CN104473636A - Brain fatigue network analysis method based on partial orientation coherence - Google Patents

Abstract

The invention discloses a brain fatigue network analysis method based on partial orientation coherence. The brain fatigue network analysis method comprises the following steps: a character of 2-back is utilized to induce a fatigue effect, electroencephalogram signals before and after fatigue are collected through an electroencephalogram collection device, and the electroencephalogram signals are preprocessed; the relationship between every two channels of the electroencephalogram signals are quantized, and the partial orientation coherence PDC value is calculated; a threshold is determined, and a cause-effect network is built; the brain network parameters are analyzed. According to the method, the brain network parameters are analyzed, so the electroencephalogram information flow orientation in the fatigue state is obtained, and the neurophysiological mechanism of fatigue is further understood. The brain fatigue network analysis method based on partial orientation coherence can also be used as a new fatigue monitoring index and provide theoretical basis for anti-fatigue.

Description

A kind of based on partially directed relevant tired brain network analysis method

Technical field

The present invention relates to analysis of network field, particularly relate to a kind of based on partially directed relevant tired brain network analysis method.

Background technology

Fatigue is declined by the operational performance that muscle power or mental work cause for a long time, subjective a kind of state with asthenia sensation.Brain fag is also called psychological fatigue, refers to a kind of amotivational subjective sensation with vigilance, and main manifestations is for having one's head in the clouds, and difficult concentrating, thinking difficulty, forgetful, desire reduces, and job performance declines and easily makes mistakes.Along with the progress of modern society, the increase of operating pressure, brain fag crowd get more and more, and therefore seems particularly important to the research of brain fag.Under fatigue state, the exception of the brain information course of processing can show in EEG signals, and information flow direction can be required mental skill, network represents.

Human brain is one of the most complicated system of occurring in nature, and in this system, multiple neuron, neuron colony or multiple brain district are interconnected to complicated network structure, and by the various functions of the brain that interacted.According to statistics, about 10 are had in adult's brain ¹¹individual neuronal cell, the neuronal cell of these enormous amount is by about 10 ¹⁵individual synapse is interconnected, and defines the brain structural network of a high complexity.This is complicated and huge network is the physiological foundation that brain carries out information processing and cognitive Expression.On the architecture basics of brain, neuronic spontaneous activity and the process of excitation being subject to environmental stimuli and producing and process of inhibition are delivered to other relevant neurons by synapse, make the carrying out that the neural activity between each neuron, between nervous system each several part can cooperatively interact, mutually coordinate.When brain is in different cognitive states, network connects also different, therefore can connect by network the change spying upon brain cognitive state.

Cerebral nerve interconnection network can be divided into structural brain network, functional brain network and because of validity brain network.Because of the cause effect relation in the statistical significance between validity brain network description node, influencing each other or information flow direction between each node of reflection brain network.At present, be still not clear for the information flow direction of fatigue state and the physiological mechanism of fatigue, the research that the brain network information flows to figure can provide a kind of new ways and means for fatigue monitoring.Because validity brain network can provide one method intuitively for the brain information course of processing under fatigue analysis state.The physiological mechanisms of current brain fag is still not clear, and brain network can be offered help for the Mechanism Study of brain fag, and contributes to the research of tired countercheck.

Summary of the invention

The invention provides a kind of based on partially directed relevant tired brain network analysis method, the present invention sets up the brain network of tired front and back, and analyzes network parameter; By the change of tired front and back network parameter, fatigue state brain information is flowed to and explores, thus contribute to explaining tired physiological mechanisms, described below:

Based on a partially directed relevant tired brain network analysis method, described tired brain network analysis method comprises the following steps:

Utilize character 2-back to bring out fatigue effect, gathered the EEG signals of tired front and back by brain wave acquisition device, pretreatment is carried out to EEG signals;

Quantize each passage of EEG signal relation between any two, calculate partially directed coherence analysis value PDC;

Definite threshold also builds because of validity network;

Brain network parameter is analyzed.

The partially directed coherence analysis value PDC of described calculating is specially:

${\mathrm{PDC}}_{x_j\→x_k}\left(f\right)=\frac{\stackrel{\‾}{a_{k,j}}\left(f\right)}{\sqrt{{\mathrm{\Σ}}_{i=1}^m{\left|\stackrel{\‾}{a_{i,j}}\left(f\right)\right|}^2}}$

Wherein, table in jth row in a kth element; represent jth row in i-th element; value be normalized, between [0,1], represent x _jflow to x _ksignal account for all from x _jthe ratio of flow-out signal.

Described definite threshold also builds because validity network is specially:

If PDC value is greater than threshold value, if the adjacency matrix element value of correspondence is 1; Otherwise, if the adjacency matrix element value of correspondence is 0; Brain network is drawn according to the matrix of binaryzation.

