Method, system and device for cognitive state evaluation and regulation of few-electrode electroencephalogram signals

Abstract

The application discloses a cognitive state evaluation and regulation method, system and device of a few-electrode electroencephalogram signal. The method selects a blood oxygen level dependent signal feature set related to a preset cognitive state. The blood oxygen level dependent signal feature set is used as a label of preprocessed scalp electroencephalogram signals. A trained first prediction model is determined based on the label and the preprocessed scalp electroencephalogram signals. A target scalp electroencephalogram signal is input into the trained first prediction model to obtain a target few-electrode electroencephalogram signal. The target few-electrode electroencephalogram signal is filtered, and the filtered target few-electrode electroencephalogram signal is input into a constructed cognitive state evaluation model to obtain a cognitive state output by the cognitive state evaluation model. The target cognitive state of a current subject is regulated according to the cognitive state output by the cognitive state evaluation model. The application improves the accuracy of cognitive state evaluation and improves the spatial specificity of online cognitive regulation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cognitive state evaluation, in particular to a cognitive state evaluation and regulation method, system and device based on few-electrode electroencephalogram signals. BACKGROUND

[0002] As the core element of human-machine interaction system, the cognitive state of human has a significant impact on the stability and safety of the system. Accurate evaluation of cognitive state and further regulation of the cognitive level of the operator are of great significance to maintain efficient work efficiency, enhance the operation experience, ensure system safety and reduce the accident rate.

[0003] Cognitive ability is related to specific brain regions, neural circuits or brain networks of the human brain. As one of the most commonly used non-invasive brain imaging techniques, nuclear magnetic resonance has very high spatial resolution and can accurately locate cognitive-related brain regions. However, the nuclear magnetic resonance device has poor portability and is expensive. As another most commonly used non-invasive brain imaging signal, electroencephalogram has high temporal resolution, good portability and economy. However, due to the low spatial resolution, it cannot accurately locate the neural signals of specific brain regions or neural circuits related to cognitive tasks. The electroencephalogram source localization method can locate the functional brain regions related to specific tasks, but generally requires 64 or more electrodes, which is less economical and portable.

[0004] Therefore, how to recover the blood oxygen level dependent (BOLD) signal of the cognitive-related specific brain region through the few-electrode electroencephalogram signal, so as to realize the evaluation of cognitive state using the few-electrode electroencephalogram signal and online cognitive regulation is a problem to be solved. SUMMARY

[0005] The present application aims to provide a cognitive state evaluation and regulation method, system and device based on few-electrode electroencephalogram signals, which can recover the blood oxygen level dependent signal of the cognitive-related specific brain region through the few-electrode electroencephalogram signal, thereby improving the accuracy of cognitive state evaluation using the few-electrode electroencephalogram signal and improving the spatial specificity of online cognitive regulation.

[0006] In a first aspect, the embodiments of the present application provide a cognitive state evaluation and regulation method based on few-electrode electroencephalogram signals, which comprises:

[0007] acquiring the simultaneously collected scalp electroencephalogram signals, functional magnetic resonance imaging data and structural magnetic resonance imaging data;

[0008] preprocessing the structural magnetic resonance imaging data to obtain preprocessed structural magnetic resonance imaging data; preprocessing the functional magnetic resonance imaging data according to the preprocessed structural magnetic resonance imaging data to obtain preprocessed functional magnetic resonance imaging data; and preprocessing the scalp electroencephalogram signal to obtain preprocessed scalp electroencephalogram signal;

[0009] selecting a blood oxygen level dependent signal feature set related to a preset cognitive state according to the preprocessed functional magnetic resonance imaging data, wherein the blood oxygen level dependent signal feature set comprises at least one of a brain region blood oxygen level dependent signal, a neural circuit blood oxygen level dependent signal and a brain network blood oxygen level dependent signal;

[0010] taking the blood oxygen level dependent signal feature set as a label of the preprocessed scalp electroencephalogram signal, and determining a trained first prediction model based on the label and the preprocessed scalp electroencephalogram signal;

[0011] inputting a target scalp electroencephalogram signal into the trained first prediction model to obtain a target few-electrode electroencephalogram signal, wherein the target few-electrode electroencephalogram signal is an electroencephalogram signal of a small number of electrodes related to the label;

[0012] filtering the target few-electrode electroencephalogram signal and inputting the filtered target few-electrode electroencephalogram signal into a constructed cognitive state evaluation model to obtain a cognitive state output by the cognitive state evaluation model;

[0013] According to the cognitive state output by the cognitive state evaluation model, the target cognitive state of the current subject is regulated.

[0014] Compared with the prior art, the first aspect of the present application has the following beneficial effects:

[0015] The method can provide a good data basis for training the first prediction model by preprocessing structural magnetic resonance imaging data, preprocessing functional magnetic resonance imaging data, and preprocessing scalp electroencephalogram signals; according to the preprocessed functional magnetic resonance imaging data, a blood oxygen level dependent signal feature set related to a preset cognitive state is selected, the blood oxygen level dependent signal feature set is taken as a label of the preprocessed scalp electroencephalogram signals, and based on the label and the preprocessed scalp electroencephalogram signals, a trained first prediction model is determined; the target scalp electroencephalogram signal is input into the trained first prediction model, and a target few-electrode electroencephalogram signal is obtained; the first prediction model is trained by associating the brain region blood oxygen level dependent signal, the neural circuit blood oxygen level dependent signal and the brain network blood oxygen level dependent signal with the preprocessed scalp electroencephalogram signal, so that the trained first prediction model can predict the electroencephalogram signal related to the brain region blood oxygen level dependent signal, the neural circuit blood oxygen level dependent signal and the brain network blood oxygen level dependent signal according to the collected electroencephalogram signal; the target few-electrode electroencephalogram signal is filtered, and the filtered target few-electrode electroencephalogram signal is input into the constructed cognitive state evaluation model to obtain a cognitive state output by the cognitive state evaluation model; and the target cognitive state of the current subject is regulated according to the cognitive state output by the cognitive state evaluation model, which can improve the recovery of the blood oxygen level dependent signal of the cognitive-related specific brain region by the few-electrode electroencephalogram signal, thereby improving the accuracy of the cognitive state evaluation using the few-electrode electroencephalogram signal and improving the spatial specificity of online cognitive regulation.

