CN103006211A - Map mapping device based on brain electrical activity network analysis - Google Patents

Abstract

The invention discloses a map mapping device based on brain electrical activity network analysis. The map mapping device comprises a brain electrical activity acquisition circuit, a brain electrical activity image lead wire, a brain electrical activity preprocessing circuit, an analog-digital converting signal and a signal processor which are sequentially connected, wherein the signal processor is sequentially connected with a brain electrical activity interference filtering module, a brain electrical activity segmenting and decomposing module, a cross-approximate entropy analysis module, a threshold value analysis module, a brain network parameter computing module and a brain network comparing module. A brain electrical activity signal is utilized to extract network interconnection characteristics, the brain electrical activity signal is convenient to acquire and good in time resolution, and interference can be eliminated by a method combining hardware and software to enable the signal to be stable. The brain electrical activity acquisition circuit based on the electrical activity signal is simple, low in manufacturing cost, beneficial to popularization in families and can bring benefits to common people. Wavelet packet decomposition is combined with nonlinear methods like cross-approximate entropy and neural network analysis, so that nonlinearity and instability natures of the brain electrical activity signal are fitted better. Compared with a conventional brain electrical activity mapping, the map mapping device is capable of providing more information about brain activities.

Description

A kind of topography drawing apparatus based on the brain Electrical network analysis

Technical field

The invention belongs to technical field of information processing, relate to human body signal and process, specifically a kind of by extracting human body electroencephalogram's EEG signal and carrying out signal processing and the brain analysis of network, with the judgement of final realization brain function state.The present invention can be used for the different brain function state of human body is qualitatively judged, and provides a kind of new tool for understanding the cerebral activity state.

Background technology

Tradition brain electrical activity mapping BEAM(brain electrical activity mapping) be exactly on the electroencephalogram technical foundation, with computer the EEG signal is carried out digital processing, One-dimension Time Series is transformed into locates and quantitative pseudo-colours Two dimensional Distribution image, show the changes of function of brain and carry out preliminary form and locate with numeral and color.

At present, neuro physiology, anatomy, pathological existing research all show, on the different level such as cell, nuclear group, tissue, organ, human brain is to form the nuclear group with certain function or further form the cerebral tissue with certain physical arrangement and function by nuclear group by neuron or by neuron to consist of, between the brain Component units of these different levels, all there is complicated internet relation, some brain functioies are unusual, such as schizophrenia, its principal character is exactly the interconnected unusual of cranial nerve network.But traditional B EAM treats each electrode of brain electricity in the composition process as signal source independently, and the topography of formation does not reflect the dependency relation between a plurality of electrode for encephalograms, has open defect.

In fact, some have been arranged about the relevant brain network research of mental status analysis at present, comprised sleep, anesthesia, attention, schizophrenia, senile dementia etc.These researchs different variations occurred from the brain network that a plurality of angles have disclosed different brain function states, the result of research mostly finds, when brain function is unusual, the brain network has more the variation that has showed Small-world Characters in various degree, and the degeneration of brain network Small-world Characters means the brain network when process information, its global efficiency and local Efficiency Decreasing.

The modal signals collecting instrument of research brain network is nuclear magnetic resonance MRI device at present, relatively more typical report comprises Silvina G.Horovitz, AllenR.Braun, " the Decouplingof the brain's default mode network during deep sleep " that Walter S.Carrd etc. delivers, see PNAS, 2009(27) 11376-11381; " the applied research progress of brain network in depression " that Zhu Junjuan, Peng Daihui, Jiang Kaida etc. deliver seen Chinese neuropsychiatric disease magazine 12 phase 60-63 pages or leaves in 2011; He Y, Chen Z, EvansA. " Structural insights into aberrant topological patterns of large-scale corticalnetworks in alzheimer ' s disease " sees J Neurosci, 2008(18) 4756-66.There is following defective in these brain analysis of network based on magnetic resonance MRI, have limited its extensive use:

(1) MRI equipment and Laboratory Fee are expensive;

(2) can not use monitoring and crash equipment in the MRI machine room, MRI is moving responsive to the patient body, easily produces pseudo-shadow, is unsuitable for emergency treatment and critical patient are checked;

(3) one or two people enter the scanning room and can produce claustrophobia, and a kind of fear that is difficult to describe of private prosecution can not cooperate inspection, often causes checking unsuccessfully.

In a word, traditional B EAM isolates each electrode for encephalograms in composition, can not reflect interkniting between the electrode.The instrument of present analysis brain network adopts MRI more, but it has many deficiencies.Functional analysis and the modern brain network theory traditional B EAM plane positioning, pseudo-colours expressed based on the topography drawing apparatus of brain Electrical network analysis combine, Acquisition Circuit is simple, cheap, signal acquisition process is also very convenient, has overcome traditional B EAM and can not reflect the relevant shortcoming of information between the electrode for encephalograms.