The beneficial effect of technical scheme provided by the invention is: this method is by the analysis to brain network parameter, and the brain electric information that can obtain fatigue state flows to, and understands tired neurophysiological mechanism further.This method also can also can provide theories integration for the antagonism of fatigue as a kind of new fatigue monitoring index.

Accompanying drawing explanation

Fig. 1 is the brain wave acquisition distribution schematic diagram that leads used;

Fig. 2 is tired front and back brain networks; A is before fatigue; B is after fatigue;

Fig. 3 is in-degree, the out-degree schematic diagram in Ge Nao district, tired front and back; A is in-degree schematic diagram; B is out-degree schematic diagram;

Fig. 4 is a kind of flow chart based on partially directed relevant tired brain network analysis method.

Detailed description of the invention

For making the object, technical solutions and advantages of the present invention clearly, below embodiment of the present invention is described further in detail.

101: utilize character 2-back to bring out fatigue effect, gathered the EEG signals of tired front and back by brain wave acquisition device, pretreatment is carried out to EEG signals;

Wherein, first utilize character 2-back to bring out fatigue, gather the EEG signals of tired front and back, brain wave acquisition device is Neuroscan4.5 system.Electrode position is laid according to international 10-20 standard, and 10-20 system electrode placement methods Shi You international electroencephalogram association specified standard electrode placement methods, electrode placement methods as shown in Figure 1.This method acquires 20 eeg datas led, and reference electrode is auris dextra mastoid process, and sample frequency is 1000Hz.Carry out pretreatment to the EEG signals collected, comprise down-sampled to 200Hz, carry out 0.1-40Hz bandpass filtering, independent component analysis (ICA) removes eye electrical interference.This part is conventionally known to one of skill in the art, and the embodiment of the present invention does not repeat this.

102: quantize each passage of EEG signal relation between any two, calculate PDC value;

First the node that electrode position is brain network is defined.The intensity connected because of validity generally adopts Causality Analysis to quantize, and the topmost analysis thought of cause effect relation is Granger Causality (GC) analysis.Based on GC thought, develop multiple in the causal method of frequency domain inner analysis, current comparative maturity mainly contain direct transfer function analysis (DTF) and partially directed coherence analysis (PDC).The present invention adopts PDC method to build the brain network of CFS patient.

When processing single channel EEG signals, suppose that EEG signals is a stable time series, and deepen EEG signals and can describe its production process with a linear filter, thus autoregression model (AR) model is used for electroencephalogramsignal signal analyzing, owing to gathering multichannel EEG signals, multichannel AR model therefore to be built.

$\left[\begin{array}{c}{x}_1\left(n\right)\\ x_2\left(n\right)\\ .\\ .\\ .\\ x_m\left(n\right)\end{array}\right]=\underset{i=1}{\overset{p}{\mathrm{\Σ}}}{C}_i\left[\begin{array}{c}{x}_1(n-i)\\ x_2(n-i)\\ .\\ .\\ .\\ x_m(n-i)\end{array}\right]+\left[\begin{array}{c}{w}_1\left(t\right)\\ w_2\left(t\right)\\ .\\ .\\ .\\ w_m\left(t\right)\end{array}\right]$

Wherein, x ₁(n), x ₂(n) ... x _mn (), represents each eeg data led respectively; P is the exponent number of regression model; x ₁(n-i), x ₂(n-i) ... x _m(n-i) represent respectively each lead before the eeg data of i time point; w ₁(t), w ₂(t) ... w _mt () represents residual sequence respectively.

Carry out Fourier transformation

$A\left(f\right)=\underset{i=1}{\overset{p}{\mathrm{\Σ}}}{C}_i{e}^{-j2\mathrm{\πif}/f_s}$

$\stackrel{\‾}{A}\left(f\right)=I-A\left(f\right)=I-\underset{i=1}{\overset{p}{\mathrm{\Σ}}}{C}_i{e}^{-j2\mathrm{\πif}/f_s}$

Wherein, the unit matrix of I to be dimension be m; A (f) represents the connection matrix in frequency domain; f _srepresent sample frequency.

PDC computing formula is

${\mathrm{PDC}}_{x_j\→x_k}\left(f\right)=\frac{\stackrel{\‾}{a_{k,j}}\left(f\right)}{\sqrt{{\mathrm{\Σ}}_{i=1}^m{\left|\stackrel{\‾}{a_{i,j}}\left(f\right)\right|}^2}}$

Wherein, represent in jth row in a kth element; represent jth row in i-th element. value be normalized, between [0,1], represent x _jflow to x _ksignal account for all from x _jthe ratio of flow-out signal.