[0016] In some embodiments, the selecting, according to the preprocessed functional magnetic resonance imaging data, a blood oxygen level dependent signal feature set related to a preset cognitive state comprises:

[0017] The preprocessed functional magnetic resonance imaging data of the whole brain is divided into a plurality of blood oxygen level dependent signal feature sets, wherein the functional magnetic resonance imaging data comprises at least one of a brain region functional magnetic resonance imaging signal, a neural circuit functional magnetic resonance imaging signal, and a brain network functional magnetic resonance imaging signal;

[0018] The multivariate pattern analysis method is used to select a blood oxygen level dependent signal feature set related to a preset cognitive state from the plurality of blood oxygen level dependent signal feature sets.

[0019] In some embodiments, the determining, based on the label and the preprocessed scalp electroencephalogram signal, a trained first prediction model comprises:

[0020] The preprocessed scalp electroencephalogram signal is filtered to obtain a multi-band scalp electroencephalogram signal;

[0021] The multi-band scalp electroencephalogram signal is processed by a sliding window method to obtain a multi-time period multi-band electroencephalogram signal.

[0022] The multi-period and multi-band EEG signals are input into a first prediction model, and the first prediction model is trained according to the labels to obtain a trained first prediction model.

[0023] In some embodiments, the first prediction model includes a channel attention layer, a spatial attention layer, and a neural network prediction layer. Inputting the multi-period multi-band EEG signal into the first prediction model, training the first prediction model according to the label, and obtaining a trained first prediction model includes:

[0024] Inputting the multi-period and multi-band EEG signals into the channel attention layer to obtain an output result of the channel attention layer;

[0025] Input the output result of the channel attention layer to the spatial attention layer to obtain the output result of the spatial attention layer;

[0026] Inputting the output result of the spatial attention layer into the neural network prediction layer to obtain the output result of the neural network prediction layer;

[0027] Constructing a first loss function according to the output result of the neural network prediction layer and the label;

[0028] The first prediction model is trained according to the first loss function until the first loss function converges or reaches a preset number of iterations, thereby obtaining a trained first prediction model.

[0029] In some embodiments, the cognitive state assessment model includes a temporal feature extraction module, a spatial feature extraction module, and a classification module. Inputting the filtered target few-electrode EEG signal into the constructed cognitive state assessment model to obtain the cognitive state output by the cognitive state assessment model includes:

[0030] Inputting the filtered target few-electrode EEG signal into the spatial feature extraction module to obtain an output result of the spatial feature extraction module;

[0031] Inputting the output result of the spatial feature extraction module into the temporal feature extraction module to obtain the output result of the temporal feature extraction module;

[0032] The output result of the time feature extraction module is input into the classification module to obtain the cognitive state output by the cognitive state assessment model.

[0033] In some embodiments, regulating the target cognitive state of the current subject according to the cognitive state output by the cognitive state assessment model includes:

[0034] obtain a few-electrode electroencephalogram signal of a current subject;

[0035] filter the few-electrode electroencephalogram signal of the current subject to obtain a few-electrode electroencephalogram signal of the same frequency band as the target few-electrode electroencephalogram signal;

[0036] calculate a first time-frequency feature according to the few-electrode electroencephalogram signal of the same frequency band as the target few-electrode electroencephalogram signal;

[0037] regulate a target cognitive state of the current subject according to the cognitive state output by the cognitive state evaluation model and the first time-frequency feature.

[0038] In some embodiments, the regulating the target cognitive state of the current subject according to the cognitive state output by the cognitive state evaluation model and the first time-frequency feature comprises:

[0039] obtain the cognitive state output by the cognitive state evaluation model in a plurality of test times;

[0040] select, from the cognitive state output by the cognitive state evaluation model in the plurality of test times, an electroencephalogram signal corresponding to each cognitive state that is correctly predicted;

[0041] calculate a second time-frequency feature corresponding to each cognitive state according to the electroencephalogram signal corresponding to each cognitive state that is correctly predicted;

[0042] compare the second time-frequency feature with the first time-frequency feature to obtain a difference value;

[0043] adjust the target cognitive state of the current subject according to the difference value.

[0044] In a second aspect, the embodiments of the present application further provide a cognitive state evaluation and regulation system of a few-electrode electroencephalogram signal, and the system comprises:

[0045] a data acquisition unit configured to acquire simultaneously collected scalp electroencephalogram signals, functional magnetic resonance imaging data, and structural magnetic resonance imaging data;

[0046] a data processing unit configured to pre-process the structural magnetic resonance imaging data to obtain pre-processed structural magnetic resonance imaging data, pre-process the functional magnetic resonance imaging data according to the pre-processed structural magnetic resonance imaging data to obtain pre-processed functional magnetic resonance imaging data, and pre-process the scalp electroencephalogram signals to obtain pre-processed scalp electroencephalogram signals;

[0047] The data selecting unit is configured to select a blood oxygen level dependent signal feature set related to a preset cognitive state according to the preprocessed functional magnetic resonance imaging data, wherein the blood oxygen level dependent signal feature set comprises at least one of a brain region blood oxygen level dependent signal, a neural circuit blood oxygen level dependent signal, and a brain network blood oxygen level dependent signal.

[0048] The model training unit is configured to take the blood oxygen level dependent signal feature set as a label of the preprocessed scalp electroencephalogram signal, and determine a trained first prediction model based on the label and the preprocessed scalp electroencephalogram signal.

[0049] The model output unit is configured to input a target scalp electroencephalogram signal into the trained first prediction model to obtain a target few-electrode electroencephalogram signal, wherein the target few-electrode electroencephalogram signal is an electroencephalogram signal of a small number of electrodes related to the label.

[0050] The data evaluation unit is configured to filter the target few-electrode electroencephalogram signal, and input a filtered target few-electrode electroencephalogram signal into the constructed cognitive state evaluation model to obtain a cognitive state output by the cognitive state evaluation model.

[0051] The data regulation unit is configured to regulate a target cognitive state of a current subject according to the cognitive state output by the cognitive state evaluation model.

[0052] In a third aspect, an electronic device is provided, which includes at least one control processor and a memory in communication connection with the at least one control processor; the memory stores instructions executable by the at least one control processor, and the instructions are executed by the at least one control processor to enable the at least one control processor to perform the cognitive state evaluation and regulation method of the few-electrode electroencephalogram signal as described above.

[0053] In a fourth aspect, a computer readable storage medium is provided, which stores computer executable instructions for causing a computer to perform the cognitive state evaluation and regulation method of the few-electrode electroencephalogram signal as described above.