Summary of the invention

The objective of the invention is to overcome the shortcoming of traditional B EAM and MRI, provide a kind of easy to use, the cheap novel topography drawing apparatus based on the brain analysis of network, the effectiveness that detects to improve brain information is realized the more accurate judgement of brain function state.

Technical scheme of the present invention is achieved in that

A kind of topography drawing apparatus based on the brain Electrical network analysis, comprise the brain wave acquisition electrode, electroencephalogram conducting wire, brain electricity pre-process circuit, analog to digital conversion circuit and the signal processor that connect successively, described signal processor is provided with brain electrical interference filtering module, brain electricity segmentation and the decomposing module that connects successively, mutual Analysis of Approximate Entropy module, analysis of threshold module, brain network parameter computing module and brain network contrast module; The interference of 50Hz power frequency, baseline drift and myoelectricity that brain electrical interference filtering module is used for the filtering EEG signals disturb; The brain wave extraction module be used for to extract EEG signals, carries out sliding window and processes as one section take per 1024; The Analysis of Approximate Entropy module is used for asking the mutual approximate entropy between each crosslinking electrode mutually; The analysis of threshold module is used to be set up brain network definite threshold and sets up corresponding two values matrix A; Brain network parameter computing module comprises node degree, cluster coefficients and Jie's number for setting up the brain network and calculate corresponding brain network parameter according to threshold value; Brain network contrast module is used for the comparing with reference to person's brain network parameters of experimenter's brain network parameters and pre-stored, and with the form demonstration of comparing result with topography.

Described brain electric network topography instrument, described brain electricity pre-process circuit comprises brain electricity prestage amplifying circuit, brain electrofiltration ripple amplifying circuit, 50Hz notch filter circuit; Brain electricity prestage amplifying circuit is selected the four high guaily unit device of low-power consumption, high input impedance, high cmrr; Brain electrofiltration ripple amplifying circuit adopts two amplifiers in the four high guaily unit device to be designed to respectively a voltage-controlled active high-pass filter of second order and the voltage-controlled active low-pass filter of second order, be combined into band filter, the input of band filter links to each other with the output of brain electricity prestage amplifying circuit; 50Hz notch filter circuit adopts double T rejector circuit.

Described brain electric network topography instrument, the computational methods of described mutual approximate entropy are: to calculate EEG signals alpha wave band x _Ft7(i) and x _Ft8(i), the mutual approximate entropy of 1≤i≤1024 is example, wherein x _Ft7(i) and x _Ft8(i) represent respectively the alpha signal at Ft7 and Ft8 place, its concrete calculation procedure is as follows:

(1) determine first two parameter m and r, wherein m is previously selected pattern dimension, and r is previously selected similar tolerance limit, and fix these two parameter m and r constant;

(2) establish x _Ft7(i) length is N, with filtered EEG signals x _Ft7(i) be divided in order N-m+1 m n dimensional vector n $X_{\mathrm{Ft}7}^1,X_{\mathrm{Ft}7}^2,...X_{\mathrm{Ft}7}^{N-m+1},$ Wherein:

$X_{\mathrm{Ft}7}^i=[x_{\mathrm{Ft}7}\left(i\right),x_{\mathrm{Ft}7}(i+1),...,x_{\mathrm{Ft}7}(i-m+1)],i=1~N-m+1$

Expression is since i m the x that point is continuous _Ft7(i) value, x _Ft8(i) also do similar processing;

(3) establish $X_{\mathrm{Ft}7}^i$ With $X_{\mathrm{Ft}8}^j$ For $X_{\mathrm{Ft}7}^1,X_{\mathrm{Ft}7}^2,...X_{\mathrm{Ft}7}^{N-m+1}$ With ${\mathrm{<figure-callout\; id="8"\; label="X\; Ft"\; filenames="HDA00002748543800091.png,HDA00002748543800101.png"\; state="\{\{state\}\}">X</figure-callout>}}_{\mathrm{Ft}}^8,{\mathrm{<figure-callout\; id="8"\; label="X\; Ft"\; filenames="HDA00002748543800091.png,HDA00002748543800101.png"\; state="\{\{state\}\}">X</figure-callout>}}_{\mathrm{Ft}}^8,...X_{\mathrm{Ft}8}^{N-m+1}$ In two m n dimensional vector ns arbitrarily, definition

With

Between distance

Be of difference maximum in both corresponding elements, that is:

$d[X_{\mathrm{Ft}7}^i,X_{\mathrm{Ft}8}^j]=\underset{1\&le;k\&le;m}{\max }|X_{\mathrm{Ft}7}^i\left(k\right)-X_{\mathrm{Ft}8}^j\left(k\right)|$

Each i value is calculated

With its complement vector

(the distance between i, j=1 ~ N-m+1);

(4) to each i, 1≤i≤N-m+1, Data-Statistics Less than the number of r and the ratio of this number and the total N-m+1 of distance, be denoted as

That is:

i＝1~N-m+1

(5) first will

Take the logarithm, ask again it to the meansigma methods of all i, be denoted as Φ ^m(r), that is:

