessing the target EEG signal to obtain a preprocessed EEG signal includes:

[0083] The target EEG signal is acquired, and the target EEG signal is processed according to the outlier removal method to obtain a first preprocessed EEG signal.

[0084] The first preprocessed EEG signal is downsampled, and the downsampled first preprocessed EEG signal is input into a band-stop filter to obtain the second preprocessed EEG signal.

[0085] The second preprocessed EEG signal is subjected to mean-degradation processing to obtain the preprocessed EEG signal.

[0086] For the acquired target EEG signal, it is first processed using outlier removal to obtain a first preprocessed EEG signal. In this outlier removal process, data points outside the six standard deviations are removed to eliminate spurious differences. The first preprocessed EEG signal is then downsampled according to a set signal frequency to reduce its dimensionality. In one embodiment, the first preprocessed EEG signals are all 2000Hz or 1000Hz signals, and downsampling uniformly reduces their frequency to 500Hz. A high-pass filter and a band-stop filter are then used to remove 50Hz power frequency interference from the downsampled data to obtain a second preprocessed EEG signal. The second preprocessed EEG signal is then subjected to mean averaging to obtain the final preprocessed EEG signal. In one embodiment, mean averaging can be implemented using the `detrend` method in Matlab.

[0087] The step of acquiring the energy density of the preprocessed EEG signal to obtain the energy corresponding to a preset frequency band in the preprocessed EEG signal includes:

[0088] The wavelet of the preprocessed EEG signal is obtained by performing wavelet transformation on the preprocessed EEG signal.

[0089] The wavelet is convolved to obtain a continuous wavelet transform, and the energy corresponding to the preset frequency band in the preprocessed EEG signal is obtained from the continuous wavelet transform according to the preset frequency band.

[0090] It should be noted that after preprocessing the target EEG signal, the preprocessed EEG signal will enter the deep learning module for extracting EEG signal features. To better meet the needs of the recurrent neural network model and extract richer information, time-frequency features are chosen as the feature representation of the EEG signal. Specifically, Morlet mother wavelet is used to perform wavelet transform on the preprocessed EEG signal. The wavelets of the preprocessed EEG signal are convolved to obtain a continuous wavelet transform. Based on a preset frequency band, the energy corresponding to the preset frequency band in the preprocessed EEG signal is obtained from the continuous wavelet transform.

[0091] The wavelet transform formula is as follows:

[0092]

[0093] Where ω is the center frequency of the Morlet wavelet, σ is the parameter controlling the Gaussian kernel scale, t is the time point, and ψ is the wavelet coefficient calculated at each time point. The continuous wavelet transform of the first preprocessed EEG signal is the convolution of the signal with the wavelet basis function, and the energy output from the wavelet convolution is expressed by the following formula:

[0094]

[0095] That is, a wavelet basis ψ(1 / α(τ-t)) is established by translating and shifting ψ(t) in the time domain, where α is the scaling factor controlling the wavelet's scale in the time domain, and τ is the translation factor controlling the wavelet's shift in the time domain. By scaling α, the first preprocessed EEG signal can be analyzed over a wider time range. X is the energy obtained by the wavelet convolution. Squaring X yields the energy corresponding to the preset frequency band in the preprocessed EEG signal. At the same time, the frequency band energy in the EEG signal is expressed as amplitude changes in different frequency bands.

[0096] The frequency energy in EEG signals manifests as amplitude variations across different frequency bands, including delta (1-3Hz), theta (3-8Hz), alpha (8-14Hz), beta (14-30Hz), gamma (30-100Hz), and high gamma (100-150Hz). To extract feature information from different frequency bands, the energy corresponding to preset frequency bands in the preprocessed EEG signal is obtained from continuous wavelet transform and input into a deep learning model for extracting EEG signals. In this embodiment of the invention, to preserve the energy corresponding to the first preprocessed EEG signal in the power preprocessing, the power spectral density of each 4-second interval is divided into four equal parts, ranging from 0-150Hz. The corresponding frequency bands include delta (1-3Hz), theta (3-8Hz), alpha (8-14Hz), beta (14-30Hz), gamma (30-100Hz), and high gamma (100-150Hz).

