Method and system for dynamic coupling analysis of spatial auditory attention based on a multi-layer network

Abstract

The present invention discloses a method and system for dynamically coupling and analyzing spatial auditory attention based on a multi-layer network. The method includes the following steps: collecting electroencephalogram (EEG) signals of multiple brain regions induced by spatial auditory attention tasks in a complex acoustic environment and performing preprocessing on the EEG data; constructing a multi-layer rhythm network and a multi-layer time-varying network; constructing a cross-time-frequency coupling hyperconnection network based on the multi-layer rhythm network and the multi-layer time-varying network, and determining the core network layer of the cross-time-frequency coupling hyperconnection network by using inter-layer correlation and inter-layer conditional probability; quantitatively analyzing the network attributes of the core network layer to obtain the connection pattern and collaborative working mode among the responsible brain regions. The present invention integrates frequency and time information by constructing a four-dimensional hyperconnection network, and accurately quantitatively analyzes the complex interaction mechanism of spatial auditory attention.

Description

Technical Field

[0001] The present invention relates to the technical field of auditory brain-computer interfaces, and particularly to a method and system for dynamic coupling analysis of spatial auditory attention based on a multi-layer network. Background Art

[0002] Spatial auditory attention refers to the ability of an individual to selectively focus on auditory information at a specific spatial location in a complex auditory environment, while suppressing other irrelevant noises or interfering sound sources. This ability is crucial for humans to perform cognitive tasks such as target sound source recognition, speech understanding, and sound source localization in a multi-source environment. Especially in a noisy environment, its role is particularly prominent. For example, in the "cocktail party effect", an individual can focus their attention on the voice of the target interlocutor through spatial auditory attention, while ignoring the surrounding background noise and other conversations. For hearing-impaired patients (such as cochlear implant users), the lack of spatial auditory attention ability will greatly reduce their speech perception ability in a complex environment. Therefore, in-depth research on its neural mechanism is of great significance for improving the performance of auditory rehabilitation and assistive devices.

[0003] In recent years, numerous studies have used a variety of technical means to deeply explore the neural basis of spatial auditory attention in complex acoustic environments. Functional magnetic resonance imaging studies have found that the activities of brain regions such as the bilateral posterior superior temporal gyrus, anterior insula, supplementary motor area, and anterior parietal network in individuals are significantly enhanced in complex environments. Voxel lesion-behavior mapping studies have shown that the right planum temporale and the left inferior frontal gyrus play key roles in the process of spatial auditory selective attention. Magnetoencephalography studies have shown that a specific activation pattern appears in the left precentral gyrus during the execution of spatial auditory attention. Electroencephalography has the advantages of being non-invasive and having high temporal resolution. Related electroencephalography studies have found that there is a correlation between spatial attention allocation and specific event-related potentials (such as mismatch negativity, P3a, and P3b components) in an environment with multiple interfering sources. Although these studies have revealed the complexity and temporality of the brain neural activities involved in spatial auditory attention in complex environments, however, the dynamic connection patterns between different brain regions have not been deeply elucidated.

[0004] Spatial auditory attention is a complex cognitive process that involves the interaction and coordination of multiple brain regions, including traditional auditory processing regions (such as the primary auditory cortex and auditory association areas), as well as brain regions such as the prefrontal lobe and parietal lobe related to attention, memory, and executive function. However, existing methods mostly use static brain network analysis or single-time / frequency dimension analysis, which are difficult to comprehensively characterize the dynamic coupling patterns of different brain regions when the brain performs spatial auditory attention tasks. Therefore, there is an urgent need to develop a method that can analyze the dynamic coupling of spatial auditory attention brain functions across time and frequency, so as to reveal the interaction mechanism of the spatial auditory attention network in the brain in a complex environment and provide an innovative solution for the development of auditory brain-computer interface technology. Summary of the Invention

[0005] The object of the present invention is to provide a method and system for dynamically coupling analysis of spatial auditory attention based on a multi-layer network in view of the technical defects existing in the prior art.