${\mathrm{\&Phi;}}^m\left(r\right)=\frac{1}{N-m+1}\underset{i=1}{\overset{N-m+1}{\mathrm{\&Sigma;}}}\ln C_i^m\left(r\right);$

(6) again dimension is added 1, dimension becomes m+1, and repeating step (4.2)～(4.5) obtain

And Φ ^M+1(r);

(7) when sequence length is N, the estimated value of approximate entropy ApEn is mutually:

CrossApEn(m,r,N)＝Φ ^m(r)-Φ ^m+1(r)。

Described brain electric network topography instrument, described Threshold is: first threshold value is made as 1, and set up corresponding two values matrix and Laplacian Matrix δ according to threshold value, judge that whether the minimal eigenvalue of Laplacian Matrix δ is greater than 0, if greater than 0, threshold value is 1, and two values matrix is the brain network matrix; Otherwise threshold value is successively decreased gradually, until the minimal eigenvalue of Laplacian Matrix δ is greater than 0.According to above-mentioned definite threshold, definite method of two values matrix A is: when the analog value of mutual approximate entropy matrix more than or equal to threshold value, then the analog value of two values matrix A is 1; Otherwise be 0.

Brain electric network topography instrument according to claim 1 is characterized in that, described brain network parameter computational methods are: (1) calculates the node strength of each node.Computing formula is as follows:

$k_i=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{a}_{\mathrm{ij}}$

(2) calculate the cluster coefficients C of each node _iComputing formula is as follows:

$C_i=\frac{{2t}_i}{k_i(k_i-1)}$

Wherein

$t_i=\frac{1}{2}\underset{j,h\&Element;N}{\mathrm{\&Sigma;}}{a}_{\mathrm{ij}}{a}_{\mathrm{ih}}{a}_{\mathrm{jh}}$

K wherein _i, a _IjBe respectively the node degree element corresponding with two values matrix A; J, the value of h has been determined certain node in the network, and n is the node sum in institute's establishing network, and making it here is 32.

(3) Jie who calculates each node counts b _i, computing formula is as follows:

$b_i=\frac{1}{(n-1)(n-2)}\underset{h,j=1}{\overset{n}{\mathrm{\&Sigma;}}}\frac{{\mathrm{\&rho;}}_{\mathrm{hj}}\left(i\right)}{{\mathrm{\&rho;}}_{\mathrm{hj}}}$

ρ wherein _HjBe node h, the shortest path between the j, ρ _Hj(i) be the h through node i, the shortest path between the j.

With respect to prior art, the present invention has following advantage:

(1) utilize EEG signals to extract the brain network characterization, the brain wave acquisition circuit is simple, and is cheap, and collection is convenient and temporal resolution is better, and can adopt the software and hardware combining method to reject and disturb, and makes signal stabilization reliable;

(2) express in traditional landform plan pseudo-colours, on the basis of functional localization, describe also will provide relevant information between the electrode for encephalograms based on the topography of brain analysis of network;

(3) will comprise that WAVELET PACKET DECOMPOSITION and mutual nonlinear methods such as Analysis of Approximate Entropy and analysis of neural network combine, more adapt to EEG signals non-linear, non-stationary is essential, resulting result is more reliable;

Description of drawings

Fig. 1 is system construction drawing of the present invention;

Fig. 2 32 leads the electrode for encephalograms scattergram;

Fig. 3 is the elementary amplification circuit diagram of the used EEG signals of the present invention;

Fig. 4 is the used brain electrofiltration ripple amplification circuit diagram of the present invention;

Fig. 5 is the used 50Hz notch filter circuit diagram of the present invention;

The WAVELET PACKET DECOMPOSITION sketch map that Fig. 6 the present invention is used;

The EEG signals WAVELET PACKET DECOMPOSITION sketch map of a passage of Fig. 7;

The used threshold value of Fig. 8 the present invention is determined flow chart;

The node degree brain network topography figure of Fig. 9 Normal Subjects attitude;

Figure 10 normal person's attention state cluster coefficients brain network topography figure;

Figure 11 schizophrenic tranquillization attitude Jie counts brain network topography figure.

The specific embodiment

Below in conjunction with specific embodiment, the present invention is described in detail.

With reference to Fig. 1, checkout gear of the present invention is to comprise brain wave acquisition electrode (being attached on the scalp during use, so sometimes also claim scalp electrode), electroencephalogram conducting wire, brain electricity pre-process circuit, analog to digital conversion circuit and the signal processor that connects successively, wherein:

Signal processor, it is provided with brain electrical interference filtering module, brain electricity segmentation and the decomposing module that connects successively, mutual Analysis of Approximate Entropy module, analysis of threshold module, brain network parameter computing module and brain network contrast module, the concrete function of these modules is: brain electrical interference filtering module, and the interference of 50Hz power frequency, baseline drift and the myoelectricity that are used for the filtering EEG signals disturb; The brain wave extraction module be used for to extract EEG signals, carries out sliding window and processes as one section take per 1024; The Analysis of Approximate Entropy module is used for asking the mutual approximate entropy between each crosslinking electrode mutually; The analysis of threshold module is used to be set up brain network definite threshold and sets up corresponding two values matrix A; Brain network parameter computing module comprises node degree, cluster coefficients and Jie's number for setting up the brain network and calculate corresponding brain network parameter according to threshold value; Brain network contrast module is used for comparing with reference to person's brain network parameters of experimenter's brain network parameters and pre-stored, and the form of comparing result with brain electrical activity mapping shown.