[0097] The step of inputting the energy corresponding to the preset frequency band in the preprocessed EEG signal into the deep learning-based EEG signal extraction model and outputting the latent features of the target EEG signal includes:

[0098] The energy corresponding to the preset frequency band in the preprocessed EEG signal is input into the deep learning-based EEG signal extraction model;

[0099] The deep learning model for extracting EEG signals processes the energy corresponding to the preset frequency band through a fully connected layer and the Tanh activation function to obtain the potential features of the target EEG signal.

[0100] It should be noted that the deep learning model for extracting EEG signals is an autoencoder model, which consists of fully connected layers, a Tanh activation function, and a Sigmoid function, as detailed below. Figure 4 As shown, the energy corresponding to the preset frequency band in the preprocessed EEG signal is input into the encoder part. After passing through multiple fully connected layers and Tanh, the output data is standardized using the Sigmoid function.

[0101] Further description reveals that the entire deep learning model for extracting EEG signals primarily consists of fully connected layers, with nonlinear transformations achieved through the addition of Tanh activation functions. During training, mean squared error is chosen as the loss function to calculate the distance between input and output features, and the PyTorch deep learning framework is used to construct and train the model. The deep learning model structure comprises an encoder and a decoder. The encoder reduces the dimensionality of the input, while the decoder restores the dimensionality of the reduced data to its original dimensions. The deep learning model is divided into three layers: an input layer, hidden layers, and an output layer, with the hidden layers consisting of multiple fully connected layers. During training, PyTorch is used as the network training framework, an RTX 2070 Super GPU is used for training, the batch size is 18, and Adam is used as the optimizer. The deep learning model consists of five fully connected layers with decreasing input dimensions, and the latent features of the target EEG signal output by the encoder are saved after 50 batch training iterations. The potential features of the target EEG signal include visual-auditory EEG potential representations and auditory-visual EEG potential representations, each of which is represented by a 128-bit sequence.

[0102] Step S300: Obtain the target component based on the first difference component and the second difference component, wherein the target component is one of the first difference component and the second difference component.

[0103] It should be noted that both the component of attention spreading from vision to hearing (the first difference component) and the component of attention spreading from hearing to vision (the second difference component) contain reaction time information and accuracy information. By comparing the reaction time information and accuracy information in the component of attention spreading from vision to hearing and the component of attention spreading from hearing to vision, the target component can be obtained.

[0104] Furthermore, the step of obtaining the target component based on the first difference component and the second difference component includes:

[0105] Obtain the first data of the first difference component and the second data of the second difference component;

[0106] Compare the first data and the second data to obtain the comparison result;

[0107] Based on the comparison results, the target component is selected from the first data of the component of attention spreading from vision to hearing and the component of attention spreading from hearing to vision.

[0108] Specifically, the first data represents the reaction rate and accuracy of the first differential component, and the second data represents the reaction rate and accuracy of the second differential component. A higher reaction rate corresponds to a higher accuracy. Therefore, by comparing the reaction rates or accuracy of the first and second data, the component with the better reaction rate or accuracy is selected as the target component. For example, if the reaction rate of the first differential component is faster than that of the second differential component, then the first differential component is selected as the target component.

[0109] Step S400: Obtain tinnitus parameters, fit the latent features with the target components and the tinnitus parameters using a linear mixture model to obtain prediction parameters, and output them.

[0110] It should be noted that the tinnitus parameters include four aspects: the Tinnitus Disability Assessment Scale (THI), pure-tone hearing threshold range, tinnitus frequency, and tinnitus loudness. Among these, the THI score, pure-tone hearing threshold range, tinnitus frequency, and tinnitus loudness are all pre-acquired data, i.e., the tinnitus parameters. The THI uses the Newman rating scale, and the tinnitus frequency is pre-measured using the two-tone selection method. Tinnitus loudness is pre-matched using pure tones or narrowband noise, with the intensity of the test tone gradually increased in 1 dB increments until the intensity of the test tone equals the tinnitus sound. The fitting using a linear mixed model is employed.