[0006] The technical solution adopted to achieve the object of the present invention is as follows:

[0007] A method for dynamically coupling analysis of spatial auditory attention based on a multi-layer network includes the following steps:

[0008] Step 1, collect electroencephalogram (EEG) signals of multiple brain regions induced by a spatial auditory attention task in a complex acoustic environment, and perform EEG data preprocessing;

[0009] Step 2, based on the preprocessed EEG data obtained in Step 1, extract EEG signals of different frequency bands, and quantitatively analyze the mutual coupling effects of EEG signals of different frequency bands and the same frequency band based on a connectivity index, and then construct a multi-layer rhythm network;

[0010] Step 3, based on the preprocessed EEG data obtained in Step 1, construct a microstate spatio-temporal sequence, and use the microstate spatio-temporal sequence as a microstate frame, and quantitatively analyze the mutual coupling effects of EEG signals of different microstate frames and the same microstate frame based on a connectivity index, and then construct a multi-layer time-varying network;

[0011] Step 4, construct a cross-time-frequency coupling hyperconnection network based on the multi-layer rhythm network obtained in Step 2 and the multi-layer time-varying network obtained in Step 3, and determine the core network layer of the cross-time-frequency coupling hyperconnection network by using inter-layer correlation and inter-layer conditional probability; quantitatively analyze the network attributes of the core network layer to obtain the connection mode and collaborative working mode between each responsible brain region.

[0012] In the above technical solution, in Step 1, a complex acoustic environment containing multiple sound sources is created by a circular speaker array, where one is the target sound source and the others are interfering sound sources, background noise is introduced into both the target sound source and the interfering sound sources, and when the participant performs the spatial auditory attention task, the direction of the target sound source is judged.

[0013] In the above technical solution, in Step 1, 64 electrode channels of a non-invasive EEG device are used to collect EEG signals of different brain regions, and each electrode channel serves as a network node.

[0014] In the above technical solution, in Step 1, the EEG data preprocessing steps include filtering, artifact removal, segmentation, baseline correction, and rereferencing to obtain EEG data with a fixed-length time window.

[0015] In the above technical solution, in step 2, the EEG data of each fixed-length time window obtained in step 1 is subjected to frequency decomposition to extract the EEG data of different frequency bands. The frequency bands include delta wave, theta wave, alpha wave, beta wave and gamma wave. The connectivity index is phase synchronization, generalized synchronization or partial directed coherence. The coupling strength between the EEG signals of all network nodes within the same frequency band is calculated using the connectivity index, and the coupling strength between the EEG signals of each network node between different frequency bands is calculated in turn, thereby constructing a multi-layer rhythm network.

[0016] In the above technical solution, in step 3, based on the Cartool software, the optimal microstate template is extracted from the EEG data of all fixed-length time windows obtained in step 1 using the spatial clustering algorithm, and then the optimal microstate template is applied to locate the EEG data of the fixed-length time window at different angles. Similar EEG activity patterns are identified through the template matching method, thereby constructing a microstate spatio-temporal sequence. The microstate spatio-temporal sequence is composed of multiple microstate frames. The connectivity index is phase synchronization, generalized synchronization or partial directed coherence. The coupling strength between the EEG signals of all network nodes within the same microstate frame is calculated using the connectivity index, and the coupling strength between the EEG signals of each network node between different microstate frames is calculated in turn, thereby constructing a multi-layer time-varying network.

[0017] In the above technical solution, in step 4, the cross-time-frequency coupling hyperconnection network is a four-dimensional hyperconnection network. The first dimension is used to represent different microstate frames, reflecting different stages of brain nerve activity in the time process; the second dimension represents different frequency bands, reflecting the activity characteristics of the brain at different frequency levels; the third dimension and the fourth dimension are respectively used to describe the connection strength within the same layer and between different layers, accurately quantifying the interaction relationship between each network node in the brain network.

[0018] In the above technical solution, in step 4, the graph theory analysis method is used to quantitatively analyze the network attributes of the core network layer. The network attributes are clustering coefficient, global efficiency, node degree or minimum characteristic path length.

[0019] In the above technical solution, in step 4, the inter-layer correlation is calculated through the Pearson correlation coefficient:

[0020] ;

[0021] where is the Pearson correlation coefficient between two network layers and are the network matrices of the m-th and n-th layers respectively, and is the average of the m-th and n-th layer network matrices, i, j= 1, 2, ……, M are all the number of network layers.

[0022] In the above technical solution, in step 4, the inter-layer conditional probability is calculated by the following formula:

[0023] ;

[0024] where, is the inter-layer conditional probability, and are the m-th and n-th layer network matrices respectively, i, j= 1, 2, ……, M are all the number of network layers.