[0054] It can be understood that the beneficial effects of the second aspect to the fourth aspect compared with the related art are the same as the beneficial effects of the first aspect compared with the related art, and reference can be made to the related description in the first aspect, which will not be repeated here. BRIEF DESCRIPTION OF DRAWINGS

[0055] The above and / or additional aspects and advantages of the present application will become apparent and more readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings, in which:

[0056] Figure 1 is a flowchart of an embodiment of the cognitive state evaluation and regulation method of the few-electrode electroencephalogram signal provided in the present application;

[0057] Figure 2 is a flowchart of the offline training stage and the online testing stage in an embodiment of the cognitive state evaluation and regulation method of the few-electrode electroencephalogram signal provided in the present application;

[0058] Figure 3 is a flowchart of the first prediction model in an embodiment of the cognitive state evaluation and regulation method of the few-electrode electroencephalogram signal provided in the present application;

[0059] Figure 4 is a flowchart of the cognitive state evaluation model in an embodiment of the cognitive state evaluation and regulation method of the few-electrode electroencephalogram signal provided in the present application;

[0060] Figure 5 is a structural schematic diagram of an embodiment of the cognitive state evaluation and regulation system of the few-electrode electroencephalogram signal provided in the present application. DETAILED DESCRIPTION

[0061] The embodiments of the present application are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be understood as a limitation of the present application.

[0062] In the description of the present application, if there is a description to the first, the second, etc., it is only for the purpose of distinguishing technical features, and cannot be understood as indicating or implying the relative importance of the indicated technical features or implicitly indicating the number of the indicated technical features or the sequence of the indicated technical features.

[0063] In the description of the present application, it is to be understood that the orientation description, such as the orientation or position relationship indicated by up, down, etc., is based on the orientation or position relationship shown in the drawings, and is only for the convenience of describing the present application and simplifying the description, and therefore cannot be understood as indicating or implying that the indicated device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present application.

[0064] In the description of the present application, it is to be noted that, unless otherwise explicitly limited, the words such as setting, installing, connecting, etc. should be broadly understood, and the person skilled in the art can reasonably determine the specific meaning of the above words in the present application in combination with the specific content of the technical solution.

[0065] Cognitive ability is related to specific brain regions, neural circuits or brain networks of human brain. As one of the most commonly used non-invasive brain imaging techniques, nuclear magnetic resonance has very high spatial resolution and can accurately locate cognitive-related brain regions. However, the nuclear magnetic resonance device has poor portability and is expensive. As another most commonly used non-invasive brain imaging signal, electroencephalogram has high temporal resolution, good portability and economy. However, due to the low spatial resolution, it cannot accurately locate the neural signals of specific brain regions or neural circuits related to cognitive tasks. The electroencephalogram source tracing method can locate the functional brain regions related to specific tasks, but generally requires 64 or more electrodes, which has poor economy and portability.

[0066] Therefore, how to recover the blood oxygen level dependent signal of the specific brain region related to cognition through the few-electrode electroencephalogram signal, so as to realize the evaluation of cognitive state using the few-electrode electroencephalogram signal and the online cognitive regulation is a problem to be solved.

[0067] To solve the problem of how to recover the blood oxygen level dependent signal of the specific brain region related to cognition through the few-electrode electroencephalogram signal, so as to improve the accuracy of cognitive state evaluation using the few-electrode electroencephalogram signal and improve the spatial specificity of online cognitive regulation, the application provides a cognitive state evaluation and regulation method, system and device of few-electrode electroencephalogram signal.

[0068] Reference Figure 1 The embodiment of the application provides a cognitive state evaluation and regulation method of few-electrode electroencephalogram signal, which comprises the following steps:

[0069] Step S100, acquiring the scalp electroencephalogram signal, functional magnetic resonance imaging data and structural magnetic resonance imaging data collected synchronously;

[0070] Step S200, preprocessing the structural magnetic resonance imaging data to obtain preprocessed structural magnetic resonance imaging data; preprocessing the functional magnetic resonance imaging data according to the preprocessed structural magnetic resonance imaging data to obtain preprocessed functional magnetic resonance imaging data; and preprocessing the scalp electroencephalogram signal to obtain preprocessed scalp electroencephalogram signal;

[0071] Step S300, selecting a blood oxygen level dependent signal feature set related to a preset cognitive state according to the preprocessed functional magnetic resonance imaging data, wherein the blood oxygen level dependent signal feature set comprises at least one of brain region blood oxygen level dependent signal, neural circuit blood oxygen level dependent signal and brain network blood oxygen level dependent signal;

[0072] Step S400, taking the blood oxygen level dependent signal feature set as a label of the preprocessed scalp electroencephalogram signal, and determining a trained first prediction model based on the label and the preprocessed scalp electroencephalogram signal;

[0073] Step S500, input the target scalp EEG signal to the trained first prediction model to obtain a target few-electrode EEG signal, wherein the target few-electrode EEG signal is a few-electrode EEG signal related to the label;

[0074] Step S600, filtering the target few-electrode EEG signal, and inputting the filtered target few-electrode EEG signal to the constructed cognitive state evaluation model to obtain a cognitive state output by the cognitive state evaluation model;

[0075] Step S700, regulating the target cognitive state of the current subject according to the cognitive state output by the cognitive state evaluation model.

[0076] In the embodiment, by preprocessing the structural magnetic resonance imaging data, preprocessing the functional magnetic resonance imaging data, and preprocessing the scalp EEG signal, a good data basis can be provided for later training of the first prediction model; according to the preprocessed functional magnetic resonance imaging data, a blood oxygen level dependent signal feature set related to the preset cognitive state is selected, the blood oxygen level dependent signal feature set is taken as a label of the preprocessed scalp EEG signal, and based on the label and the preprocessed scalp EEG signal, a trained first prediction model is determined; the target scalp EEG signal is input to the trained first prediction model to obtain a target few-electrode EEG signal; the first prediction model is trained by associating the brain region blood oxygen level dependent signal, the neural circuit blood oxygen level dependent signal, and the brain network blood oxygen level dependent signal with the preprocessed scalp EEG signal, so that the trained first prediction model can predict the EEG signal related to the brain region blood oxygen level dependent signal, the neural circuit blood oxygen level dependent signal, and the brain network blood oxygen level dependent signal according to the collected EEG signal; the target few-electrode EEG signal is filtered, and the filtered target few-electrode EEG signal is input to the constructed cognitive state evaluation model to obtain a cognitive state output by the cognitive state evaluation model; the target cognitive state of the current subject is regulated according to the cognitive state output by the cognitive state evaluation model, which can improve the recovery of the blood oxygen level dependent signal of the cognitive-related specific brain region by the few-electrode EEG signal, thereby improving the accuracy of the cognitive state evaluation using the few-electrode EEG signal and improving the spatial specificity of the online cognitive regulation.

