P300-based electroencephalogram signal analysis method and brain-computer interface system

Abstract

The invention discloses a P300-based electroencephalogram signal analysis method and an electroencephalogram interface system.The method comprises the following steps that a plurality of task targets are configured in advance, corresponding evaluation strategies are configured for all the task targets, and the evaluation strategies are used for indicating used stimulation normal forms and cognitive classification models; determining a current task target based on user operation, obtaining an evaluation strategy corresponding to the current task target, and determining a corresponding target stimulation normal form and a target cognitive classification model; performing visual stimulation and / or auditory stimulation based on the target stimulation normal form, and synchronously collecting corresponding electroencephalogram signals to obtain first electroencephalogram signals; preprocessing the first electroencephalogram signal to obtain a corresponding second electroencephalogram signal; performing feature extraction on the second electroencephalogram signal to obtain a corresponding P300 feature; and based on the obtained P300 features, performing cognitive state classification by using the target cognitive classification model to obtain a corresponding classification result. Through the design of the evaluation strategy, the method can be suitable for various scenes, and is high in universality.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of brain-computer interface technology and artificial intelligence, in particular to a P300-based electroencephalogram signal analysis method, and a P300-based brain-computer interface system. BACKGROUND

[0002] The brain-computer interface system converts neural electrical activity into executable control instructions by real-time detection of characteristic electrical signals (such as event-related potentials P300) generated by central nervous system activity, feature extraction and pattern recognition using signal processing algorithms, thereby constructing a direct human-computer interaction paradigm independent of traditional peripheral nerve-muscle pathways. The system has multiple application values such as neural function replacement, repair (stroke rehabilitation), enhancement (cognitive training), and supplementation (environmental control), and its core technologies include three key links: high-precision signal acquisition, real-time decoding algorithm, and closed-loop feedback mechanism.

[0003] The brain-computer interface (BCI) technology was initially a revolutionary breakthrough in the field of medical rehabilitation, aiming to build a direct brain-external device interaction channel for patients with motor dysfunction, enabling them to achieve prosthetic control, environmental interaction, and other basic life function reconstruction through neural electrical signal decoding. With the cross-disciplinary integration of neural decoding technology, biosensors, and artificial intelligence algorithms, modern BCI systems have broken through the traditional medical boundaries, forming a development pattern of "medical rehabilitation as the foundation, multi-field expansion": continuously deepening the application of neural function repair disease intervention in the medical field; at the same time, rapidly penetrating into emerging scenarios such as education and cognitive enhancement (attention training), consumer electronics (brain-controlled game interaction), industrial control (high-risk operation assistance), and national defense and aerospace (pilot state monitoring), showing a technical evolution path of "technology medicalization" and "technology popularization". SUMMARY

[0004] The present application provides a P300-based electroencephalogram signal analysis method and a P300-based brain-computer interface system to overcome the low universality and weak functional expansion of existing BCI technology. The evaluation strategy is designed to be suitable for multiple scenarios.

[0005] In a first aspect, a P300-based electroencephalogram signal analysis method includes the following steps:

[0006] Pre-configure a plurality of task targets, and configure a corresponding evaluation strategy for each task target, the evaluation strategy being used to indicate the used stimulation paradigm and cognitive classification model;

[0007] Determine the current task target based on user operation, obtain the evaluation strategy corresponding to the current task target, and determine the target stimulation paradigm and target cognitive classification model;

[0008] perform visual stimulation and / or auditory stimulation based on the target stimulation paradigm, and synchronously collect a corresponding electroencephalogram signal to obtain a first electroencephalogram signal;

[0009] perform preprocessing on the first electroencephalogram signal to obtain a corresponding second electroencephalogram signal;

[0010] perform feature extraction on the second electroencephalogram signal to obtain a corresponding P300 feature;

[0011] perform cognitive state classification based on the obtained P300 feature by using the target cognitive classification model to obtain a corresponding classification result.

[0012] As an implementable manner:

[0013] divide the second electroencephalogram signal into a plurality of electroencephalogram signal segments according to the target stimulation paradigm, evaluate the concentration level corresponding to each electroencephalogram signal segment, and based on the evaluation result and a preset over-limit duration, obtain a plurality of electroencephalogram signal segments as target signal segments for superposition averaging according to a preset superposition data amount to obtain a corresponding ERP waveform;

[0014] perform feature extraction on the ERP waveform to obtain a corresponding P300 feature.