[0025] On the other hand, the present invention further includes a system based on the above-mentioned method for dynamically coupling and analyzing spatial auditory attention based on a multi-layer network, including an experimental design module, an electroencephalogram signal acquisition and preprocessing module, a multi-layer network construction and analysis module, a hyper-connection network construction module, and a visualization and report generation module, wherein:

[0026] The experimental design module automatically generates and adjusts the experimental paradigm, and the experimental design module includes an acoustic scene design unit and an experimental stimulus and process generation unit;

[0027] The electroencephalogram signal acquisition and preprocessing module includes an electroencephalogram data acquisition unit and a data preprocessing unit;

[0028] The multi-layer network construction and analysis module includes a hyper-connection network generation unit for generating the cross-time-frequency coupling hyper-connection network, a correlation and conditional probability analysis unit for quantitatively analyzing the cross-time-frequency coupling hyper-connection network, and a core network layer screening unit for screening the core network layer;

[0029] The visualization and report generation module is used to visualize the analysis results of the multi-layer network construction and analysis module.

[0030] Compared with the prior art, the beneficial effects of the present invention are:

[0031] (1) Highly simulate a complex acoustic environment: The present invention uses a speaker array to construct a multi-source space, combines sentences with high semantic complexity, music segments, etc. as stimulus materials, and at the same time introduces real background noises such as street noise and multi-person conversation sounds to enhance the authenticity and ecological validity of the experimental environment, making the experimental results closer to the actual auditory scene;

[0032] (2)Comprehensively capture the spatio-temporal-frequency characteristics of EEG signals: This invention uses methods such as wavelet transform and short-time Fourier transform for multi-band decomposition to accurately extract rhythm signals such as delta and theta. At the same time, by constructing spatio-temporal sequences through microstate analysis, it can capture the characteristics of EEG signals from both time and space dimensions, making the dynamic changes of brain activities clearer and more measurable.

[0033] (3)Construct multi-layer networks and hyper-connected networks: This invention constructs multi-layer rhythm networks and multi-layer time-varying networks to quantify the coupling effect of EEG signals across rhythms and time scales. By constructing a four-dimensional hyper-connected network, it integrates frequency and time information to accurately quantify the connection strength of each network node. At the same time, it determines the core network layer and analyzes its network properties, revealing the complex interaction mechanism of the responsible brain regions in the brain during spatial auditory attention at different frequencies and time periods, providing a new perspective and a more in-depth research method for understanding the spatial auditory attention function of the brain.

[0034] (4)Construct an efficient automated analysis system: The dynamic analysis system of spatial auditory attention coupling based on multi-layer networks constructed in this invention integrates modules such as experimental design, EEG data acquisition and preprocessing, network construction, hyper-connection matrix analysis, visualization, and report generation to achieve full-process automated analysis. The system has adjustable parameters and high-efficient data processing capabilities, supports flexible adjustment of different noise environments, stimulus materials, frequency bands, and time windows, significantly improves research efficiency, reduces manual operation errors, and provides a convenient and efficient tool for auditory cognitive research and clinical evaluation. Description of the Drawings

[0035] Figure 1 It is a flow chart of the dynamic coupling analysis method of spatial auditory attention based on multi-layer networks.

[0036] Figure 2 It is a diagram of the experimental paradigm of spatial auditory attention in a complex acoustic environment, where (a) is a schematic diagram of the experiment and (b) is a flow chart of the experiment.

[0037] Figure 3 It is a schematic diagram of constructing a multi-layer rhythm network.

[0038] Figure 4 It is a schematic diagram of constructing a microstate spatio-temporal sequence.

[0039] Figure 5 It is a schematic diagram of constructing a multi-layer time-varying network.

[0040] Figure 6 It is a schematic diagram of the core network layer of the hyper-connection matrix.

[0041] Figure 7 It is a schematic diagram of the dynamic coupling analysis system of spatial auditory attention based on multi-layer networks. Detailed Implementation Manner

[0042] The present invention will be further described in detail below in conjunction with specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0043] Embodiment 1

[0044] As Figure 1 shown, a method for dynamically coupling and analyzing spatial auditory attention based on a multi-layer network includes the following steps:

[0045] Step 1, collect the electroencephalogram (EEG) signals induced by multiple brain regions during a spatial auditory attention task in a complex acoustic environment, and perform preprocessing on the EEG data;

[0046] Step 2, based on the preprocessed EEG data obtained in Step 1, extract the EEG signals in different frequency bands, and quantitatively analyze the mutual coupling effects of the EEG signals in different frequency bands and the same frequency band based on connectivity metrics, and then construct a multi-layer rhythm network;

[0047] Step 3, based on the preprocessed EEG data obtained in Step 1, construct a microstate spatio-temporal sequence, and use the microstate spatio-temporal sequence as a microstate frame to quantitatively analyze the mutual coupling effects of the EEG signals in different microstate frames and the same microstate frame based on connectivity metrics, and construct a multi-layer time-varying network;

[0048] Step 4, construct a cross-time-frequency coupling hyperconnection network based on the multi-layer rhythm network obtained in Step 2 and the multi-layer time-varying network obtained in Step 3, and use inter-layer correlation and inter-layer conditional probability to determine the core network layer of the cross-time-frequency coupling hyperconnection network; quantitatively analyze the network attributes of the core network layer to obtain the connection mode and collaborative working mode between the responsible brain regions.