[0077] In some embodiments, according to the preprocessed functional magnetic resonance imaging data, the blood oxygen level dependent signal feature set related to the preset cognitive state is selected, including:

[0078] The preprocessed functional magnetic resonance imaging data of the whole brain is divided into a plurality of blood oxygen level dependent signal feature sets, wherein the functional magnetic resonance imaging data includes at least one of a brain region functional magnetic resonance imaging signal, a neural circuit functional magnetic resonance imaging signal, and a brain network functional magnetic resonance imaging signal.

[0079] The multivariate pattern analysis method is used to select a blood oxygen level dependent signal feature set related to the preset cognitive state from a plurality of blood oxygen level dependent signal feature sets.

[0080] In the embodiment, the blood oxygen level dependent signal feature set composed of the blood oxygen level dependent signals of the brain region, neural circuit or brain network related to the preset cognitive state can lay a good data foundation for training the first prediction model later.

[0081] In some embodiments, based on the label and the preprocessed scalp electroencephalogram signal, a trained first prediction model is determined, including:

[0082] The preprocessed scalp electroencephalogram signal is filtered to obtain a multi-band scalp electroencephalogram signal;

[0083] The multi-band scalp electroencephalogram signal is obtained by using a sliding window method;

[0084] The multi-time period and multi-band electroencephalogram signal is input into the first prediction model, and the first prediction model is trained according to the label to obtain the trained first prediction model.

[0085] In the embodiment, the brain region blood oxygen level dependent signal, neural circuit blood oxygen level dependent signal and brain network blood oxygen level dependent signal are used as the label for model training, and the first prediction model is trained by associating the preprocessed scalp electroencephalogram signal, so that the trained first prediction model can predict the electroencephalogram signal related to the brain region blood oxygen level dependent signal, neural circuit blood oxygen level dependent signal and brain network blood oxygen level dependent signal according to the collected electroencephalogram signal.

[0086] In some embodiments, the first prediction model includes a channel attention layer, a spatial attention layer and a neural network prediction layer, the multi-time period and multi-band electroencephalogram signal is input into the first prediction model, the first prediction model is trained according to the label, and the trained first prediction model is obtained, including:

[0087] The multi-time period and multi-band electroencephalogram signal is input into the channel attention layer to obtain an output result of the channel attention layer;

[0088] The output result of the channel attention layer is input into the spatial attention layer to obtain an output result of the spatial attention layer;

[0089] The output result of the spatial attention layer is input into the neural network prediction layer to obtain an output result of the neural network prediction layer;

[0090] The first loss function is constructed according to the output result of the neural network prediction layer and the label;

[0091] The first prediction model is trained according to the first loss function until the first loss function converges or a preset number of iterations is reached, and a trained first prediction model is obtained.

[0092] In the embodiment, the first loss function is constructed according to the output result of the neural network prediction layer and the label, and then the first prediction model is trained according to the first loss function until the first loss function converges or a preset number of iterations is reached, and a trained first prediction model is obtained. The trained first prediction model can be used to predict the few-electrode electroencephalogram signal related to the brain region blood oxygen level dependent signal, the neural circuit blood oxygen level dependent signal and the brain network blood oxygen level dependent signal, and improve the prediction accuracy of the first prediction model.

[0093] In some embodiments, the cognitive state evaluation model includes a time feature extraction module, a spatial feature extraction module and a classification module, and the filtered target few-electrode electroencephalogram signal is input into the constructed cognitive state evaluation model to obtain the cognitive state output by the cognitive state evaluation model, including:

[0094] The filtered target few-electrode electroencephalogram signal is input into the spatial feature extraction module to obtain the output result of the spatial feature extraction module;

[0095] The output result of the spatial feature extraction module is input into the time feature extraction module to obtain the output result of the time feature extraction module;

[0096] The output result of the time feature extraction module is input into the classification module to obtain the cognitive state output by the cognitive state evaluation model.

[0097] In the embodiment, the filtered target few-electrode electroencephalogram signal is processed through the time feature extraction module, the spatial feature extraction module and the classification module, and a more accurate cognitive state can be obtained, and the evaluation of the cognitive state using the few-electrode electroencephalogram signal is realized.

[0098] In some embodiments, the target cognitive state of the current subject is regulated according to the cognitive state output by the cognitive state evaluation model, including:

[0099] The few-electrode electroencephalogram signal of the current subject is obtained;

[0100] The few-electrode electroencephalogram signal of the current subject is filtered to obtain a few-electrode electroencephalogram signal in the same frequency band as the target few-electrode electroencephalogram signal;

[0101] The first time-frequency feature is calculated according to the few-electrode electroencephalogram signal in the same frequency band as the target few-electrode electroencephalogram signal;

[0102] The target cognitive state of the current subject is regulated according to the cognitive state output by the cognitive state evaluation model and the first time-frequency feature.

[0103] In the embodiment, the target cognitive state of the current subject is regulated according to the cognitive state output by the cognitive state evaluation model and the first time-frequency feature, and the spatial specificity of the cognitive state evaluation and the cognitive state regulation is further improved by training the prediction model according to the accurate data obtained in the later period.

[0104] In some embodiments, regulating the target cognitive state of the current subject according to the cognitive state output by the cognitive state evaluation model and the first time-frequency feature comprises:

[0105] obtaining the cognitive state output by the cognitive state evaluation model in multiple test times;

[0106] selecting the electroencephalogram corresponding to each cognitive state correctly predicted from the cognitive state output by the cognitive state evaluation model in multiple test times;

[0107] calculating the second time-frequency feature corresponding to each cognitive state according to the electroencephalogram corresponding to each cognitive state correctly predicted;

[0108] comparing the second time-frequency feature with the first time-frequency feature to obtain a difference value;

[0109] adjusting the target cognitive state of the current subject according to the difference value.

[0110] In the embodiment, the second time-frequency feature corresponding to each cognitive state is calculated according to the electroencephalogram corresponding to each cognitive state correctly predicted, then the second time-frequency feature is compared with the first time-frequency feature to obtain a difference value, and finally the target cognitive state of the current subject is adjusted according to the difference value, so that the spatial specificity of the target cognitive state regulation is improved.