[0015] As an implementable manner, the step of obtaining the target signal segment corresponding to the current feature extraction stage is:

[0016] evaluate the concentration level corresponding to the electroencephalogram signal segment and update the signal accumulation duration corresponding to the current feature extraction stage;

[0017] when the concentration level reaches a preset first threshold value, determine that the corresponding concentration level is excellent, and mark the corresponding electroencephalogram signal segment as an optimal signal segment;

[0018] when the concentration level reaches a preset second threshold value but is less than the first threshold value, determine that the corresponding concentration level is good, and mark the corresponding electroencephalogram signal segment as a candidate signal segment;

[0019] when the concentration level does not reach the second threshold value, determine that the corresponding concentration level is poor, and discard the corresponding electroencephalogram signal segment;

[0020] when the number of optimal signal segments reaches a preset superposition data amount, the corresponding optimal signal segment is used as a target signal segment;

[0021] when the signal accumulation duration reaches the over-limit duration, the corresponding optimal signal segment is used as a target signal segment, and a plurality of candidate signal segments are selected as target signal segments, so that the number of target signal segments meets the superposition data amount.

[0022] As an implementable mode:

[0023] The focus degree is calculated based on the alpha wave power, beta wave power and theta wave power corresponding to the brain electrical signal segment;

[0024] The baseline focus degree is calculated based on the alpha wave power, beta wave power and theta wave power corresponding to the user's focus state;

[0025] The baseline focus degree is weighted by the focus evaluation coefficient, and a corresponding second threshold is calculated and obtained, the value range of the focus evaluation coefficient is [0.5, 1].

[0026] As an implementable mode, the preprocessing mode includes:

[0027] A Butterworth band-pass filter is used to filter based on the existing published adaptive filtering technology;

[0028] Blind source separation technology is used to eliminate electrooculogram / electromyogram artifacts;

[0029] The signal drift is eliminated based on the baseline correction algorithm. As an implementable mode:

[0030] The evaluation strategy is also used to indicate the corresponding evaluation index, and the evaluation index includes the task completion degree index, the cognitive load index, and the neural response characteristic index:

[0031] Based on the task completion degree index, the success rate of the user's response to the target stimulation paradigm is indicated;

[0032] Based on the cognitive load index, the cognitive load level borne by the user in real time is indicated;

[0033] Based on the neural response characteristic index, the fluctuation degree of P300 characteristics is indicated.

[0034] As an implementable mode, the evaluation index further includes a pathological index for indicating a pathological evaluation result, and the evaluation mode of the pathological evaluation result includes:

[0035] Based on the pathological index, the corresponding abnormal mode feature is extracted from the second brain electrical signal, and the abnormal mode feature is matched with each sample mode feature in the preset pathological mode library, and the corresponding pathological evaluation result is generated based on the matching result.

[0036] As an implementable mode:

[0037] The stimulation paradigm includes the Oddball paradigm, the row-column matrix, and the motion trigger paradigm;

[0038] The cognitive classification model corresponding to the Oddball paradigm adopts an LDA algorithm; the cognitive classification model corresponding to the row-column matrix adopts a CNN algorithm; and the cognitive classification model corresponding to the motion trigger paradigm adopts an SVM.

[0039] The second application is a P300-based brain-computer interface system, comprising:

[0040] A storage module is configured to store a plurality of stimulation paradigms, and is further configured to store a plurality of cognitive classification models and an evaluation strategy corresponding to a task target, wherein the evaluation strategy is used to indicate the stimulation paradigms and the cognitive classification models used.

[0041] A display feedback module is configured to collect user operations, obtain a current task target, and perform visual stimulation and / or auditory stimulation based on a target stimulation paradigm corresponding to the current task target.

[0042] An electroencephalogram signal acquisition module is configured to acquire electroencephalogram signals of a user responding to the target stimulation paradigm, and obtain first electroencephalogram signals.

[0043] The electroencephalogram signal analysis module comprises:

[0044] A preprocessing submodule is configured to preprocess the first electroencephalogram signals, and obtain corresponding second electroencephalogram signals.

[0045] A feature extraction submodule is configured to extract features from the second electroencephalogram signals, and obtain corresponding P300 features.

[0046] A classification processing submodule is configured to perform cognitive state classification based on the obtained P300 features and a target cognitive classification model corresponding to the current task target, and obtain a corresponding classification result.

[0047] As an implementation manner, the feature extraction submodule comprises:

[0048] A differential processing unit is configured to divide the second electroencephalogram signals into a plurality of electroencephalogram signal segments according to the target stimulation paradigm, evaluate the concentration degree corresponding to each electroencephalogram signal segment, and based on the evaluation result and a preset over-limit time length, obtain a plurality of electroencephalogram signal segments as target signal segments for superimposed averaging according to a preset superimposed data amount, and obtain a corresponding ERP waveform.

[0049] A feature extraction unit is configured to extract features from the ERP waveform, and obtain corresponding P300 features.