Brain electricity pre-process circuit comprises brain electricity prestage amplifying circuit, brain electrofiltration ripple amplifying circuit, 50Hz notch filter circuit.As shown in Figure 3, brain electricity prestage amplifying circuit is selected low-power consumption, high input impedance, the four high guaily unit device of high cmrr, it is by the two-stage differential the electric circuit constitute, wherein first order difference channel is by the first operational amplifier U1 and the second operational amplifier U2 and resistance R 0, R1 and R2 form, and R1=R2, the forward end of the first operational amplifier U1 and the second operational amplifier U2 is drawn by brain electric conductance on line, link to each other according to the terminal electrode of international 10-20 standard assigned address with the experimenter, wherein the first operational amplifier U1 anode links to each other with the electrode of forehead position, the second operational amplifier U2 anode links to each other with the electrode of corresponding measured position in the 10-20 international standard, links to each other with the electrode of ear location with reference to ground wire 3; Second level difference channel is comprised of the 3rd operational amplifier U3 and peripheral resistance, and R3=R4, R5=R6, the output of the first operational amplifier U1 links to each other through the negative terminal of resistance R 3 with the 3rd operational amplifier U3, the output of the second operational amplifier U2 links to each other with the anode of the 3rd operational amplifier U3 through R4, and total amplification of this pre-amplification circuit is:

$A_1=(1+\frac{2R1}{R0})\frac{R5}{\mathrm{<figure-callout\; id="3"\; label="R"\; filenames="HDA00002748543800021.png"\; state="\{\{state\}\}">R</figure-callout>}3}=100$

The regulating resistance value is so that the front end circuit amplification is about 100 times.

As shown in Figure 4, this brain electrofiltration ripple amplifying circuit adopts two amplifiers of level Four amplifier device to be designed to respectively a voltage-controlled active high-pass filter of second order and the voltage-controlled active low-pass filter of second order, be combined into band filter, the input of band filter links to each other with the output of brain electricity prestage amplifying circuit, wherein four-operational amplifier U4 and C1, R7, C2 and R8 form high pass filter, the 5th operational amplifier U5 and R9, R11, C3 and C4 form low pass filter, and the 6th operational amplifier U6 and R10, R12 form secondary amplifier.

In order not lose the low-frequency component of EEG signals, the value of R7, R8, C1 and C2 in the Circuit tuning is so that the cut-off frequency of high pass filter is

$f_L=\frac{1}{2\mathrm{\&pi;}\sqrt{R7R8C1\mathrm{<figure-callout\; id="2"\; label="C"\; filenames="HDA00002748543800021.png,HDA00002748543800101.png"\; state="\{\{state\}\}">C</figure-callout>}2}}=0.1\mathrm{Hz}$

It should be noted that C1, C2 are positioned on the signalling channel, the noiseproof feature of itself is vital.Should select leaded multilayer ceramic capacitor or tantalum electric capacity, not select electrochemical capacitor.

Consider the frequency characteristic of EEG signals, the value of R9, R11, C3 and C4 in the Circuit tuning is so that the cut-off frequency of low pass filter is

$f_H=\frac{1}{2\mathrm{\&pi;}\sqrt{R9R11C3\mathrm{<figure-callout\; id="4"\; label="C"\; filenames="HDA00002748543800101.png,HDA00002748543800111.png"\; state="\{\{state\}\}">C</figure-callout>}4}}=100\mathrm{Hz}$

For the intermediate amplifier that is formed by the 6th operational amplifier U6, the resistance of regulating resistance R10 and R12, so that secondary amplification is about 400, the overall gain that preposition and secondary two-stage is amplified is 40000, so that the amplitude of EEG signals is fit to the requirement of follow-up analog to digital conversion circuit.Therefore, the amplification of this secondary amplification circuit is:

$A_2=1+\frac{R12}{\mathrm{<figure-callout\; id="10"\; label="R"\; filenames="HDA00002748543800101.png,HDA00002748543800111.png"\; state="\{\{state\}\}">R</figure-callout>}10}=400$

As shown in Figure 5,50Hz notch filter circuit adopts double T rejector circuit, is made of the 7th operational amplifier U7 and peripheral cell, and its output is as the input of analog to digital conversion circuit.Resistance capacitance value among the choose reasonable figure just can realize the notch filter of 50Hz.