[0111] For the convenience of those skilled in the art, a set of best embodiments is provided as follows:

[0112] As shown in Figure 2 , the embodiments of the present application are divided into two stages of offline training and online testing. The offline stage trains a prediction model (i.e., a first prediction model) of scalp electroencephalogram and specific brain region BOLD signal (i.e., blood oxygen level dependent signal) and a cognitive state evaluation model of few-electrode electroencephalogram, and the online testing stage acquires few-electrode electroencephalogram signals of the subject in different cognitive states in real time, evaluates the cognitive state, and constructs visual / auditory feedback signals for neural feedback training. Specifically:

[0113] The offline training stage obtains multi-modal brain imaging data containing simultaneously collected scalp electroencephalogram signals, functional magnetic resonance imaging data and structural magnetic resonance imaging data by designing an experimental paradigm (which can be a task paradigm of having the subjects watch multiple types of emotional video); pre-processes the multi-modal imaging data to obtain pre-processed multi-modal imaging data; uses the fMRI signals in the pre-processed multi-modal imaging data (i.e., pre-processed functional magnetic resonance imaging data) to locate the BOLD signals, neural circuit BOLD signals or brain network BOLD signals of specific brain regions related to specific cognition by using a multivariate pattern analysis method; uses a deep neural network to establish a first prediction model of the association between the scalp electroencephalogram signals and the BOLD signals of specific brain regions, neural circuit BOLD signals or brain network BOLD signals, and finds the few-electrode electroencephalogram signals of the electrode channels and frequency bands that are more important for predicting blood oxygen level dependent signals (i.e., BOLD signals); and uses the found few-electrode electroencephalogram signals of specific electrode channels and frequency bands to establish a few-electrode electroencephalogram cognitive state evaluation model with the cognitive state under a specific experimental paradigm. The online testing stage collects few-electrode electroencephalogram signals in real time under different cognitive paradigms, calculates the electroencephalogram signals in specific frequency bands through pre-processing filtering, inputs the trained cognitive state evaluation model to determine the cognitive state, and constructs visual / auditory feedback neural feedback signals according to the determined cognitive state for cognitive enhancement.

[0114] The embodiment specifically includes the following steps:

[0115] 1. Designing an experimental paradigm to obtain multi-modal brain imaging data containing simultaneously collected scalp electroencephalogram signals, functional magnetic resonance imaging data and structural magnetic resonance imaging data.

[0116] The acquisition device used can be a nuclear magnetic resonance scanner, and the fast gradient echo method is selected. Before scanning, the scanning parameters such as scanning period, layer thickness, number of layers, scanning matrix and scanning field of view need to be set. First, the individual structural magnetic resonance imaging data of the subject is collected, and then the functional magnetic resonance imaging data and scalp electroencephalogram signals under the condition that the subject watches four types of emotional video stimuli paradigm are simultaneously collected. During the scanning process, the subject needs to wear a 64-channel magnetic electrode cap, and the subject is required to keep his head still. The scalp electroencephalogram signal data is collected at the same time as the functional magnetic resonance imaging data. The placement position of the electroencephalogram electrode is arranged according to the international electrode 10-10 system. During the collection process, the subject repeatedly watches emotional stimulus videos, and the videos contain four types of emotions: happy, sad, neutral and fear.

[0117] 2. Pre-processing the multi-modal imaging data.

[0118] (1) Pre-process the structural magnetic resonance imaging data, including head motion correction, registration, segmentation, spatial normalization and smoothing, which is realized by the tool package Statistic Parametric Mapping (SPM12) in MATLAB.

[0119] (2) Pre-process the scalp electroencephalogram signal, according to the marker of the scalp electroencephalogram signal at the beginning of fMRI acquisition, extract the scalp electroencephalogram signal in the fMRI scanning period, calculate the noise template, subtract the fMRI gradient noise template contained in the scalp electroencephalogram signal using adaptive artifact reduction, and remove the gradient artifact of the EEG signal using principal component analysis method. The above pre-processing process is realized in the software Brain Vision; baseline correction is performed on the electroencephalogram signal, band-pass filtering (0.1 Hz to 45 Hz) is performed, and independent component analysis method is used to remove the influence of electrooculogram, electrocardiogram, head motion and other noises, which is completed by the MATLAB tool package eeglab.

[0120] (3) The pre-processing of functional magnetic resonance imaging data includes: first, interlayer time correction and head motion correction; then, registration of functional magnetic resonance image and structural magnetic resonance image, mapping all data to MNI (Montreal Neurological Institute) standard space according to the segmentation information of structural magnetic resonance image; using a specified frequency band space to remove linear drift in time domain filtering of functional magnetic resonance imaging data, and the specified frequency band interval in this embodiment is 0.01 Hz to 0.08 Hz. The above pre-processing process is realized by the tool package SPM12 in MATLAB.

[0121] It should be noted that when the functional magnetic resonance imaging signal starts scanning, a marker of starting scanning is marked on the scalp electroencephalogram signal. This can ensure that the functional magnetic resonance imaging data and the scalp electroencephalogram signal are collected at the same time.

[0122] The pre-processed structural magnetic resonance imaging data and the pre-processed functional magnetic resonance imaging data are registered, and the functional magnetic resonance data is mapped to the MNI standard space; the pre-processed functional magnetic resonance imaging data is used for step 3 to locate the BOLD signal of the brain region, neural circuit or brain network related to specific cognition; and the pre-processed scalp electroencephalogram signal is used for step 4 to establish a first prediction model of the association between the scalp electroencephalogram signal and the BOLD signal of the specific brain region, neural circuit or brain network;

[0123] 3. Locate the brain region, neural circuit or brain network related to specific cognition using fMRI signal.

[0124] The pre-processed fMRI signal (i.e. pre-processed functional magnetic resonance imaging data) is analyzed using a multivariate pattern analysis method, and a whole-brain-based method (i.e. using whole-brain fMRI signal) is used to locate the brain regions related to the emotional video stimulation paradigm. Specifically, the fMRI signal of a combination of several brain regions is selected as a variable from the whole-brain fMRI signal, and the whole-brain fMRI signal is traversed. The multivariate pattern analysis method fully considers the relationship between the feature combination pattern and the cognitive state variable, and selects a feature subset (i.e. blood oxygenation level dependent signal feature set) related to the cognitive state variable for decoding to determine the BOLD signal of the brain region, neural circuit or brain network related to the emotional video stimulation paradigm (i.e. pre-set cognitive state). The feature subset includes at least one of the selected brain region BOLD signal, neural circuit BOLD signal and brain network BOLD signal.

[0125] The BOLD signal of the brain region, neural circuit or brain network selected in step 3 is used as a label for the pre-processed scalp electroencephalogram signal. The pre-processed scalp electroencephalogram signal is input into a first prediction model for training to select the scalp electroencephalogram signal that can predict the BOLD signal of the brain region, neural circuit or brain network. Step 4 is to select certain specific scalp electroencephalogram signals from the scalp electroencephalogram signals of all electrodes through the established first prediction model of the association between the scalp electroencephalogram signal and the fMRI signal.