[0050] The application has the following technical effects:

[0051] The prior art brain-computer interface can only complete a single function through signal acquisition, and the application can automatically call corresponding stimulation paradigms and algorithm models (such as cognitive classification models) according to the specific application scene of a user through the design of an evaluation strategy, has a multi-scene application function, and realizes a scene set. BRIEF DESCRIPTION OF DRAWINGS

[0052] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description only constitute some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0053] Figure 1 is a flowchart of the P300-based electroencephalogram signal analysis method of the present application;

[0054] Figure 2 is a module schematic diagram of the P300-based brain-computer interface system of the present application. DETAILED DESCRIPTION

[0055] The present application will be further described in detail below in combination with embodiments, and the following embodiments are used to explain the present application but the present application is not limited to the following embodiments.

[0056] As an implementable manner of the P300-based electroencephalogram signal analysis method proposed in the present application, the following steps are included:

[0057] S100, data preparation;

[0058] S110, presetting a stimulation paradigm;

[0059] The stimulation paradigm includes an Oddball paradigm, a row-column matrix, and a motion trigger paradigm;

[0060] Those skilled in the art can select several existing publicly disclosed stimulation paradigms according to actual needs, and the present embodiment does not limit them in detail.

[0061] S120, presetting an evaluation strategy;

[0062] A plurality of task targets are preconfigured, and corresponding evaluation strategies are configured for each task target, the evaluation strategy is used to indicate the task configuration corresponding to the current task, such as a stimulation paradigm, a duration, etc., and is also used to indicate the evaluation index related to the execution performance of the current task by a feedback user, that is, a corresponding evaluation index, and an algorithm model for obtaining the evaluation index.

[0063] S121, a task target;

[0064] The task target corresponds to a specific application scenario selectable by the user, and attention is paid to function evaluation, working memory training, sensory-motor integration analysis, etc. Those skilled in the art can pre-configure the corresponding task target according to the application scenario involved, such as a training task, a state monitoring task, etc., which is not limited in detail in the specification.

[0065] S122, evaluating the index and the algorithm model;

[0066] Cognitive state indicator :

[0067] A corresponding cognitive state classification model is pre-constructed, the input of the model is the P300 feature (latency, amplitude), and the output is the corresponding classification result;

[0068] Those skilled in the art can determine the cognitive state type and specific category corresponding to the task target according to the actual situation to evaluate the cognitive state of the user, which is not limited in detail in the embodiment;

[0069] Those skilled in the art can select an existing publicly disclosed neural network based on the characteristics of the stimulation paradigm, for example, the cognitive classification model corresponding to the Oddball paradigm in the embodiment adopts LDA (Linear Discriminant Analysis); the cognitive classification model corresponding to the row matrix adopts CNN (Convolutional Neural Network); and the cognitive classification model corresponding to the motion trigger paradigm adopts SVM (Support Vector Machine);

[0070] Under the premise that the input is the P300 feature (latency, amplitude), the classifiable category of the cognitive state is determined based on the actual situation, and the neural network is determined based on the stimulation paradigm, those skilled in the art can train the corresponding cognitive state classification model based on the existing publicly disclosed training method.

[0071] Task completion indicator :

[0072] The task completion degree index is used to indicate the success rate of the user in responding to the target stimulation paradigm;

[0073] The method for calculating the task completion degree in real time in the embodiment is as follows:

[0074] Based on the operation of the user, the correct selection rate and the average response time of the user are obtained;

[0075] The confidence corresponding to the classification result output by the cognitive classification model is obtained;

[0076] The P300 average amplitude in the P300 feature is obtained;

[0077] The correct selection rate, the average response time, the confidence and the P300 average amplitude are weighted and summed to obtain the corresponding task completion degree.

[0078] The person skilled in the art can self-allocate the weights according to actual needs, and the sum of the weights corresponding to the four items of correct selection rate, average response time, confidence and P300 average amplitude is 1, which can be selected correctly. The present specification does not limit it in detail.

[0079] Cognitive load indicator

[0080] The cognitive load indicator is used to indicate the cognitive load level borne by the user in real time.

[0081] The method for calculating the cognitive load level in real time in the embodiment is as follows:

[0082] After obtaining the P300 features, the change trend of each type of P300 feature is analyzed, such as the extension amount of the P300 latency feature, the attenuation amount of the P300 amplitude, and the change of the P300 amplitude stability.

[0083] Based on the change trend of the power spectral density of the electroencephalogram data (pre-processed electroencephalogram data) corresponding to the P300 features, such as the change trend of the theta frequency band and the change trend of the alpha frequency band.

[0084] These signal change features are integrated through a mathematical model, thereby outputting a quantitative cognitive load level to feedback the cognitive load level borne by the user in real time.

[0085] The person skilled in the art can allocate appropriate weight coefficients to the change amounts of the above-mentioned electroencephalogram features (such as the change amounts of the P300 latency, the P300 amplitude, the theta frequency band power, and the alpha frequency band power) according to actual needs; in actual application scenarios, the change amounts are weighted and summed to obtain a corresponding load score, and the cognitive load level corresponding thereto is determined based on the load score according to a pre-set grading rule.