[0111] Furthermore, the step of obtaining tinnitus parameters, fitting the latent features with the target components and the tinnitus parameters using a linear mixture model to obtain predicted parameters, and outputting the predicted parameters includes:

[0112] Obtain tinnitus parameters;

[0113] The latent features, target components, and tinnitus parameters are fitted using the linear mixture model to obtain the fitting parameters corresponding to each tinnitus parameter;

[0114] Compare the correlation coefficients of the fitted parameters corresponding to each tinnitus parameter, and select the fitted parameter with the largest correlation coefficient as the current prediction parameter;

[0115] A preset number of prediction parameters are obtained based on the current prediction parameters corresponding to each tinnitus parameter, and then output.

[0116] After obtaining the tinnitus parameters, since the latent features include potential visual-auditory EEG representations and potential auditory-visual EEG representations, and the tinnitus parameters include the Tinnitus Disability Assessment Scale, pure-tone hearing threshold range, tinnitus frequency, and tinnitus loudness, a linear mixture model is used for processing. For each feature in the tinnitus parameters, a linear mixture is performed with the target component and the tinnitus parameter to obtain the fitting parameters corresponding to each tinnitus parameter. For example, regarding tinnitus loudness as a parameter of tinnitus, the following parameters are used: y-value is the latent visual-auditory EEG representation, x1 is the accuracy of the target component, x2 is the reaction speed, and x3 is the tinnitus loudness. A linear mixture model is used to perform linear mixing to obtain the first fitting parameter. The second fitting parameter is obtained by using the latent visual-auditory EEG representation as the y-value, x1 is the accuracy of the target component, x2 is the reaction speed, and x3 is the tinnitus loudness. The third fitting parameter is obtained by using the combined latent visual-auditory EEG representation and the latent auditory-visual EEG representation as the y-value, x1 is the accuracy of the target component, x2 is the reaction speed, and x3 is the tinnitus loudness. The first, second, and third fitting parameters are the fitting parameters corresponding to the tinnitus loudness parameter. The fitting parameters obtained by linear mixing of the linear mixture model can be used to obtain the magnitude of the correlation coefficient. The fitting parameter with the largest correlation coefficient among the first, second and third fitting parameters is selected as the current prediction parameter corresponding to the tinnitus loudness parameter.

[0117] The current predicted parameter corresponding to each tinnitus parameter is obtained, which yields a preset number of predicted parameters, and these are then output. In one embodiment of this application, the preset number is 4, corresponding to the number of features contained in the tinnitus parameters.

[0118] The obtained prediction parameters include the current prediction parameters corresponding to the four features of the tinnitus parameters. Each current prediction parameter is represented as a fitting curve. Through these prediction parameters, the correlation between the EEG signal and the tinnitus parameter can be obtained, thereby obtaining accurate processing results of the EEG data. This also allows users to obtain information from the EEG data more accurately and intuitively, thus facilitating the processing of EEG signals.

[0119] As can be seen from the above, compared with existing technologies, the current methods of combining tinnitus-related data with wireless EEG data typically only utilize behavioral paradigms or deep learning, failing to accurately link tinnitus-related parameters with wireless EEG data and thus failing to obtain tinnitus parameter information from wireless EEG data. This invention addresses this problem by using deep learning to extract EEG signal models to process preprocessed target EEG signals, combining spatial selective visual attention paradigms and spatial selective auditory attention paradigms to obtain target components, and finally combining tinnitus parameters to output predicted parameters. This approach utilizes behavioral paradigms and employs deep learning to process EEG signals, linking wireless EEG data with tinnitus-related parameters. Furthermore, the predicted parameters are the result of fitting wireless EEG data and tinnitus-related parameters, allowing for more accurate acquisition of tinnitus parameter information.

[0120] Exemplary device

[0121] like Figure 5 As shown, corresponding to the aforementioned wireless EEG data processing method, this embodiment of the invention also provides a wireless EEG data processing system, the wireless EEG data processing system comprising:

[0122] The data acquisition module 51 is used to acquire a pre-generated target EEG signal, a first differential component, and a second differential component, wherein the first differential component includes the reaction rate and accuracy obtained using the visual diffusion to auditory attention paradigm, and the second differential component includes the reaction rate and accuracy obtained using the auditory diffusion to visual attention paradigm.

[0123] The feature acquisition module 52 is used to preprocess the target EEG signal, input it into a deep learning model for extracting EEG signals, and output the potential features of the target EEG signal.