[0126] 4. A first prediction model of the association between the scalp electroencephalogram signal and the BOLD signal of the specific brain region, neural circuit or brain network is established using a deep neural network to find a few-electrode electroencephalogram signal (i.e. target few-electrode electroencephalogram signal) that is important for predicting the BOLD signal.

[0127] The pre-processed scalp electroencephalogram signal is band-pass filtered into five frequency bands of delta (0.5-3Hz), theta (3-8Hz), alpha (8-12Hz), beta (12-30Hz), and gamma (30-60Hz) (i.e. multi-frequency band scalp electroencephalogram signal);

[0128] The response time of the BOLD signal may vary, and the collected scalp electroencephalogram signal is aligned with the synchronously collected fMRI signal using a sliding window method, with a window length of 2s and a step length of 0.5s, to obtain a multi-time period and multi-frequency band electroencephalogram sample (i.e. multi-time period and multi-frequency band electroencephalogram signal);

[0129] The deep neural network prediction model (i.e., the first prediction model) is built by combining a channel attention layer, a spatial attention layer, and a neural network prediction layer. The input of the deep neural network prediction model is the multi-time period and multi-frequency band electroencephalogram signal, and the output of the deep neural network prediction model is the predicted few-electrode electroencephalogram signal (i.e., the target few-electrode electroencephalogram signal) related to the BOLD signal of a specific brain region, neural circuit, or brain network in the feature subset. The loss function (i.e., the first loss function) of the deep neural network prediction model is the root mean square error value of the predicted BOLD signal of the brain region, neural circuit, or brain network related to the few-electrode electroencephalogram signal and the corresponding real BOLD signal of the real brain region, real neural circuit, or real brain network (i.e., the brain region BOLD signal, neural circuit BOLD signal, or brain network BOLD signal in the feature subset), i.e.,

[0130] loss1 = ||y-y'|'|2

[0131] where y' represents the predicted BOLD signal of the brain region, neural circuit, or brain network related to the few-electrode electroencephalogram signal, and y represents the real BOLD signal of the real brain region, real neural circuit, or real brain network corresponding to the predicted BOLD signal of the brain region, neural circuit, or brain network.

[0132] Referring to Figure 3 , first, the collected 64-channel scalp electroencephalogram signal is subjected to band-pass filtering in the delta, theta, alpha, beta, and gamma 5 frequency bands. Then, considering the time delay of the BOLD signal, multi-frequency band electroencephalogram signals (i.e., multi-time period and multi-frequency band electroencephalogram signals) of different time periods are obtained through a sliding window method. The multi-time period and multi-frequency band electroencephalogram signals are input into the channel attention layer to attach different weights to the electroencephalogram signals of different channels. Then, the output of the channel attention layer is input into the spatial attention layer. Finally, the output of the spatial attention layer is input into the neural network prediction layer to attach different weights to the electroencephalogram signals of different frequency bands. The channel attention layer can be constructed using Squeeze-and-Excitation Networks, the spatial attention layer can be constructed using Spatial Transformer Networks, and the neural network prediction layer can be constructed using 2D-Convolut ional Neural Networks. The channel attention layer determines the electrode channel related to the BOLD signal of a specific brain region, neural circuit, or brain network, and the spatial attention layer determines the frequency band range related to the BOLD signal of a specific brain region, neural circuit, or brain network.

[0133] 5. Using the selected specific channel and frequency band few-electrode electroencephalogram signal, a few-electrode electroencephalogram cognitive state evaluation model is established for the cognitive state under a specific paradigm.

[0134] According to the step 4, the selected few-electrode electroencephalogram signals with good BOLD signal prediction effect for specific brain regions or neural circuits or brain networks and specific frequency bands are used to filter the few-electrode electroencephalogram signals selected in the execution of the emotional stimulation task paradigm, and the few-electrode electroencephalogram signals of specific frequency bands (i.e. the filtered target few-electrode electroencephalogram signals) are extracted.

[0135] Referring to Figure 4 , a cognitive state evaluation model of the few-electrode electroencephalogram signal based on a deep neural network is built, the deep neural network includes a time feature extraction module, a spatial feature extraction module and a classification module, wherein the time feature extraction module can be constructed by using a long short-term memory network, the spatial feature extraction module can be constructed by using a 2D CNN network, and the classification module can be constructed by using an MLP multi-layer perceptron network. The input of the cognitive state evaluation model is the few-electrode electroencephalogram signal of the specific frequency band, and the output of the cognitive state evaluation model is the discriminated cognitive state. Specifically, the spatial feature of the few-electrode electroencephalogram signal of the specific frequency band is extracted by the spatial feature extraction module (i.e. the output result of the spatial feature extraction module); then the spatial feature is input into the time feature extraction module to extract the feature containing the spatial feature and the time feature (i.e. the output result of the time feature extraction module), and then the feature containing the spatial feature and the time feature is input into the classification module to obtain the discriminated cognitive state (i.e. the cognitive state output by the cognitive state evaluation model).

[0136] The loss function of the deep neural network is the cross-entropy error value of the cognitive state predicted by the cognitive state evaluation model and the real cognitive state, and the correlation features of the multi-frequency few-electrode electroencephalogram signal and the cognitive state are learned by the deep neural network. Wherein, the loss function is:

[0137]

[0138] Wherein, p i represents the cognitive state output by the cognitive state evaluation model, y i represents the real cognitive state.

[0139] 6. Using the selected few-electrode electroencephalogram signals of specific channels and frequency bands, a neurofeedback regulation signal is constructed.

[0140] According to the few-electrode electroencephalogram signals selected in step 4 and the few-electrode electroencephalogram signal cognitive state evaluation model established in step 5, only the few-electrode electroencephalogram signals of the subjects in the execution of the emotional stimulation paradigm task are collected in the online experiment, and the few-electrode electroencephalogram signals are filtered and preprocessed according to the specific frequency bands determined in step 4;

[0141] The filtered electroencephalogram signal is used as a neural feedback signal, time-frequency characteristics (i.e., first time-frequency characteristics) of the electroencephalogram signal, such as power spectral density, entropy and the like, are calculated, visual and auditory feedback signals are constructed according to the calculated time-frequency characteristics, and the visual and auditory feedback signals can be presented in real time on a display screen or earphones;

[0142] According to the few-electrode electroencephalogram cognitive state evaluation model established in step 5, in offline analysis, the number of trials in which the prediction is correct in each type of cognitive state is counted, and certain time-frequency characteristics (i.e., second time-frequency characteristics) of the electroencephalogram signal in these trials are calculated, such as power spectral density, entropy and the like. The calculated time-frequency characteristics are used as the control target value of the online feedback experiment of the cognitive state, the difference between the time-frequency characteristics of the current state electroencephalogram signal of the subject and the control target value is displayed in real time, the difference between the subject's own state and the control target value is adjusted in real time according to the difference, and the subject is helped to control the current target cognitive state (i.e., any one of the four types of emotions: happy, sad, neutral and fear).