[0086] The sum of the weight coefficients of each feature should be normalized to 1 to ensure the normalization and comparability of the calculation results; and when determining the weight, the polarity (i.e. the sign) thereof should be set according to the known relationship between the feature change and the cognitive load level: a feature positively correlated with the load (such as the extension of the P300 latency and the enhancement of the theta power) is given a positive weight; and a feature negatively correlated with the load (such as the reduction of the P300 amplitude and the weakening of the alpha power) is given a negative weight.

[0087] The person skilled in the art can also select an existing publicly disclosed neural network model according to actual needs, take the change amounts of the electroencephalogram features as inputs, take the cognitive load level classification results as outputs, and construct the corresponding data model according to the existing model construction scheme; in actual application scenarios, only the change amounts of the electroencephalogram features need to be input into the pre-trained data model, and the corresponding cognitive load level can be obtained.

[0088] The embodiment quantifies the cognitive load level of the user in real time by monitoring the speed of cognitive processing (whether the P300 latency is significantly prolonged) and the intensity of neural resource investment (whether the P300 amplitude systematically decays as the task progresses, whether the waveform stability decreases), and comprehensively and accurately analyzing the electroencephalogram features such as enhancement of slow wave (theta band) activity or inhibition of wave (alpha band) activity.

[0089] Neural response characteristic indicator :

[0090] The neural response characteristic index is used to indicate the fluctuation degree of the P300 feature.

[0091] In the embodiment, the neural response characteristic index includes the neural response intensity corresponding to the P300 amplitude feature, the reaction speed corresponding to the P300 latency, and the response stability corresponding to the fluctuation degree. A person skilled in the art can calculate and quantify the response stability by weighted summation based on the latency standard deviation and the amplitude coefficient of variation.

[0092] Pathology indicator :

[0093] The pathology index is used to indicate the pathology evaluation result, and the evaluation method of the pathology evaluation result includes:

[0094] After completing the stimulation task corresponding to the entire stimulation paradigm, the corresponding abnormal pattern feature is extracted from the second electroencephalogram signal, and the abnormal pattern feature is matched with each sample pattern feature in the preset pathology pattern library. The corresponding pathology evaluation result is generated based on the matching result.

[0095] A person skilled in the art can configure one or more evaluation indexes for the task target according to actual needs, and the embodiment does not limit them in detail.

[0096] S130, a preset differential processing strategy;

[0097] In the embodiment, the differential processing strategy is used to filter the preprocessed electroencephalogram signal based on the corresponding concentration level to determine the signal segment participating in superposition averaging;

[0098] The differential processing strategy includes:

[0099] S131, superposition data amount;

[0100] The feature extraction of the P300 potential usually needs to superimpose and analyze a plurality of continuous electroencephalogram data to obtain a clear enough signal. In the embodiment, the superposition data amount indicates the number of electroencephalogram data for each superposition analysis;

[0101] S132, over-limit time length;

[0102] Considering the limitation of real-time processing delay, the accumulated data duration should not be too long, so the embodiment avoids excessive screening affecting the real-time performance of the system by indicating the maximum signal range (the length of the maximum electroencephalogram signal) of the signal segment screening when the signal quality is poor.

[0103] S133, screening rule;

[0104] According to the corresponding stimulation paradigm, the preprocessed electroencephalogram signal is divided into a plurality of electroencephalogram signal segments, the concentration corresponding to each electroencephalogram signal segment is evaluated, and based on the evaluation result and the preset over-limit duration, a plurality of electroencephalogram signal segments are obtained as target signal segments for superposition and averaging according to the preset superposition data amount, and the corresponding ERP waveform is obtained.

[0105] Specifically:

[0106] Step 1, segmenting the preprocessed electroencephalogram signal according to the corresponding stimulation paradigm, i.e., the length required for subsequent superposition analysis, to obtain a plurality of electroencephalogram signal segments;

[0107] Step 2, evaluating the concentration corresponding to the electroencephalogram signal segment and updating the signal accumulation duration corresponding to the current feature extraction stage;

[0108] In this embodiment, the concentration corresponding to the electroencephalogram signal segment is calculated based on the alpha wave power f α , beta wave power f β and theta wave power f θ , and the calculation formula is:

[0109]

[0110] The start time of the electroencephalogram signal segment entering the current feature extraction stage is taken as the start time, and the end time of the currently evaluated electroencephalogram signal segment is taken as the end time, and the signal accumulation duration is determined based on the start time and the end time;

[0111] Step 3, determining the concentration level based on the obtained concentration, and determining the priority of the corresponding electroencephalogram signal segment based on the determination result;

[0112] When the concentration reaches a preset first threshold, it is determined that the corresponding concentration level is excellent, and the corresponding electroencephalogram signal segment is marked as a preferred signal segment;

[0113] When the concentration reaches a preset second threshold but is less than the first threshold, it is determined that the corresponding concentration level is good, and the corresponding electroencephalogram signal segment is marked as a candidate signal segment;

[0114] When the concentration does not reach the second threshold, it is determined that the corresponding concentration level is poor, and the corresponding electroencephalogram signal segment is discarded to avoid invalid analysis and error result output.