[0049] Embodiment 2

[0050] This embodiment will be further described in detail on the basis of Embodiment 1.

[0051] A method for dynamically coupling and analyzing spatial auditory attention based on a multi-layer network includes the following steps.

[0052] Step 1: Design a spatial auditory attention experimental paradigm in a complex acoustic environment based on a loudspeaker array, which includes:

[0053] Step 1.1, build a complex multi-source environment. Create an acoustic environment containing multiple sound sources through a circular loudspeaker array to simulate a complex auditory scene. Taking the setting of four sound sources as an example, arrange the four loudspeakers at the 90° and 30° positions on both sides of the midplane of the subject. The specific layout is as Figure 2As shown in (a) below. In addition, sentences, music segments, etc. with semantic complexity and emotional color are selected as stimulus materials, and one of them is designated as the target sound source, while the remaining sound sources are used as interfering sound sources. This design aims to enhance the ecological validity of the experiment and effectively induce the brain responses of the subjects in the spatial auditory attention task;

[0054] Step 1.2, construct a background noise environment. Based on the stimulus materials selected in Step 1.1, background noise is introduced to enhance the simulation of real-life scenarios in the experiment. The background noise includes, but is not limited to, street noise, voices of multiple people talking indoors, etc. An audio processing software (such as Adobe Audition) is used to mix the background noise with the original stimulus materials to form noise environments with different signal-to-noise ratios, such as 0 dB and 10 dB signal-to-noise ratio conditions. In addition, all stimulus materials are normalized using audio processing software to ensure that their root mean square amplitudes and durations are consistent;

[0055] Step 1.3, design the experimental procedure. The MATLAB and Psychtoolbox software are used to generate an experimental stimulus sequence, and a specific sequence is shown as Figure 2 in (b) below. The experiment consists of 200 groups of stimuli and is divided into two modules. During the experiment, four sound sources (one target sound source and three interfering sound sources) are played synchronously and randomly from the speakers, and the subjects are required to judge the direction of the target sound source by pressing a key within 2000 ms after each group of stimuli ends. To reduce the influence of the previous stimulus on subsequent judgments, a silent interval of 1000 ms is set between each stimulus. The experiment is carried out according to the above procedure until all 200 groups of stimuli are played, and the experiment ends.

[0056] Step 2: Collect EEG signals based on a non-invasive EEG device and perform data preprocessing

[0057] Step 2.1, EEG data collection. The subjects perform the spatial auditory attention task according to the experimental procedure designed in Step 1.3, and at the same time, professional EEG devices such as the 64-channel NeuroScan SynAmps2 are used to record the EEG data induced by the subjects during the task. To ensure the accuracy and reliability of the collected data, the experimental conditions need to be strictly controlled during the recording process so that the impedance of all electrodes always remains below 10 kΩ, and the sampling rate is set to 1000 Hz. At the same time, the collected data is preliminarily processed using a 0.1 - 150 Hz band-pass filter and a 50 Hz notch filter to reduce noise interference;

[0058] Step 2.2, preprocessing of EEG data. Use EEGLAB plug-ins to preprocess the EEG data. Specifically, filter the EEG data using a 0.1 - 48 Hz band-pass filter and downsample the data to 250 Hz. Then, use independent component analysis to remove artifacts caused by eye movement or blinking. Further, divide the data into 1200 ms time windows (epochs), each epoch including 200 ms before the stimulus and 1000 ms after the stimulus, and use the data of the first 200 ms for baseline correction. In addition, exclude epochs with a maximum-minimum amplitude difference exceeding 100 μV to ensure data quality. Finally, re-reference all the remaining epoch data to the mean of all electrodes to unify the data reference standard and ensure the consistency and comparability of the data.