[0143] In the embodiment, by effectively fusing the multi-modal brain imaging data synchronously collected under the emotional stimulation paradigm, a first prediction model is established, and a few-electrode electroencephalogram cognitive state evaluation model is established. These models can improve the spatial specificity of cognitive state evaluation and neural regulation.

[0144] Reference Figure 5 The embodiment of the present application also provides a few-electrode electroencephalogram signal cognitive state evaluation and regulation system, which comprises a data acquisition unit 100, a data processing unit 200, a data selection unit 300, a model training unit 400, a model output unit 500, a data evaluation unit 600 and a data regulation unit 700, wherein:

[0145] The data acquisition unit 100 is used to acquire the synchronously collected scalp electroencephalogram signal, functional magnetic resonance imaging data and structural magnetic resonance imaging data;

[0146] The data processing unit 200 is used to pre-process the structural magnetic resonance imaging data to obtain pre-processed structural magnetic resonance imaging data; pre-process the functional magnetic resonance imaging data according to the pre-processed structural magnetic resonance imaging data to obtain pre-processed functional magnetic resonance imaging data; and pre-process the scalp electroencephalogram signal to obtain pre-processed scalp electroencephalogram signal;

[0147] The data selection unit 300 is used to select a blood oxygen level dependent signal feature set related to a preset cognitive state according to the pre-processed functional magnetic resonance imaging data, wherein the blood oxygen level dependent signal feature set comprises at least one of a brain region blood oxygen level dependent signal, a neural circuit blood oxygen level dependent signal and a brain network blood oxygen level dependent signal;

[0148] The model training unit 400 is configured to take the blood oxygen level dependent signal feature set as a label of the preprocessed scalp electroencephalogram signal, and determine a trained first prediction model based on the label and the preprocessed scalp electroencephalogram signal.

[0149] The model output unit 500 is configured to input a target scalp electroencephalogram signal into the trained first prediction model to obtain a target few-electrode electroencephalogram signal, where the target few-electrode electroencephalogram signal is a few-electrode electroencephalogram signal related to the label.

[0150] The data evaluation unit 600 is configured to filter the target few-electrode electroencephalogram signal, and input the filtered target few-electrode electroencephalogram signal into the constructed cognitive state evaluation model to obtain a cognitive state output by the cognitive state evaluation model.

[0151] The data regulation unit 700 is configured to regulate a target cognitive state of a current subject according to the cognitive state output by the cognitive state evaluation model.

[0152] It should be noted that, since the cognitive state evaluation and regulation system of the few-electrode electroencephalogram signal in the embodiment and the cognitive state evaluation and regulation method of the few-electrode electroencephalogram signal are based on the same inventive concept, the corresponding contents in the method embodiment are also applicable to the system embodiment, which will not be described in detail here.

[0153] The embodiments of the present application also provide an electronic device, which comprises at least one control processor and a memory connected in communication with the at least one control processor.

[0154] The memory, as a non-transitory computer readable storage medium, can be used to store non-transitory software programs and non-transitory computer executable programs. In addition, the memory can include a high-speed random access memory, and can also include a non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state memory device. In some embodiments, the memory can optionally include a memory remotely arranged with respect to the processor, and these remote memories can be connected to the processor through a network. Examples of the above-mentioned network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0155] The non-transitory software programs and instructions required for the cognitive state evaluation and regulation method of the few-electrode electroencephalogram signal are stored in the memory, and when executed by the processor, the cognitive state evaluation and regulation method of the few-electrode electroencephalogram signal in the above embodiments is executed, for example, the method steps S100 to S700 in the above description Figure 1 are executed.

[0156] The system embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separated, i.e., can be located in one place, or can be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purposes of the embodiments.

[0157] The embodiment of the present application further provides a computer readable storage medium, which stores computer executable instructions, and the computer executable instructions are executed by one or more control processors, so that the one or more control processors execute a cognitive state evaluation and regulation method of a few-electrode electroencephalogram signal in the method embodiments, for example, execute the functions of the method steps S100 to S700 in the above description. Figure 1

[0158] Those skilled in the art can understand that all or some steps in the above disclosed method and system can be implemented as software, firmware, hardware and appropriate combinations thereof. Some or all physical components can be implemented as software executed by a processor, such as a central processing unit, a digital signal processor or a microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit. Such software can be distributed on a computer readable medium, which can include computer storage media (or non-transitory media) and communication media (or transitory media). As known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. In addition, as known to those skilled in the art, communication media generally includes computer readable instructions, data structures, program modules or other data in modulated data signals such as carrier waves or other transport mechanisms, and can include any information delivery medium.

[0159] The above is a specific description of the preferred embodiments of the present application, but the embodiments of the present application are not limited to the above described embodiments. Those skilled in the art can make various equivalent modifications or replacements without departing from the spirit of the embodiments of the present application, and these equivalent modifications or replacements are all included in the scope defined by the claims of the embodiments of the present application.​

Claims

1. A method for cognitive state assessment and regulation of a few-electrode electroencephalogram signal, characterized in that, The method comprises: acquiring simultaneously collected scalp electroencephalogram signals, functional magnetic resonance imaging data and structural magnetic resonance imaging data; preprocessing the structural magnetic resonance imaging data to obtain preprocessed structural magnetic resonance imaging data, preprocessing the functional magnetic resonance imaging data according to the preprocessed structural magnetic resonance imaging data to obtain preprocessed functional magnetic resonance imaging data, and preprocessing the scalp electroencephalogram signals to obtain preprocessed scalp electroencephalogram signals; selecting a blood oxygen level dependent signal feature set related to a preset cognitive state according to the preprocessed functional magnetic resonance imaging data, wherein the blood oxygen level dependent signal feature set comprises at least one of a brain region blood oxygen level dependent signal, a neural circuit blood oxygen level dependent signal and a brain network blood oxygen level dependent signal; taking the blood oxygen level dependent signal feature set as a label of the preprocessed scalp electroencephalogram signals, and determining a trained first prediction model based on the label and the preprocessed scalp electroencephalogram signals; inputting target scalp electroencephalogram signals into the trained first prediction model to obtain target few-electrode electroencephalogram signals, wherein the target few-electrode electroencephalogram signals are electroencephalogram signals of a small number of electrodes related to the label; filtering the target few-electrode electroencephalogram signals, and inputting filtered target few-electrode electroencephalogram signals into a constructed cognitive state evaluation model to obtain a cognitive state output by the cognitive state evaluation model; regulating a target cognitive state of a current subject according to the cognitive state output by the cognitive state evaluation model.