[0115] In this embodiment, the first threshold and the second threshold are determined based on the baseline concentration of the user, specifically:

[0116] The alpha wave power, beta wave power and theta wave power corresponding to the concentration state of the user are collected in advance to calculate the corresponding baseline concentration;

[0117] The baseline concentration is weighted by the concentration evaluation coefficient to obtain the corresponding second threshold, and the value range of the concentration evaluation coefficient is [0.5, 1];

[0118] The first threshold is determined based on the second threshold, and the first threshold is greater than the second threshold;

[0119] In this embodiment, the second threshold is 0.5 times the baseline concentration, and the first threshold is the baseline concentration.

[0120] Step 4, when the number of preferred signal segments reaches the preset superposition data amount, the corresponding preferred signal segment is taken as the target signal segment;

[0121] Step 5, when the signal accumulation duration reaches the super limit duration, the corresponding preferred signal segment is taken as the target signal segment, and a plurality of candidate signal segments are selected as target signal segments, so that the number of target signal segments meets the superposition data amount;

[0122] That is, when the total length of the signals participating in the concentration evaluation in the current stage exceeds the preset maximum length, in addition to using the electroencephalogram signal segment with excellent concentration level, the electroencephalogram signal segment with good concentration level is also used for supplement, so that the number of target signal segments participating in superposition analysis meets the requirement.

[0123] The skilled person can issue an alarm to the user to remind the user to adjust the state when the concentration level is poor or the amount of data with poor concentration level reaches the preset value according to the actual situation.

[0124] In summary, the design of the differentiated processing strategy in this application prioritizes the selection of preferred signal segments as core input to ensure the reliability of the analysis results. At the same time, considering the limitation of real-time processing delay, when the data accumulation amount of the preferred signal segment fails to meet the analysis requirement, the adaptive candidate signal segment is supplemented for processing, thereby achieving a dynamic balance between ensuring signal quality and controlling system delay.

[0125] The advantages brought by the differentiated processing strategy include:

[0126] It can significantly enhance the robustness and accuracy of the system, and greatly reduce the error risk of subsequent signal processing and recognition tasks by shielding noise data during low concentration periods;

[0127] Optimizing the utilization of computing resources, concentrating limited computing resources on high-quality data with the most value, improving processing efficiency and effective data throughput;

[0128] Improving user experience and interaction efficiency, real-time alarm enables users to immediately know the status decline, timely adjustment to maintain efficient interaction, while avoiding user frustration caused by invalid operations (such as misspelling) due to low-quality data;

[0129] Reducing the overall energy consumption of the system, selective processing and discarding of low-value data saves unnecessary computational load, which is beneficial to prolong the battery life of battery-powered devices.

[0130] S200, user interaction, referring to Figure 1 , including the following steps;

[0131] S210, determining the current task target based on user operation, obtaining the evaluation strategy corresponding to the current task target, determining the target stimulus paradigm and the target cognitive classification model;

[0132] As can be seen from the above, the target stimulus paradigm and the corresponding target evaluation index can be determined based on the evaluation strategy. In this embodiment, each evaluation strategy includes a cognitive state index, so the target cognitive classification model can be determined based on the evaluation strategy.

[0133] S220, visual stimulation and / or auditory stimulation based on the target stimulus paradigm, and synchronous acquisition of the corresponding electroencephalogram signal to obtain the first electroencephalogram signal;

[0134] In this embodiment, the signals of the user's parietal midline are collected. Those skilled in the art can also collect signals of the forehead, occipital lobe or other areas according to actual needs;

[0135] The stimulation presentation process and the electroencephalogram acquisition process need to be synchronized to ensure the alignment of the experimental paradigm events (such as the stimulation starting point, the reaction marker) and the EEG signal in the time dimension, providing a reliable time-locked basis for subsequent analysis.

[0136] S230, preprocessing based on the first electroencephalogram signal to obtain the corresponding second electroencephalogram signal;

[0137] In this embodiment, the preprocessing method includes:

[0138] A Butterworth band-pass filter (0.5-20Hz) is used to filter based on existing publicly available adaptive filtering technology to eliminate environmental electromagnetic interference;

[0139] The electrooculogram / electromyogram artifacts are removed by using blind source separation technology. In this embodiment, the electrooculogram (0.5-4 Hz) and electromyogram (20-100 Hz) artifacts are separated based on independent component analysis (ICA), and the main component of the electroencephalogram signal is retained.