103: definite threshold also builds because of validity network;

Estimate the significance level of PDC value, the significance threshold value estimated is got and does threshold value, (such as, by the PDC value that calculates by sorting from small to large, choose the value of the some correspondence of front 60% as threshold value), if PDC value is greater than threshold value, then think to there is significant cause effect relation between this sequence corresponding to PDC value, and set corresponding adjacency matrix element value as 1; Otherwise, then not think to there is cause effect relation, if the adjacency matrix element value of correspondence is 0.Brain network is drawn according to the matrix of binaryzation.As shown in Figure 2, in figure, each leads and represents each node network before and after tired, and the connection between node represents with being with the line of arrow, and the node of arrow points is the node of inflow.

104: brain network parameter is analyzed.

There are some topological parameters can be used for carrying out quantitative description to network in network, such as: node degree, cluster coefficient, shortest path length are several important network parameters.

(1) node degree k _i: the node degree of node i is defined as the number on the limit be connected with node i, namely

$k_i=\underset{j\∈N}{\mathrm{\Σ}}{a}_{\mathrm{ij}}$

In directed graph, node degree is divided into again in-degree and out-degree, and in-degree is the number on the limit flowing into a certain node, and out-degree is the number on the limit of flowing out from a certain node.So, the node degree of a certain node is the summation of out-degree and in-degree.In a network, the in-degree of a certain node is larger, shows that this node is larger by the impact of other nodes.Fig. 3 gives out-degree and the in-degree change in each brain district, tired front and back, and transverse axis represents each brain district, and the longitudinal axis is the summation of all node degrees in Ge Nao district.

(2) cluster coefficient is an important indicator of characterizing network grouping of the world economy degree, represents the probability size be connected to each other between two nodes being connected with some nodes.The cluster coefficient of a certain node is defined as, the limit number that between the node be connected with this node, reality is connected and the ratio of maximum possible fillet number, that is:

$A_i=\frac{{2e}_i}{k_i(k_i-1)}$

And the cluster coefficient of whole network is the meansigma methods of all node cluster coefficients in network, as shown in the formula:

$A=\frac{1}{N}\underset{i\∈V}{\mathrm{\Σ}}{A}_i$

Wherein, A _ifor the cluster coefficient of each node; V is the set of all nodes.

When being in fatigue state, there is change in the message transmission capability of brain, and the communication for information in each brain district occurs abnormal, and thus the parameter of brain network there will be change.According to the change of network parameter, can inquire into the brain information interchange of fatigue state and neuromechanism thereof.

The present invention proposes a kind of tired brain network analysis method based on PDC, by gathering the eeg data of tired front and back, set up brain network and network parameter is analyzed, the Brain electrical information flow of fatigue state can be obtained, and the neuromechanism of fatigue is explored, can resist as a kind of new fatigue monitoring method and for fatigue and theories integration is provided.Optimum implementation adopts patent transfer, technological cooperation.

It will be appreciated by those skilled in the art that accompanying drawing is the schematic diagram of a preferred embodiment, the invention described above embodiment sequence number, just to describing, does not represent the quality of embodiment.

The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, within the spirit and principles in the present invention all, any amendment done, equivalent replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (3)

1., based on a partially directed relevant tired brain network analysis method, it is characterized in that, described tired brain network analysis method comprises the following steps:

Utilize character 2-back to bring out fatigue effect, gathered the EEG signals of tired front and back by brain wave acquisition device, pretreatment is carried out to EEG signals;

Quantize each passage of EEG signal relation between any two, calculate partially directed coherence analysis value PDC;

Definite threshold also builds because of validity network;

Brain network parameter is analyzed.

2. according to claim 1 a kind of based on partially directed relevant tired brain network analysis method, it is characterized in that, the partially directed coherence analysis value PDC of described calculating is specially:

${\mathrm{PDC}}_{x_j\→x_k}\left(f\right)=\frac{\stackrel{\‾}{a_{k,j}}\left(f\right)}{\sqrt{{\mathrm{\Σ}}_{i=1}^m{\left|\stackrel{\‾}{a_{i,j}}\left(f\right)\right|}^2}}$

Wherein, represent in jth row in a kth element; represent jth row in i-th element; value be normalized, between [0,1], represent x _jflow to x _ksignal account for all from x _jthe ratio of flow-out signal.

3. a kind of tired brain network analysis method relevant based on orientation partially according to claim 1, is characterized in that, described definite threshold also builds because validity network is specially:

If PDC value is greater than threshold value, if the adjacency matrix element value of correspondence is 1; Otherwise, if the adjacency matrix element value of correspondence is 0; Brain network is drawn according to the matrix of binaryzation.