The operation principle of above-mentioned testing circuit is: obtain simulating EEG signals from brain wave acquisition electrode and brain electricity pre-process circuit, after 12 analog digital conversion, obtain original brain electricity digital signal, and be transferred in the signal processor, functional module in the signal processor is processed brain electricity digital signal, extract operation such as filtering baseline drift, the interference of filtering myoelectricity, the interference of filtering 50Hz power frequency and brain wave, obtain EEG signals.Brain electric separation root module in the signal processor carries out selections artificially to the EEG signals that collects, and the EEG signals after selections are decomposed is in that Analysis of Approximate Entropy module and analysis of threshold module are carried out corresponding mutually approximate entropy and threshold calculations mutually, thereby set up corresponding brain network, and further calculate corresponding brain network parameter and comprise node degree, cluster coefficients and Jie's number, contrast module and compare with reference to module brain network parameter at the brain network, finally obtain brain network topography figure.

The concrete steps that the present invention tests brain electric network topography are as follows:

Step 1. installing electrodes.Such as Fig. 2 the brain wave acquisition electrode is placed on the brain scalp according to international 10-20 standard.Wherein with the ears electrode as ground electrode, the forehead electrode is as the reference electrode, the electrode of measuring position is positioned at the corresponding position of 10-20 standard.

It should be noted that during installing electrodes on the scalp that as far as possible electrode is placed the head corresponding site, require electrode and scalp impedance less than 5K Ω.In addition, should carefully clean scalp, usually need to adopt conductive paste.

Step 2. brain wave acquisition.Hardware circuit gets final product in the original EEG signals of preliminary filtering in the brain wave acquisition process baseline drift, myoelectricity disturb and the interference of 50Hz power frequency, after analog digital conversion, and eeg signal acquisition entering signal processor unit.

The electricity segmentation of step 3. brain and decomposition.Signal processor unit carries out segmentation to filtered EEG signals, carries out sliding window and processes as one section take per 1024, and adopt wavelet packet to carry out 9 layers of decomposition, can obtain the delta ripple of electroencephalogram, theta ripple, alpha ripple and beta ripple.Wherein the WAVELET PACKET DECOMPOSITION structural representation as shown in Figure 6, the signal after channel Wavelet bag decomposes is as shown in Figure 7.In Fig. 7, be followed successively by from top to bottom the delta ripple, the theta ripple, alpha ripple and beta ripple, wherein vertical coordinate represents amplitude, what abscissa represented to sample counts.

Step 4. is calculated the mutual approximate entropy of above-mentioned every band signal (delta ripple, theta ripple, alpha ripple and beta ripple) successively, and its result of calculation is as the input vector of brain network calculations.

To calculate EEG signals alpha wave band x _Ft7(i) and x _Ft8(i), the mutual approximate entropy of 1≤i≤1024 is example, wherein x _Ft7(i) and x _Ft8(i) represent respectively the alpha signal at Ft7 and Ft8 place, its concrete calculation procedure is as follows:

(4.1) determine first two parameter m and r, wherein m is previously selected pattern dimension, and r is previously selected similar tolerance limit, and fix these two parameter m and r constant;

(4.2) establish x _Ft7(i) length is N, with filtered EEG signals x _Ft7(i) be divided in order N-m+1 m n dimensional vector n $X_{\mathrm{Ft}7}^1,X_{\mathrm{Ft}7}^2,...X_{\mathrm{Ft}7}^{N-\mathrm{<figure-callout\; id="1"\; label="m\; +"\; filenames="HDA00002748543800021.png,HDA00002748543800051.png"\; state="\{\{state\}\}">m</figure-callout>}+1},$ Wherein:

$X_{\mathrm{Ft}7}^i=[x_{\mathrm{Ft}7}\left(i\right),x_{\mathrm{Ft}7}(i+1),...,x_{\mathrm{Ft}7}(i-m+1)],i=1~N-\mathrm{<figure-callout\; id="1"\; label="m\; +"\; filenames="HDA00002748543800021.png,HDA00002748543800051.png"\; state="\{\{state\}\}">m</figure-callout>}+1$

Expression is since i m the x that point is continuous _Ft7(i) value, x _Ft8(i) also do similar processing.

(4.3) establish With

For $X_{\mathrm{Ft}7}^1,X_{\mathrm{Ft}7}^2,...X_{\mathrm{Ft}7}^{N-\mathrm{<figure-callout\; id="1"\; label="m\; +"\; filenames="HDA00002748543800021.png,HDA00002748543800051.png"\; state="\{\{state\}\}">m</figure-callout>}+1}$ With ${\mathrm{<figure-callout\; id="8"\; label="X\; Ft"\; filenames="HDA00002748543800091.png,HDA00002748543800101.png"\; state="\{\{state\}\}">X</figure-callout>}}_{\mathrm{Ft}}^8,{\mathrm{<figure-callout\; id="8"\; label="X\; Ft"\; filenames="HDA00002748543800091.png,HDA00002748543800101.png"\; state="\{\{state\}\}">X</figure-callout>}}_{\mathrm{Ft}}^8,...X_{\mathrm{Ft}8}^{N-\mathrm{<figure-callout\; id="1"\; label="m\; +"\; filenames="HDA00002748543800021.png,HDA00002748543800051.png"\; state="\{\{state\}\}">m</figure-callout>}+1}$ In two m n dimensional vector ns arbitrarily, definition With Between distance