[0059] Step 3: Construct a multi-layer rhythm network based on connectivity metrics, which includes:

[0060] Step 3.1, extract EEG signals of different frequency bands. Use signal processing techniques such as wavelet transform or short-time Fourier transform to perform frequency decomposition on the epoch data obtained in Step 2.2, and extract EEG signals of different frequency bands, including but not limited to delta waves, theta waves, alpha waves, beta waves, and gamma waves. This step aims to separate different frequency components for subsequent construction of the multi-layer rhythm network;

[0061] Step 3.2, construct a multi-layer rhythm network. Use connectivity metrics such as phase synchronization, generalized synchronization, and partial directed coherence to quantitatively analyze the mutual coupling effects of EEG signals in different frequency bands (cross-frequency bands) and their respective frequency bands (same-frequency bands) obtained in Step 3.1, and then construct a multi-layer rhythm network, as Figure 3 shown. Specifically, calculate the coupling strength within each frequency band. For example, use the above metrics to calculate the coupling strength between all network nodes (such as 64 electrode channels) in the theta frequency band; for the network coupling strength between different frequency bands, use the above metrics to calculate the coupling strength between cross-frequency bands in turn (such as the coupling strength between the first channel in theta and the second channel in the alpha frequency band);

[0062] Step 3.3, manifestation form of the multi-layer rhythm network. The construction principle of the M-layer rhythm brain network can be shown by the following formulas (1)-(3):

[0063] (1);

[0064] (2);

[0065] (3);

[0066] In the above formula system, represents a diagonal matrix , matrix, t = 1, 2, 3, ……, M , M is the number of decomposed frequency bands, represents the coupling strength of different network nodes within their respective frequency bands. represents the coupling strength of different network nodes across frequency bands. Matrix is a matrix with all elements equal to 1, that is, it means that any network node in each layer can be connected to any network node in another layer. represents the interaction matrix between layers, M is the number of decomposed frequency bands, N is the number of network nodes. In this visualization expression, the matrix on the diagonal represents the coupling within the same frequency band, such as the coupling between network nodes within the theta frequency band. This kind of connection shows the coupling relationship between different brain regions under the same frequency band; while the matrix off the diagonal represents the coupling across frequency bands, such as the coupling of network nodes between the theta frequency band and the gamma frequency band. The cross-frequency band connection reflects the synergy between different frequency band EEG activities, which helps to reveal the mechanism of multi-frequency band information integration in the brain during the spatial auditory attention task.

[0067] Step 4: Construct a multi-layer time-varying network based on the connectivity index, including:

[0068] Step 4.1, construct the microstate spatio-temporal sequence. Based on the epoch data processed in Step 2.2, use the microstate analysis method to construct the microstate spatio-temporal sequence. This process is mainly implemented in the Cartool software. The specific steps are as follows: Based on all epoch data, use the spatial clustering algorithm to extract the optimal microstate template. This template represents the characteristic distribution of EEG data in the spatial dimension and reflects the neural electrical activity pattern of the brain in a specific cognitive task. Subsequently, apply the obtained optimal microstate template to the epoch data at different orientation angles (30°, -30°, 90°, -90°), and identify similar EEG activity patterns through the template matching method, thereby constructing four microstate spatio-temporal sequences (all four microstate spatio-temporal sequences include W1 to W6, the difference is the microstate frames of W1 to W6). Finally, the obtained microstate spatio-temporal sequence serves as the microstate frame of the multi-layer network, providing a basis for the subsequent construction of the multi-layer time-varying network, as specifically shown in Figure 4 shown;

[0069] Step 4.2, Construction of multi-layer time-varying network. Based on the microstate frames (W1 to W6) obtained in Step 4.1 and combined with the method of constructing a multi-layer rhythm network in Step 3.2, a multi-layer time-varying network is constructed. The specific method is as follows: Connectivity metrics such as phase synchronization, generalized synchronization, and partial directed coherence are used to quantitatively analyze the mutual coupling effects of each network node (such as 64 electrode channels) across time and at the same time scale, and then a multi-layer time-varying network is constructed. As Figure 5 shown, at the cross-time scale, the coupling relationships between network nodes under different microstate frames (such as W1 and W6 microstate frames) are analyzed to capture the dynamic changes of neural activities over time; at the same time scale, the coupling relationships between network nodes under the same microstate frame (such as W1 and W1 microstate frames) are analyzed to characterize the functional connectivity of this time window. The constructed multi-layer time-varying network can comprehensively and dynamically present the time-varying functional connectivity characteristics between different brain regions during the spatial auditory attention task;

[0070] Step 4.3, Representation form of the multi-layer network. Based on the multi-layer time-varying network constructed in Step 4.2, the principle and method of visualization in Step 3.3 are used to visualize the multi-layer time-varying network. Representation form of the multi-layer time-varying network. The construction principle of the P-layer time-varying brain network can be shown by the following formulas (4)-(6):

[0071] (4);

[0072] (5);

[0073] (6);