[0124] The component acquisition module 53 is used to acquire a target component based on the first differential component and the second differential component, wherein the target component is one of the first differential component and the second differential component;

[0125] The result output module 54 is used to obtain tinnitus parameters, fit the latent features with the target components and the tinnitus parameters through a linear mixture model to obtain predicted parameters, and output them.

[0126] like Figure 6 As shown, the feature acquisition module includes:

[0127] The preprocessing unit is used to preprocess the target EEG signal to obtain a preprocessed EEG signal;

[0128] An energy acquisition unit is used to acquire the energy density of the preprocessed EEG signal to obtain the energy corresponding to a preset frequency band in the preprocessed EEG signal.

[0129] The latent feature generation unit is used to input the energy corresponding to a preset frequency band in the preprocessed EEG signal into the deep learning-based EEG signal extraction model and output the latent features of the target EEG signal.

[0130] like Figure 7 As shown, the component acquisition module includes:

[0131] A component data acquisition unit is used to acquire first data of the first differential component and second data of the second differential component;

[0132] A comparison unit is used to compare the first data and the second data to obtain a comparison result;

[0133] A target component generation unit is configured to select the target component from first data of components in which attention spreads from vision to hearing and components in which attention spreads from hearing to vision, based on the comparison result.

[0134] like Figure 8 As shown, the result output module includes:

[0135] The parameter acquisition unit is used to acquire tinnitus parameters;

[0136] Multiple fitting parameter generation units are used to fit the latent features with the target components and the tinnitus parameters through the linear mixture model to obtain fitting parameters corresponding to each tinnitus parameter;

[0137] The current prediction parameter acquisition unit is used to compare the correlation coefficients of the fitted parameters corresponding to each tinnitus parameter and select the fitted parameter with the largest correlation coefficient as the current prediction parameter.

[0138] The prediction parameter acquisition unit is used to obtain a preset number of prediction parameters based on the current prediction parameters corresponding to each tinnitus parameter, and then output them.

[0139] It should be noted that the specific structure and implementation of the wireless EEG data processing system and its various modules or units can be referred to the corresponding description in the method embodiments, and will not be repeated here.

[0140] It should be noted that the division of the various modules in the wireless EEG data processing system is not unique and is not intended as a specific limitation.

[0141] Based on the above embodiments, the present invention also provides a terminal, the principle block diagram of which can be as follows: Figure 9As shown. The terminal includes a processor 10, a memory 20, a network interface, and a display screen 30 connected via a system bus. In one embodiment, when the processor 10 executes the wireless EEG data processing program 40 in the memory 20, the following steps are implemented:

[0142] Acquire a pre-generated target EEG signal, a first differential component, and a second differential component, wherein the first differential component includes the reaction rate and accuracy obtained using a visual diffusion to auditory attention paradigm, and the second differential component includes the reaction rate and accuracy obtained using an auditory diffusion to visual attention paradigm.

[0143] After preprocessing the target EEG signal, it is input into a deep learning model for extracting EEG signals, and the potential features of the target EEG signal are output.

[0144] A target component is obtained based on the first difference component and the second difference component, wherein the target component is one of the first difference component and the second difference component;

[0145] The tinnitus parameters are obtained, and the latent features are fitted with the target components and the tinnitus parameters through a linear mixture model to obtain the predicted parameters, which are then output.

[0146] Optionally, the step of preprocessing the target EEG signal and then inputting it into a deep learning model for extracting EEG signals to output the latent features of the target EEG signal includes:

[0147] The target EEG signal is preprocessed to obtain a preprocessed EEG signal;

[0148] The energy density of the preprocessed EEG signal is obtained by acquiring the energy corresponding to a preset frequency band in the preprocessed EEG signal.

[0149] The energy corresponding to the preset frequency band in the preprocessed EEG signal is input into the deep learning-based EEG signal extraction model, and the potential features of the target EEG signal are output.

[0150] Optionally, the step of preprocessing the target EEG signal to obtain a preprocessed EEG signal includes:

[0151] The target EEG signal is acquired, and the target EEG signal is processed according to the outlier removal method to obtain a first preprocessed EEG signal.

[0152] The first preprocessed EEG signal is downsampled, and the downsampled first preprocessed EEG signal is input into a band-stop filter to obtain the second preprocessed EEG signal.