Be of difference maximum in both corresponding elements, that is:

$d[X_{\mathrm{Ft}7}^i,X_{\mathrm{Ft}8}^j]=\underset{1\&le;k\&le;m}{\mathrm{<figure-callout\; id="1"\; label="max"\; filenames="HDA00002748543800021.png,HDA00002748543800051.png"\; state="\{\{state\}\}">max</figure-callout>}}|X_{\mathrm{Ft}7}^i\left(k\right)-X_{\mathrm{Ft}8}^j\left(k\right)|$

Each i value is calculated

With its complement vector

(the distance between i, j=1 ~ N-m+1);

(4.4) to each i, 1≤i≤N-m+1, Data-Statistics Less than the number of r and the ratio of this number and the total N-m+1 of distance, be denoted as

That is:

i＝1~N-m+1

(4.5) first will

Take the logarithm, ask again it to the meansigma methods of all i, be denoted as Φ ^m(r), that is:

${\mathrm{\&Phi;}}^m\left(r\right)=\frac{1}{N-\mathrm{<figure-callout\; id="1"\; label="m\; +"\; filenames="HDA00002748543800021.png,HDA00002748543800051.png"\; state="\{\{state\}\}">m</figure-callout>}+1}\underset{i=1}{\overset{N-\mathrm{<figure-callout\; id="1"\; label="m\; +"\; filenames="HDA00002748543800021.png,HDA00002748543800051.png"\; state="\{\{state\}\}">m</figure-callout>}+1}{\mathrm{\&Sigma;}}}\ln C_i^m\left(r\right);$

(4.6) again dimension is added 1, dimension becomes m+1, and repeating step (4.2)～(4.5) obtain

And Φ ^M+1(r);

(4.7) when sequence length is N, the estimated value of approximate entropy ApEn is mutually:

CrossApEn(m,r,N)＝Φ ^m(r)-Φ ^m+1(r)

Grope to draw according to practical experience, work as m=2, r=0.1 ~ 0.25COV (x _Ft7, x _Ft8) time, approximate entropy CrossApEn has comparatively reasonably statistical property, the COV (x here mutually _Ft7, x _Ft8) be initial data x _Ft7(i) and x _Ft7(i), the covariance of i=1 ~ N.Thus, get m=2 in the calculating of native system approximate entropy, r=0.18COV (x _Ft7, x _Ft8).

Step 5. is determined the threshold value of this brain network, and sets up corresponding brain network connection matrix A.

The flow chart of Threshold of the present invention as shown in Figure 8, employing be the method that guarantees the largest connected property of network, and finally set up two values matrix and be brain network connection matrix A.Its basic thought is, first threshold value is made as 1, and sets up corresponding two values matrix and Laplacian Matrix δ according to threshold value, whether judges the minimal eigenvalue of Laplacian Matrix δ greater than 0, if greater than 0, threshold value is 1, and two values matrix is brain network connection matrix; Otherwise threshold value is successively decreased gradually, until the minimal eigenvalue of Laplacian Matrix δ is greater than 0.

T among Fig. 8 (w) _nBe threshold value, the establishment method of two values matrix A is

$a_{\mathrm{ij}}=\left\{\begin{array}{cc}1& {\mathrm{CA}}_{\mathrm{ij}}\&GreaterEqual;{T\left(w\right)}_{\mathrm{<figure-callout\; id="0"\; label="n"\; filenames="HDA00002748543800041.png,HDA00002748543800061.png"\; state="\{\{state\}\}">n</figure-callout>}}\\ 0& {\mathrm{CA}}_{\mathrm{ij}}<{T\left(w\right)}_n\end{array}\right.$

Wherein the CA matrix is the mutual approximate entropy matrix of gained in the step 4.Definite method of Laplacian Matrix δ is:

${\mathrm{\&delta;}}_{\mathrm{ij}}=\left\{\begin{array}{cc}1& i=j\\ 0& i\&NotEqual;j\end{array}\right.$

Step 6. is calculated corresponding brain network parameter according to the resulting brain network connection of step 5 matrix A.

(6.1) calculate the node strength of each node.Computing formula is as follows:

$k_i=\underset{j}{\mathrm{\&Sigma;}}{a}_{\mathrm{ij}}$

(6.2) calculate the cluster coefficients C of each node _iComputing formula is as follows:

$C_i=\frac{{2t}_i}{k_i(k_i-1)}$

Wherein

$t_i=\frac{1}{2}\underset{j,h=1}{\overset{n}{\mathrm{\&Sigma;}}}{a}_{\mathrm{ij}}{a}_{\mathrm{ih}}{a}_{\mathrm{jh}}$

K wherein _i, a _IjBe respectively the node degree element corresponding with two values matrix A.