[0074] In the above formula system, represents a diagonal matrix, and the matrix , t = 1, 2, 3,..., P , P is the number of microstate frames, and the matrix represents the coupling strength of different network nodes within their respective microstate frames. represents the coupling strength of different network nodes across microstate frames, and the matrix is a matrix with all elements equal to 1, that is, it means that any network node in each layer can be connected to any network node in another layer. represents the interaction matrix between layers, P is the number of microstate frames, Nis the number of network nodes. In this visual representation, the matrix on the diagonal represents the coupling at the same time scale, such as the coupling between W1 and W1. This kind of connection reflects the interaction relationship between different network nodes in the same microstate frame, which helps to observe the functional connection state of the brain at a certain instant. While the matrix off the diagonal represents the coupling across time scales, such as the coupling between W1 and W6. This type of coupling shows the dynamic connection between network nodes in different microstate frames, and can intuitively present the evolution law of brain nerve activities on the time axis, providing an intuitive basis for deeply understanding the dynamic functional mechanism of the brain during spatial auditory attention.

[0075] Step 5: Construct a cross-time-frequency coupling hyperconnection matrix based on multi-layer rhythms and time-varying networks, including:

[0076] Step 5.1, construct the hyperconnection matrix. In this step, based on the characteristics of cross-frequency bands and cross-time scales, a coupling operation is performed on the multi-layer rhythm network constructed in Step 3 and the multi-layer time-varying network constructed in Step 4 to construct a four-dimensional hyperconnection matrix. Each dimension of this four-dimensional hyperconnection matrix has clear and important meanings: the first dimension is used to represent different microstate frames, and these time periods are determined by the microstate sequence constructed in Step 4.1, reflecting different stages of brain nerve activities in the time process; the second dimension represents different frequency bands, covering various electroencephalogram rhythms such as delta wave and theta wave, reflecting the activity characteristics of the brain at different frequency levels; the third dimension and the fourth dimension are respectively used to describe the connection strength within the same layer and between different layers, precisely quantifying the interaction relationship between network nodes in the multi-layer network. By constructing such a four-dimensional matrix, the dynamic connection information of the brain in the time and frequency dimensions is comprehensively integrated, providing a powerful analysis method for deeply analyzing the dynamic coupling of the spatial auditory attention brain network;

[0077] Step 5.2, quantitative analysis of the hyperconnection matrix.

[0078] Use inter-layer correlation and inter-layer conditional probability to quantitatively analyze the cross-time-frequency coupling hyperconnection network.

[0079] The inter-layer correlation is defined as the Pearson correlation coefficient between two network layers, as shown in Equation (7):

[0080] (7);

[0081] where, is the Pearson correlation coefficient between two network layers and are the network matrices of the m-th and n-th layers respectively, and are the averages of the network matrices of the m-th and n-th layers,i, j= 1, 2, ……, M are all the numbers of network layers.

[0082] The inter-layer conditional probability is defined as the conditional probability that the connection edges in the m-th layer still exist in the n-th layer, as shown in Equation (8):

[0083] (8);

[0084] where, is the inter-layer conditional probability, and are the network matrices of the m-th and n-th layers respectively, i, j= 1, 2, ……, M are all the numbers of network layers.

[0085] Through the above correlation analysis, for example, when it is found that the 200 - 300 ms connection matrix (W2 - W3) has a high correlation with the theta and alpha connection matrices (band2 - band3), it can be speculated that within this time window, the coupled oscillation between theta and alpha plays an important role in the spatial auditory attention task, which helps to further reveal the cooperative mechanism of different frequency EEG activities in this task.

[0086] Step 5.3, determine the core network layer. According to the results of correlation and conditional probability analysis, the coupled layer with the highest correlation is selected as the core network layer. This core network layer plays a key role in the realization of brain functions in the spatial auditory attention task, and it centrally reflects the key neural activity patterns and functional connection relationships of the brain in this task. To more intuitively explore the characteristics of the core network layer, the BrainNetViewer toolbox is used to visualize the topological structure of the core network layer, and the specific visualization effect is as Figure 6 shown. Through visual analysis, the connection patterns and collaborative working methods between each responsible brain region can be intuitively observed (such as the connection line schematic in Figure 6 ), so as to deeply understand the functional coordination mechanism of the brain in the process of spatial auditory attention, providing an intuitive basis for studying the spatial auditory cognitive process of the brain. At the same time, the graph theory analysis method is used to quantitatively study the network properties of the core network layer, and the network properties involved include but are not limited to clustering coefficient, global efficiency, node degree, and minimum eigenpath, etc.