[0153] The second preprocessed EEG signal is subjected to mean-degradation processing to obtain the preprocessed EEG signal.

[0154] Optionally, the step of obtaining the target component based on the first difference component and the second difference component includes:

[0155] Obtain the first data of the first difference component and the second data of the second difference component;

[0156] Compare the first data and the second data to obtain the comparison result;

[0157] Based on the comparison results, the target component is selected from the first data of the component of attention spreading from vision to hearing and the component of attention spreading from hearing to vision.

[0158] Optionally, the step of obtaining tinnitus parameters, fitting the latent features with the target components and the tinnitus parameters using a linear mixture model to obtain predicted parameters, and outputting the predicted parameters includes:

[0159] Obtain tinnitus parameters;

[0160] The latent features, target components, and tinnitus parameters are fitted using the linear mixture model to obtain the fitting parameters corresponding to each tinnitus parameter;

[0161] Compare the correlation coefficients of the fitted parameters corresponding to each tinnitus parameter, and select the fitted parameter with the largest correlation coefficient as the current prediction parameter;

[0162] A preset number of prediction parameters are obtained based on the current prediction parameters corresponding to each tinnitus parameter, and then output.

[0163] This invention also provides a computer-readable storage medium storing a wireless EEG data processing program. When executed by a processor, the wireless EEG data processing program implements the steps of any wireless EEG data processing method provided in this invention.

[0164] It should be understood that the sequence number of each step in the embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0165] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the division of the functional units and modules is only described as an example. In practical applications, the functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this invention. The specific working process of the units and modules in the device can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0166] In the embodiments described, each embodiment has its own emphasis. For parts not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0167] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0168] In the embodiments provided by this invention, it should be understood that the disclosed system / terminal device and method can be implemented in other ways. For example, the system / terminal device embodiments described above are merely illustrative. For instance, the division of modules or units is merely a logical functional division, and in actual implementation, it can be divided in other ways. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0169] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content contained in the computer-readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction.

[0170] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not mean that the essence of the corresponding technical solutions deviates from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A wireless EEG data processing method, characterized in that, The wireless EEG data processing method includes: Acquire a pre-generated target EEG signal, a first differential component, and a second differential component, wherein the first differential component includes the reaction rate and accuracy obtained using a visual diffusion to auditory attention paradigm, and the second differential component includes the reaction rate and accuracy obtained using an auditory diffusion to visual attention paradigm. After preprocessing the target EEG signal, it is input into a deep learning model for extracting EEG signals, and the potential features of the target EEG signal are output. The potential features of the target EEG signal include: visual-auditory EEG potential representation and auditory-visual EEG potential representation. A target component is obtained based on the first difference component and the second difference component, wherein the target component is one of the first difference component and the second difference component; Obtain tinnitus parameters, fit the latent features with the target components and the tinnitus parameters using a linear mixture model to obtain predicted parameters, and output them; The step of obtaining the target component based on the first difference component and the second difference component includes: Obtain the first data of the first difference component and the second data of the second difference component; Compare the first data and the second data to obtain the comparison result; Based on the comparison results, the target component is selected from the first difference component of attention diffusing from vision to hearing and the second difference component of attention diffusing from hearing to vision; The steps of obtaining tinnitus parameters, fitting the latent features with the target components and the tinnitus parameters using a linear mixture model to obtain predicted parameters, and outputting the predicted parameters include: Obtain tinnitus parameters, including: Tinnitus Disability Assessment Scale, Pure Tone Audiometry Range, Tinnitus Frequency, and Tinnitus Loudness; The latent features, target components, and tinnitus parameters are fitted using the linear mixture model to obtain the fitting parameters corresponding to each tinnitus parameter; Compare the correlation coefficients of the fitted parameters corresponding to each tinnitus parameter, and select the fitted parameter with the largest correlation coefficient as the current prediction parameter; A preset number of prediction parameters are obtained based on the current prediction parameters corresponding to each tinnitus parameter, and then output. The first differential component is obtained by calculating the differential waves between the visual and auditory stimuli presented simultaneously at the attention and non-attention positions and the visual stimuli, respectively, to extract the effect of the ERP induced by the auditory component under multi-sensory stimulation conditions, and then comparing the auditory components extracted under visual attention and visual non-attention conditions. The second difference component is obtained by calculating the difference wave between visual and auditory stimuli presented simultaneously at attentional and non-attentional locations and the auditory stimuli, respectively, to extract the effect of the ERP induced by the visual component under multi-sensory stimulation conditions, and then comparing the visual components extracted under auditory attention and auditory non-attention conditions.