(6.3) Jie who calculates each node counts b _iComputing formula is as follows:

$b_i=\frac{1}{(n-1)(n-2)}\underset{h,j\&Element;N}{\mathrm{\&Sigma;}}\frac{{\mathrm{\&rho;}}_{\mathrm{hj}}\left(i\right)}{{\mathrm{\&rho;}}_{\mathrm{hj}}}$

ρ wherein _HjBe node h, the shortest path between the j, ρ _Hj(i) be node h, the shortest path between the j is through node i.

Step 7. makes up standard brain network parameter storehouse.Take the brain wave acquisition data of 14 Normal Subjects attitudes as primary signal, carry out the brain analysis of network of above-mentioned each step, calculate above-mentioned three brain network parameters, average and be stored as standard brain network parameter storehouse.

Step 8. will calculate the brain network parameter of gained and pre-stored with reference to person's brain network parameters relatively, with carrying out interpolation calculation apart from the inverse ratio interpolation method in the two-dimensional interpolation algorithm, specific algorithm is as follows:

$v_e=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_j{v}_j$

V wherein _j(j=1 ..., n) be point (x _j, y _j) variate-value located, w _jIt is its corresponding weight coefficient.

Weight coefficient w _jGenerally provided by following formula:

$w_j=\frac{f\left(d_{\mathrm{ej}}\right)}{\underset{j-1}{\overset{n}{\mathrm{\&Sigma;}}}f\left(d_{\mathrm{ej}}\right)}$

Wherein n is known counting, f (d _Ej) represent for interpolation point (x _e, y _e) and known point (x _j, y _j) between apart from d _EjWeight coefficient, n=32 here.

$f\left(d_{\mathrm{ej}}\right)=\frac{1}{d_{\mathrm{ej}}^b}$

The general value of b is 1 or 2, respective distances interpolation reciprocal and inverse distance square interpolation, here b=1.

Step 9. shows comparing result with pseudo-colours Two dimensional Distribution image.Wherein larger to represent the tester larger than Normal Subjects attitude relevant parameter for number, and it is less than Normal Subjects attitude relevant parameter that negative value represents the tester, and it is substantially the same with Normal Subjects attitude relevant parameter that null value represents the tester.

Node degree brain network topography figure sketch map, normal person that Fig. 9, Figure 10, Figure 11 are respectively the Normal Subjects attitude notice that attitude cluster coefficients brain network topography figure sketch map, schizophrenic's tranquillization attitude Jie count brain network topography figure sketch map, and wherein A, B, C, D represent respectively delta, theta, alpha, beta ripple.

The present invention can be used for immediately detecting easily people's brain function state, realizes the accurate judgement to the brain function state.

Should be understood that, for those of ordinary skills, can be improved according to the above description or conversion, and all these improvement and conversion all should belong to the protection domain of claims of the present invention.

Claims (5)

1. topography drawing apparatus based on the brain Electrical network analysis, it is characterized in that, comprise the brain wave acquisition electrode, electroencephalogram conducting wire, brain electricity pre-process circuit, analog to digital conversion circuit and the signal processor that connect successively, described signal processor is provided with brain electrical interference filtering module, brain electricity segmentation and the decomposing module that connects successively, mutual Analysis of Approximate Entropy module, analysis of threshold module, brain network parameter computing module and brain network contrast module; The interference of 50Hz power frequency, baseline drift and myoelectricity that brain electrical interference filtering module is used for the filtering EEG signals disturb; The brain wave extraction module be used for to extract EEG signals, carries out sliding window and processes as one section take per 1024, to calculate relevant parameter in the EEG signals; The Analysis of Approximate Entropy module is used for asking the mutual approximate entropy between each crosslinking electrode mutually; The analysis of threshold module is used to be set up brain network definite threshold and sets up corresponding two values matrix A; Brain network parameter computing module is used for setting up the brain network and calculating corresponding brain network parameter according to threshold value; Brain network contrast module is used for the comparing with reference to person's brain network parameters of experimenter's brain network parameters and pre-stored, and with the form demonstration of comparing result with brain electrical activity mapping.

2. brain electric network topography instrument according to claim 1 is characterized in that, described brain electricity pre-process circuit comprises brain electricity prestage amplifying circuit, brain electrofiltration ripple amplifying circuit, 50Hz notch filter circuit; Brain electricity prestage amplifying circuit is selected the four high guaily unit device of low-power consumption, high input impedance, high cmrr; Brain electrofiltration ripple amplifying circuit adopts two amplifiers in the four high guaily unit device to be designed to respectively a voltage-controlled active high-pass filter of second order and the voltage-controlled active low-pass filter of second order, be combined into band filter, the input of band filter links to each other with the output of brain electricity prestage amplifying circuit; 50Hz notch filter circuit adopts double T rejector circuit.