[0087] Example 3

[0088] Based on the analysis steps in Embodiment 1 or Embodiment 2, this embodiment constructs a dynamic coupling analysis system for spatial auditory attention based on a multi-layer network. This system can realize the whole-process automation processing from the design of the spatial auditory attention task experimental paradigm in a complex acoustic environment to the acquisition and preprocessing of EEG signals, network analysis, and visualization, and improve the analysis efficiency. As Figure 7 shown, this system specifically includes the following component modules:

[0089] (1) Experimental design module: This module can automatically generate and adjust the experimental paradigm, including the following units: ① Acoustic scene design unit: By precisely controlling the arrangement of the speaker array and the configuration of background noise, it simulates a complex acoustic environment. The acoustic scene design unit includes a speaker array controller, a noise generator, and a background noise adjustment module, which can realize the environmental setting under different signal-to-noise ratio conditions. ② Experimental stimulus and process generation unit: It can automatically generate and schedule the experimental stimulus sequence. The system supports adjusting the stimulus type, presentation order, and experimental duration according to user needs, and has functions such as stimulus signal normalization processing and dynamic signal-to-noise ratio adjustment;

[0090] (2) EEG signal acquisition and preprocessing module. This module mainly realizes the acquisition, preprocessing, and preliminary analysis of EEG data. It includes: ① EEG data acquisition unit: Compatible with EEG devices such as NeuroScan SynAmps2, it realizes the high-precision acquisition of EEG signals when the subject performs spatial auditory tasks. ② Data preprocessing unit: It performs band-pass filtering, notch filtering, and artifact removal operations in real time to ensure the signal quality, and at the same time performs preprocessing steps such as downsampling and time window division;

[0091] (3) Multi-layer network construction and analysis module. This module supports multi-layer network analysis based on different rhythms and time windows to dynamically reveal the characteristics of brain functional connectivity, including: ① Multi-layer rhythm construction module: It uses a variety of signal processing techniques to extract EEG signals in different frequency bands, and calculates the EEG coupling strength within and across frequency bands using connectivity indicators, so as to construct a multi-layer rhythm network. ③ Multi-layer time-varying network construction module: Based on the microstate spatio-temporal sequence, it uses the connectivity analysis method to calculate the EEG coupling strength across time scales to construct a multi-layer time-varying network;

[0092] (4) Hyperconnection network construction module. This module is responsible for coupling multi-layer rhythm and multi-layer time-varying networks, generating a cross-time-frequency coupling hyperconnection network, and conducting quantitative analysis, including: ① Hyperconnection network generation unit: Combining the time and frequency dimensions, through cross-band and cross-time-scale coupling operations, generate a cross-time-frequency coupling hyperconnection network (four-dimensional hyperconnection network) to comprehensively integrate the neural activity patterns of the brain in spatial auditory tasks. ② Correlation and conditional probability analysis unit: Select methods such as Pearson correlation and conditional probability to analyze the hyperconnection network to reveal the cooperative mechanism between different frequency bands and time windows. ③ Core network layer screening module: Automatically screen the core network layer with important functions based on the correlation analysis results;

[0093] (5) Visualization and report generation module. This module is used for the visual display of the analysis results of the multi-layer network and automatically generates an experimental analysis report, including: ① Network visualization unit: Display the topological structures of the multi-layer rhythm network and the multi-layer time-varying network, and support the visualization of the brain region connection patterns at different frequency bands and time scales. ② Dynamic connection graph visualization unit: Real-time draw the dynamic graph of the functional connections of the brain network changing with time to intuitively present the brain functional activities in spatial auditory attention tasks. ③ Automatic report generation unit: Automatically generate an experimental report based on the analysis results, and the report includes but is not limited to key research information such as graph theory analysis and dynamic functional connection results.

[0094] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for dynamically coupling analysis of spatial auditory attention based on a multi-layer network, characterized in that Including the following steps: Step 1: Collect the electroencephalogram (EEG) signals of multiple brain regions induced by spatial auditory attention tasks in a complex acoustic environment, and perform EEG data preprocessing; Step 2: Based on the preprocessed EEG data obtained in Step 1, extract EEG signals in different frequency bands, quantitatively analyze the mutual coupling effects of EEG signals in different frequency bands and the same frequency band based on connectivity metrics, and then construct a multi-layer rhythm network; Step 3: Based on the preprocessed EEG data obtained in Step 1, construct a microstate spatio-temporal sequence, use the microstate spatio-temporal sequence as a microstate frame, quantitatively analyze the mutual coupling effects of EEG signals in different microstate frames and the same microstate frame based on connectivity metrics, and then construct a multi-layer time-varying network; Step 4: Construct a cross-time-frequency coupling hyperconnection network based on the multi-layer rhythm network obtained in Step 2 and the multi-layer time-varying network obtained in Step 3, and determine the core network layer of the cross-time-frequency coupling hyperconnection network using inter-layer correlation and inter-layer conditional probability; Quantitatively analyze the network properties of the core network layer to obtain the connection patterns and collaborative working methods among the responsible brain regions.