[0140] The signal drift is eliminated based on the existing baseline correction algorithm.

[0141] S240, feature extraction is performed based on the second electroencephalogram signal to obtain a corresponding P300 feature;

[0142] In this embodiment, the second electroencephalogram signal is divided into a plurality of electroencephalogram signal segments according to a target stimulation paradigm, the concentration corresponding to each electroencephalogram signal segment is evaluated, and based on the evaluation result and a preset over-limit duration, a plurality of electroencephalogram signal segments are obtained as target signal segments for superposition averaging according to a preset superposition data amount, to obtain a corresponding ERP (event-related potential) waveform. Feature extraction is performed on the ERP waveform to obtain a corresponding P300 feature.

[0143] Specifically, the step of obtaining the target signal segment corresponding to the current feature extraction stage is:

[0144] The concentration corresponding to the electroencephalogram signal segment is evaluated and the signal accumulation duration corresponding to the current feature extraction stage is updated.

[0145] When the concentration reaches a preset first threshold value, it is determined that the corresponding concentration level is excellent, and the corresponding electroencephalogram signal segment is marked as a preferred signal segment;

[0146] When the concentration reaches a preset second threshold value but is less than the first threshold value, it is determined that the corresponding concentration level is good, and the corresponding electroencephalogram signal segment is marked as a candidate signal segment;

[0147] When the concentration does not reach the second threshold value, it is determined that the corresponding concentration level is poor, and the corresponding electroencephalogram signal segment is discarded.

[0148] When the number of preferred signal segments reaches a preset superposition data amount, the corresponding preferred signal segment is used as a target signal segment.

[0149] When the signal accumulation duration reaches the over-limit duration, the corresponding preferred signal segment is used as a target signal segment, and a plurality of candidate signal segments are selected as target signal segments, so that the number of target signal segments meets the superposition data amount.

[0150] Specifically, feature extraction is performed on the ERP waveform to obtain a corresponding P300 amplitude feature and P300 latency feature.

[0151] S250, based on the obtained P300 features, using the cognitive classification model to classify the cognitive state, and obtaining a corresponding classification result.

[0152] That is, the obtained P300 amplitude feature and P300 latency feature are input into the pre-constructed cognitive classification model, and the cognitive classification model outputs a corresponding classification result.

[0153] The construction of the cognitive classification model is not the innovation point of the present application. The innovation point of the present application lies in the design of the evaluation strategy. The task target can be determined through the selection of the user, the evaluation strategy is determined based on the task target, and the analysis of the stimulus and the electroencephalogram signal is automatically performed based on the evaluation strategy. The electroencephalogram interface system is suitable for various scenes, has high universality, various functions, and good user experience.

[0154] Further, based on the target evaluation index, the corresponding algorithm model is called to automatically generate a corresponding evaluation result.

[0155] The calculation method corresponding to each evaluation index is described in detail in the evaluation index and algorithm model in the above step S122, and therefore will not be described here.

[0156] Further, after the execution of the current task target corresponding stimulus paradigm is completed, a corresponding comprehensive feedback report is generated based on the obtained evaluation results for feedback.

[0157] As an implementable manner of the P300-based electroencephalogram interface system proposed in the present application, referring to Figure 2 , comprising:

[0158] A storage module 100 is configured to store a plurality of stimulus paradigms, and is further configured to store a plurality of cognitive classification models, and is further configured to store an evaluation strategy corresponding to a task target. The evaluation strategy is used to indicate the used stimulus paradigm and cognitive classification model.

[0159] A display feedback module 200 is configured to collect the operation of a user, obtain a current task target, and is further configured to perform visual stimulation and / or auditory stimulation based on a target stimulus paradigm corresponding to the current task target.

[0160] An electroencephalogram signal acquisition module 300 is configured to acquire the electroencephalogram signal of the user in response to the target stimulus paradigm, and obtain a first electroencephalogram signal. In the present embodiment, the electroencephalogram signal acquisition module 300 includes a first electrode (Pz / Cz) located on the midline of the parietal lobe, and further includes a second electrode (Fz) located on the forehead and / or a third electrode (Oz) located on the occipital lobe.

[0161] The electroencephalogram signal analysis module 400 comprises:

[0162] A preprocessing submodule is configured to pre-process the first electroencephalogram signal to obtain a corresponding second electroencephalogram signal.

[0163] a feature extraction sub-module, configured to extract features from the second electroencephalogram signal to obtain P300 features;

[0164] a classification processing sub-module, configured to perform cognitive state classification based on the P300 features and a target cognitive classification model corresponding to the current task target to obtain a classification result.