2. The method for cognitive state assessment and regulation of few- electrode electroencephalogram signals according to claim 1, characterized in that, The method comprises: dividing preprocessed functional magnetic resonance imaging data of the whole brain into a plurality of blood oxygen level dependent signal feature sets, wherein the functional magnetic resonance imaging data comprises at least one of functional magnetic resonance imaging signals of a brain region, functional magnetic resonance imaging signals of a neural circuit and functional magnetic resonance imaging signals of a brain network; selecting a blood oxygen level dependent signal feature set related to a preset cognitive state from the plurality of blood oxygen level dependent signal feature sets by using a multivariate pattern analysis method.

3. The method for cognitive state assessment and regulation of few- electrode electroencephalogram signals according to claim 1, characterized in that, The method comprises: filtering the preprocessed scalp electroencephalogram signals to obtain multi-band scalp electroencephalogram signals; obtaining multi-time period multi-band electroencephalogram signals by using a sliding window method on the multi-band scalp electroencephalogram signals; inputting the multi-time period multi-band electroencephalogram signals into a first prediction model, training the first prediction model according to the label, and obtaining a trained first prediction model.

4. The method for cognitive state assessment and regulation of few- electrode electroencephalogram signals according to claim 3, characterized in that, The first prediction model comprises a channel attention layer, a spatial attention layer and a neural network prediction layer, and the method comprises: inputting the multi-time period multi-band electroencephalogram signals into the channel attention layer to obtain an output result of the channel attention layer; inputting the output result of the channel attention layer into the spatial attention layer to obtain an output result of the spatial attention layer; inputting the output result of the spatial attention layer into the neural network prediction layer to obtain an output result of the neural network prediction layer; constructing a first loss function according to the output result of the neural network prediction layer and the label; training the first prediction model according to the first loss function until the first loss function converges or a preset iteration number is reached, to obtain a trained first prediction model.

5. The method for cognitive state assessment and regulation of few- electrode electroencephalogram signals according to claim 1, characterized in that, The cognitive state evaluation model comprises a time feature extraction module, a spatial feature extraction module and a classification module, and the filtered target few-electrode electroencephalogram signal is input into the constructed cognitive state evaluation model to obtain a cognitive state output by the cognitive state evaluation model, comprising: inputting the filtered target few-electrode electroencephalogram signal into the spatial feature extraction module to obtain an output result of the spatial feature extraction module; inputting the output result of the spatial feature extraction module into the time feature extraction module to obtain an output result of the time feature extraction module; inputting the output result of the time feature extraction module into the classification module to obtain the cognitive state output by the cognitive state evaluation model.

6. The method for cognitive state assessment and regulation of few- electrode electroencephalogram signals according to claim 1, characterized in that, The cognitive state output by the cognitive state evaluation model is used to regulate the target cognitive state of the current subject, comprising: acquiring a few-electrode electroencephalogram signal of a current subject; filtering the few-electrode electroencephalogram signal of the current subject to obtain a few-electrode electroencephalogram signal of the same frequency band as the target few-electrode electroencephalogram signal; calculating a first time-frequency feature according to the few-electrode electroencephalogram signal of the same frequency band as the target few-electrode electroencephalogram signal; regulating the target cognitive state of the current subject according to the cognitive state output by the cognitive state evaluation model and the first time-frequency feature.

7. The method for cognitive state assessment and regulation of few- electrode electroencephalogram signals according to claim 6, characterized in that, The cognitive state output by the cognitive state evaluation model is used to regulate the target cognitive state of the current subject, comprising: acquiring cognitive states output by the cognitive state evaluation model in multiple test times; selecting, from the cognitive states output by the cognitive state evaluation model in the multiple test times, electroencephalogram signals corresponding to each cognitive state correctly predicted; calculating second time-frequency features corresponding to each cognitive state according to the electroencephalogram signals corresponding to each cognitive state correctly predicted; comparing the second time-frequency features with the first time-frequency feature to obtain a difference value; adjusting the target cognitive state of the current subject according to the difference value.

8. A cognitive state evaluation and regulation system of few-electrode electroencephalogram signals, characterized in that, The system comprises: a data acquisition unit configured to acquire simultaneously collected scalp electroencephalogram signals, functional magnetic resonance imaging data and structural magnetic resonance imaging data; a data processing unit configured to pre-process the structural magnetic resonance imaging data to obtain pre-processed structural magnetic resonance imaging data, pre-process the functional magnetic resonance imaging data based on the pre-processed structural magnetic resonance imaging data to obtain pre-processed functional magnetic resonance imaging data, and pre-process the scalp electroencephalogram signals to obtain pre-processed scalp electroencephalogram signals; The data selection unit is configured to select a blood oxygen level dependent signal feature set related to a preset cognitive state according to the preprocessed functional magnetic resonance imaging data, wherein the blood oxygen level dependent signal feature set comprises at least one of a brain region blood oxygen level dependent signal, a neural circuit blood oxygen level dependent signal, and a brain network blood oxygen level dependent signal. The model training unit is configured to take the blood oxygen level dependent signal feature set as a label of the preprocessed scalp electroencephalogram signal, and determine a trained first prediction model based on the label and the preprocessed scalp electroencephalogram signal. The model output unit is configured to input a target scalp electroencephalogram signal into the trained first prediction model to obtain a target few-electrode electroencephalogram signal, wherein the target few-electrode electroencephalogram signal is a few-electrode electroencephalogram signal related to the label. The data evaluation unit is configured to filter the target few-electrode electroencephalogram signal, and input a filtered target few-electrode electroencephalogram signal into a constructed cognitive state evaluation model to obtain a cognitive state output by the cognitive state evaluation model. The data regulation unit is configured to regulate a target cognitive state of a current subject according to the cognitive state output by the cognitive state evaluation model.

9. An electronic device, comprising: The at least one control processor and the memory connected in communication with the at least one control processor; the memory stores instructions executable by the at least one control processor, and the instructions are executed by the at least one control processor to enable the at least one control processor to perform the cognitive state evaluation and regulation method of the few-electrode electroencephalogram signal according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions for causing a computer to perform the cognitive state evaluation and regulation method of the few-electrode electroencephalogram signal according to any one of claims 1 to 7.

Images