2. The wireless EEG data processing method according to claim 1, characterized in that, The step of preprocessing the target EEG signal, inputting it into a deep learning model for extracting EEG signals, and outputting the latent features of the target EEG signal includes: The target EEG signal is preprocessed to obtain a preprocessed EEG signal; The preprocessed EEG signal is subjected to energy density acquisition processing to obtain the energy corresponding to a preset frequency band in the preprocessed EEG signal. The energy corresponding to the preset frequency band in the preprocessed EEG signal is input into the deep learning-based EEG signal extraction model, and the potential features of the target EEG signal are output.

3. The wireless EEG data processing method according to claim 2, characterized in that, The step of preprocessing the target EEG signal to obtain a preprocessed EEG signal includes: The target EEG signal is acquired, and the target EEG signal is processed according to the outlier removal method to obtain a first preprocessed EEG signal; The first preprocessed EEG signal is downsampled, and the downsampled first preprocessed EEG signal is input into a band-stop filter to obtain the second preprocessed EEG signal. The second preprocessed EEG signal is subjected to mean-down processing to obtain the preprocessed EEG signal.

4. A wireless EEG data processing system, characterized in that, The wireless EEG data processing system is used to implement the wireless EEG data processing method according to any one of claims 1-3, and the wireless EEG data processing system includes: The data acquisition module is used to acquire a pre-generated target EEG signal, a first differential component, and a second differential component, wherein the first differential component includes the reaction rate and accuracy obtained using the visual diffusion to auditory attention paradigm, and the second differential component includes the reaction rate and accuracy obtained using the auditory diffusion to visual attention paradigm. The feature acquisition module is used to preprocess the target EEG signal, input it into a deep learning model for extracting EEG signals, and output the potential features of the target EEG signal. A component acquisition module is used to acquire a target component based on the first differential component and the second differential component, wherein the target component is one of the first differential component and the second differential component; The result output module is used to obtain tinnitus parameters, fit the latent features with the target components and the tinnitus parameters through a linear mixture model to obtain predicted parameters, and output them.

5. The wireless EEG data processing system according to claim 4, characterized in that, The feature acquisition module includes: The preprocessing unit is used to preprocess the target EEG signal to obtain a preprocessed EEG signal; An energy acquisition unit is used to perform energy density acquisition processing on the preprocessed EEG signal to obtain the energy corresponding to a preset frequency band in the preprocessed EEG signal. The latent feature generation unit is used to input the energy corresponding to a preset frequency band in the preprocessed EEG signal into the deep learning-based EEG signal extraction model and output the latent features of the target EEG signal.

6. The wireless EEG data processing system according to claim 4, characterized in that, The component acquisition module includes: A component data acquisition unit is used to acquire first data of the first differential component and second data of the second differential component; A comparison unit is used to compare the first data and the second data to obtain a comparison result; The target component generation unit is configured to select the target component from a first difference component in which attention spreads from vision to hearing and a second difference component in which attention spreads from hearing to vision, based on the comparison result.

7. The wireless EEG data processing system according to claim 4, characterized in that, The result output module includes: The parameter acquisition unit is used to acquire tinnitus parameters; Multiple fitting parameter generation units are used to fit the latent features with the target components and the tinnitus parameters through the linear mixture model to obtain fitting parameters corresponding to each tinnitus parameter; The current prediction parameter acquisition unit is used to compare the correlation coefficients of the fitted parameters corresponding to each tinnitus parameter and select the fitted parameter with the largest correlation coefficient as the current prediction parameter. The prediction parameter acquisition unit is used to obtain a preset number of prediction parameters based on the current prediction parameters corresponding to each tinnitus parameter, and then output them.

8. A terminal, characterized in that, The terminal includes a memory, a processor, and a wireless EEG data processing program stored in the memory and executable on the processor. When executed by the processor, the wireless EEG data processing program implements the steps of the wireless EEG data processing method as described in any one of claims 1-3.

Images