[0165] In this embodiment, the feature extraction sub-module includes:

[0166] a differential processing unit, configured to divide the second electroencephalogram signal into a plurality of electroencephalogram signal segments according to a target stimulation paradigm, evaluate the concentration degree corresponding to each electroencephalogram signal segment, and based on the evaluation result and a preset over-limit duration, perform superposition averaging on a plurality of electroencephalogram signal segments as target signal segments according to a preset superposition data amount to obtain an ERP waveform.

[0167] a feature extraction unit, configured to extract features from the ERP waveform to obtain P300 features.

[0168] Specifically, the differential processing unit is configured to:

[0169] evaluate the concentration degree corresponding to the electroencephalogram signal segment and update the signal accumulation duration corresponding to the current feature extraction stage, and in this embodiment, the concentration degree is calculated based on the alpha wave power, beta wave power and theta wave power corresponding to the electroencephalogram signal segment.

[0170] when the concentration degree reaches a preset first threshold, it is determined that the corresponding concentration level is excellent, and the corresponding electroencephalogram signal segment is marked as a preferred signal segment;

[0171] when the concentration degree reaches a preset second threshold but is less than the first threshold, it is determined that the corresponding concentration level is good, and the corresponding electroencephalogram signal segment is marked as a candidate signal segment;

[0172] when the concentration degree does not reach the second threshold, it is determined that the corresponding concentration level is poor, and the corresponding electroencephalogram signal segment is discarded;

[0173] when the number of preferred signal segments reaches the preset superposition data amount, the corresponding preferred signal segments are used as target signal segments;

[0174] when the signal accumulation duration reaches the over-limit duration, the corresponding preferred signal segments are used as target signal segments, and a plurality of candidate signal segments are selected as target signal segments, so that the number of target signal segments meets the superposition data amount.

[0175] For apparatus embodiments, since they are substantially similar to the method embodiments, the description is relatively simple, and the relevant parts are referred to the part of the description of the method embodiments.

[0176] Each of the embodiments in the present specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same and similar parts between the embodiments are referred to each other.

[0177] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, apparatus, or computer program product. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage, etc.) containing computer-usable program code.

[0178] The present application is described with reference to flowcharts and / or block diagrams of the method, terminal device (system), and computer program product according to the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate a means for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in one or more blocks.

[0179] These computer program instructions can also be stored in a computer-readable memory that can cause the computer or other programmable data processing terminal device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including instruction means, which implements the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in one or more blocks.

[0180] These computer program instructions can also be loaded into a computer or other programmable data processing terminal device, so that a series of operation steps are performed on the computer or other programmable terminal device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable terminal device provide a means for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in one or more blocks.

[0181] It is to be noted that:

[0182] Reference throughout this specification to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the application. Thus, the appearances of the phrases "in one embodiment" or "an embodiment" in various places throughout this specification are not necessarily all referring to the same embodiment.

[0183] Although preferred embodiments of the application have been described herein, additional changes and modifications can be made by those skilled in the art once given the benefit of the present disclosure. It is therefore intended that the appended claims shall cover all such changes and modifications as fall within the true spirit and scope of the application.

[0184] Furthermore, it is to be understood that the specific embodiments described herein can be different from one another, and that the description of the specific embodiments is not intended to be limiting. Any features described as being in one embodiment can be in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one embodiment or in any other embodiment. Any features described as being in an embodiment can be in one

Claims

1. A P300-based electroencephalogram signal analysis method, characterized by, The method comprises the following steps: a plurality of task targets are pre-configured, and an evaluation strategy corresponding to each task target is configured, the evaluation strategy being used to indicate a used stimulation paradigm and a cognitive classification model; a current task target is determined based on a user operation, an evaluation strategy corresponding to the current task target is obtained, and a target stimulation paradigm and a target cognitive classification model are determined; visual stimulation and / or auditory stimulation is performed based on the target stimulation paradigm, and a corresponding electroencephalogram signal is synchronously collected to obtain a first electroencephalogram signal; the first electroencephalogram signal is pre-processed to obtain a corresponding second electroencephalogram signal; P300 features are extracted from the second electroencephalogram signal to obtain corresponding P300 features; a cognitive state is classified by using the target cognitive classification model based on the obtained P300 features to obtain a corresponding classification result.

2. The P300-based electroencephalogram signal analysis method according to claim 1, characterized in that: the second electroencephalogram signal is divided into a plurality of electroencephalogram signal segments according to the target stimulation paradigm, the concentration of each electroencephalogram signal segment is evaluated, and based on the evaluation result and a preset over-limit time length, a preset superposition data amount is used to obtain a plurality of electroencephalogram signal segments as target signal segments for superposition averaging to obtain a corresponding ERP waveform; P300 features are extracted from the ERP waveform.