2. The method for dynamically coupling analysis of spatial auditory attention based on a multi-layer network according to claim 1, wherein In Step 1, a complex acoustic environment containing multiple sound sources is created through a circular loudspeaker array, where one is the target sound source and the others are interference sound sources. Background noise is introduced into both the target sound source and the interference sound sources. When the participant performs the spatial auditory attention task, the direction of the target sound source is judged.

3. The method for dynamically coupling and analyzing spatial auditory attention based on a multi-layer network according to claim 1, characterized in that, In Step 1, 64 electrode channels of a non-invasive EEG device are used to collect EEG signals of different brain regions, and each electrode channel serves as a network node.

4. The method for dynamically coupling and analyzing spatial auditory attention based on a multi-layer network according to claim 1, characterized in that In Step 1, the EEG data preprocessing steps include filtering, artifact removal, segmentation, baseline correction, and rereferencing to obtain EEG data in a fixed-length time window.

5. The method for dynamically coupling analysis of spatial auditory attention based on a multi-layer network according to claim 1, wherein In Step 2, the EEG data in each fixed-length time window obtained in Step 1 is decomposed in frequency to extract EEG data in different frequency bands. The frequency bands include delta wave, theta wave, alpha wave, beta wave, and gamma wave. The connectivity metric is phase synchronization, generalized synchronization, or partial directed coherence. The coupling strength between the EEG signals of all network nodes in the same frequency band is calculated using the connectivity metric, and the coupling strength between the EEG signals of each network node between different frequency bands is calculated in turn, and then a multi-layer rhythm network is constructed.

6. The method for dynamically coupling analysis of spatial auditory attention based on a multi-layer network according to claim 1, wherein In step 3, based on the Cartool software, the optimal microstate templates are extracted from the EEG data of all fixed-length time windows obtained in step 1 using a spatial clustering algorithm, and then the optimal microstate templates are applied to localize the EEG data of fixed-length time windows at different angles. Similar EEG activity patterns are identified through a template matching method, thereby constructing a microstate spatio-temporal sequence. The microstate spatio-temporal sequence is composed of multiple microstate frames. The connectivity index is phase synchronization, generalized synchronization, or partial directed coherence. The coupling strength between the EEG signals of all network nodes within the same microstate frame is calculated using the connectivity index, and the coupling strength between the EEG signals of each network node between different microstate frames is calculated in turn, thereby constructing a multi-layer time-varying network.

7. The method for dynamically coupling and analyzing spatial auditory attention based on a multi-layer network according to claim 1, wherein In step 4, a graph theory analysis method is used to quantitatively analyze the network properties of the core network layer. The network properties are clustering coefficient, global efficiency, node degree, or minimum characteristic path length.

8. The method for dynamically coupling analysis of spatial auditory attention based on a multi-layer network according to claim 1, wherein In step 4, the inter-layer correlation is calculated through the Pearson correlation coefficient: ； Among them, is the Pearson correlation coefficient between two network layers and are the network matrices of the m-th and n-th layers respectively, and are the averages of the network matrices of the m-th and n-th layers, i, j = 1, 2, ……, M are all the numbers of network layers.

9. The method for dynamically coupling analysis of spatial auditory attention based on a multi-layer network according to claim 1, wherein In step 4, the inter-layer conditional probability is calculated by the following formula: ； Among them, is the inter-layer conditional probability, and are the network matrices of the m-th and n-th layers respectively, i, j = 1, 2, ……, M are all the number of network layers.

10. A system based on the method for dynamically coupling analysis of spatial auditory attention based on a multi-layer network as claimed in claim 1, characterized in that, It includes an experimental design module, an EEG signal acquisition and preprocessing module, a multi-layer network construction and analysis module, a hyper-connected network construction module, and a visualization and report generation module, where: The experimental design module automatically generates and adjusts the experimental paradigm. The experimental design module includes an acoustic scene design unit and an experimental stimulus and process generation unit; The EEG signal acquisition and preprocessing module includes an EEG data acquisition unit and a data preprocessing unit; The multi-layer network construction and analysis module includes a hyper-connected network generation unit for generating the cross-time-frequency coupling hyper-connected network, a correlation and conditional probability analysis unit for quantitatively analyzing the cross-time-frequency coupling hyper-connected network, and a core network layer screening unit for screening the core network layer; The visualization and report generation module is used to visualize the analysis results of the multi-layer network construction and analysis module.

Images