3. brain electric network topography instrument according to claim 1 is characterized in that the computational methods of described mutual approximate entropy are: to calculate EEG signals alpha wave band x _Ft7(i) and x _Ft8(i), the mutual approximate entropy of 1≤i≤1024 is example, wherein x _Ft7(i) and x _Ft8(i) represent respectively the alpha signal at Ft7 and Ft8 place, its concrete calculation procedure is as follows:

(1) determine first two parameter m and r, wherein m is previously selected pattern dimension, and r is previously selected similar tolerance limit, and fix these two parameter m and r constant;

(2) establish x _Ft7(i) length is N, with filtered EEG signals x _Ft7(i) be divided in order N-m+1 m n dimensional vector n $X_{\mathrm{Ft}7}^1,X_{\mathrm{Ft}7}^2,...X_{\mathrm{Ft}7}^{N-m+1},$ Wherein:

$X_{\mathrm{Ft}7}^i=[x_{\mathrm{Ft}7}\left(i\right),x_{\mathrm{Ft}7}(i+1),...,x_{\mathrm{Ft}7}(i-m+1)],i=1~N-m+1$

Expression is since i m the x that point is continuous _Ft7(i) value, x _Ft8(i) also do similar processing;

(3) establish

With

For $X_{\mathrm{Ft}7}^1,X_{\mathrm{Ft}7}^2,...X_{\mathrm{Ft}7}^{N-m+1}$ With $X_{\mathrm{Ft}8}^1,X_{\mathrm{Ft}8}^2,...X_{\mathrm{Ft}8}^{N-m+1}$ In two m n dimensional vector ns arbitrarily, definition

With

Between distance

Be of difference maximum in both corresponding elements, that is:

$d[X_{\mathrm{Ft}7}^i,X_{\mathrm{Ft}8}^j]=\underset{1\&le;k\&le;m}{\max }|X_{\mathrm{Ft}7}^i\left(k\right)-X_{\mathrm{Ft}8}^j\left(k\right)|$

Each i value is calculated

With its complement vector

(the distance between i, j=1 ~ N-m+1);

(4) to each i, 1≤i≤N-m+1, Data-Statistics

Less than the number of r and the ratio of this number and the total N-m+1 of distance, be denoted as

That is:

i＝1~N-m+1

(5) first will

Take the logarithm, ask again it to the meansigma methods of all i, be denoted as Φ ^m(r), that is:

${\mathrm{\&Phi;}}^m\left(r\right)=\frac{1}{N-m+1}\underset{i=1}{\overset{N-m+1}{\mathrm{\&Sigma;}}}\ln C_i^m\left(r\right);$

(6) again dimension is added 1, dimension becomes m+1, and repeating step (2)～(5) obtain

And Φ ^M+1(r);

(7) when sequence length is N, the estimated value of approximate entropy ApEn is mutually:

CrossApEn(m,r,N)＝Φ ^m(r)-Φ ^m+1(r)。

4. brain electric network topography instrument according to claim 1, it is characterized in that, definite method of described threshold value and two values matrix A is: first threshold value is made as 1, and set up corresponding two values matrix A and Laplacian Matrix δ according to threshold value, judge that whether the minimal eigenvalue of Laplacian Matrix δ is greater than 0, if greater than 0, threshold value is 1, and two values matrix is the brain network matrix; Otherwise threshold value is successively decreased gradually by 0.01, until the minimal eigenvalue of Laplacian Matrix δ is greater than 0, the threshold value of determining this moment is the threshold value that guarantees the largest connected property of network;

According to above-mentioned definite threshold, definite method of two values matrix A is: when the analog value of mutual approximate entropy matrix more than or equal to threshold value, then the analog value of two values matrix A is 1; Otherwise be 0.

5. brain electric network topography instrument according to claim 1 is characterized in that, described brain network parameter computational methods are:

(1) calculates the node strength k of each node _iComputing formula is as follows:

$k_i=\underset{j}{\mathrm{\&Sigma;}}{a}_{\mathrm{ij}}$

A wherein _IjBe element corresponding to two values matrix A;

(2) calculate the cluster coefficients C of each node _iComputing formula is as follows:

$C_i=\frac{{2t}_i}{k_i(k_i-1)}$

$t_i=\frac{1}{2}\underset{j,h=1}{\overset{n}{\mathrm{\&Sigma;}}}{a}_{\mathrm{ij}}{a}_{\mathrm{ih}}{a}_{\mathrm{jh}}$

Wherein: j, the value of h has been determined certain node in the network, and n is the node sum in institute's establishing network, and making it here is 32;

(3) Jie who calculates each node counts b _i, computing formula is as follows:

$b_i=\frac{1}{(n-1)(n-2)}\underset{h,j=1}{\overset{n}{\mathrm{\&Sigma;}}}\frac{{\mathrm{\&rho;}}_{\mathrm{hj}}\left(i\right)}{{\mathrm{\&rho;}}_{\mathrm{hj}}}$

ρ wherein _HjBe node h, the shortest path between the j, ρ _Hj(i) be the h through node i, the shortest path between the j.

Images