3. The P300-based electroencephalogram signal analysis method of claim 2, wherein, The steps of obtaining the target signal segments corresponding to the current feature extraction stage are as follows: the concentration of the electroencephalogram signal segment is evaluated and the signal accumulation time length corresponding to the current feature extraction stage is updated; when the concentration reaches a preset first threshold value, it is determined that the corresponding concentration level is excellent, and the corresponding electroencephalogram signal segment is marked as an optimal signal segment; when the concentration reaches a preset second threshold value but is less than the first threshold value, it is determined that the corresponding concentration level is good, and the corresponding electroencephalogram signal segment is marked as a candidate signal segment; when the concentration does not reach the second threshold value, it is determined that the corresponding concentration level is poor, and the corresponding electroencephalogram signal segment is discarded; when the number of optimal signal segments reaches the preset superposition data amount, the corresponding optimal signal segments are used as target signal segments; when the signal accumulation time length reaches the over-limit time length, the corresponding optimal signal segments are used as target signal segments, and a plurality of candidate signal segments are selected as target signal segments, so that the number of target signal segments meets the superposition data amount.

4. The P300-based electroencephalogram signal analysis method according to claim 3, characterized in that: the concentration of the electroencephalogram signal segment is calculated based on the alpha wave power, beta wave power and theta wave power corresponding to the electroencephalogram signal segment; a baseline concentration is calculated by pre-collecting the alpha wave power, beta wave power and theta wave power corresponding to the user's concentration state; the baseline concentration is weighted by using a concentration evaluation coefficient to obtain a corresponding second threshold value, and the value range of the concentration evaluation coefficient is [0.5, 1].

5. The P300-based electroencephalogram signal analysis method according to any one of claims 1 to 4, characterized in that, The pre-processing mode comprises: a Butterworth band-pass filter is used to filter based on an existing publicly disclosed adaptive filtering technology; a blind source separation technology is used to remove electrooculogram / electromyogram artifacts; signal drift is eliminated based on a baseline correction algorithm.

6. The P300-based electroencephalogram signal analysis method according to any one of claims 1 to 4, characterized in that: the evaluation strategy is further used to indicate a corresponding evaluation index, and the evaluation index comprises a task completion degree index, a cognitive load index, and a neural response characteristic index: the task completion degree index is used to indicate a success rate of a user in responding to a target stimulation paradigm; the cognitive load index is used to indicate a cognitive load level borne by the user in real time; the neural response characteristic index is used to indicate a fluctuation degree of a P300 feature.

7. The P300-based electroencephalogram signal analysis method of claim 6, wherein, the evaluation index further comprises a pathological index used to indicate a pathological evaluation result, and the evaluation manner of the pathological evaluation result comprises: based on the pathological index, a corresponding abnormal pattern feature is extracted from the second electroencephalogram signal, and the abnormal pattern feature is matched with each sample pattern feature in a preset pathological pattern library, and a corresponding pathological evaluation result is generated based on a matching result.

8. The P300-based electroencephalogram signal analysis method according to any one of claims 1 to 4, characterized in that: the stimulation paradigm comprises an Oddball paradigm, a row-column matrix, and a motion trigger paradigm; the cognitive classification model corresponding to the Oddball paradigm adopts an LDA algorithm; the cognitive classification model corresponding to the row-column matrix adopts a CNN algorithm; and the cognitive classification model corresponding to the motion trigger paradigm adopts an SVM.

9. A P300-based brain-computer interface system, characterized by comprises: a storage module configured to store a plurality of stimulation paradigms, a plurality of cognitive classification models, and an evaluation strategy corresponding to a task target, the evaluation strategy being used to indicate the stimulation paradigms and the cognitive classification models used; a display feedback module configured to collect a user's operation, acquire a current task target, and perform visual stimulation and / or auditory stimulation based on a target stimulation paradigm corresponding to the current task target; an electroencephalogram signal acquisition module configured to acquire an electroencephalogram signal of the user responding to the target stimulation paradigm, and obtain a first electroencephalogram signal; the electroencephalogram signal analysis module comprises: a preprocessing sub-module configured to pre-process the first electroencephalogram signal, and obtain a corresponding second electroencephalogram signal; a feature extraction sub-module configured to extract a feature from the second electroencephalogram signal, and obtain a corresponding P300 feature; a classification processing sub-module configured to perform cognitive state classification based on the obtained P300 feature and a target cognitive classification model corresponding to a current task target, and obtain a corresponding classification result.

10. The P300-based brain-computer interface system of claim 9, wherein, the feature extraction sub-module comprises: a differential processing unit configured to divide the second electroencephalogram signal into a plurality of electroencephalogram signal segments according to the target stimulation paradigm, evaluate a concentration degree corresponding to each electroencephalogram signal segment, and based on an evaluation result and a preset over-limit time length, obtain a plurality of electroencephalogram signal segments as target signal segments for superposition averaging according to a preset superposition data amount, and obtain a corresponding ERP waveform; a feature extraction unit configured to extract a feature from the ERP waveform, and obtain a corresponding P300 feature.

Images