CN104083163A - Method for obtaining nonlinearity parameter electroencephalogram mapping - Google Patents

Abstract

The invention provides a method for obtaining nonlinearity parameter electroencephalogram mapping. The method comprises the following steps that (1) multi-lead electroencephalograms synchronously collected on the scalp are coded through dynamic nonlinearity coding rules to form a multi-dimensional code sequence; (2) the multi-dimensional code sequence obtained in the step (1) is calculated to obtain multi-dimensional information increment entropy; (3) the multi-dimensional information increment entropy obtained in the step (2) is mapped to a simulated head top view according to electrode positions collected through the electroencephalograms, two-dimensional interpolation is conduced, and finally the nonlinearity parameter electroencephalogram mapping is drawn. The nonlinearity parameter electroencephalogram mapping drawn through the method can well reflect different activity states of the brain in a visualized mode, the preprocessing such as artificial artifact removing does not need to be conducted on the collected electroencephalograms, and the effect is quite good based on practical data testing.

Description

A kind of method of nonlinear parameter brain electrical activity mapping

Technical field

The present invention relates to brain-electrical signal processing method and brain electrical activity mapping technology.

Background technology

Electroencephalogram (Electroencephalogram, EEG) is the important means that we understand cerebral activity and state, and brain electrical activity mapping (EEG mapping) is widely used a kind of electroencephalogram diagnostic techniques clinically at present.First brain electrical activity mapping technology extracts characteristic parameter from electroencephalogram, the putting position of electrode during then in conjunction with collection EEG, characteristic parameter is plotted on two dimensional surface as the scalar function by determining positions, thereby the time domain waveform of EEG is become and can be located and quantitative topography simultaneously.

At present, brain electrical activity mapping is drawn and is mainly depended on frequency domain parameter,, difference according to frequency content is divided into brain electricity the different rhythm and pace of moving things such as δ (0.5Hz～4Hz), θ (4Hz～8Hz), α (8Hz～13Hz), β (13Hz～30Hz), and the absolute energy of the corresponding rhythm and pace of moving things or relative energy form the characteristic parameter in topographic map drawing.

There is following limitation in current brain electrical activity mapping technology:

First, frequency-domain analysis is a kind of linear analysis method in essence, and analysis result does not possess robustness for the data of non-stationary.Yet EEG signals is rich in non-linear component, and usually right and wrong stably, therefore, frequency domain parameter is limited for the capturing ability of brain electrical feature.

In addition, EEG signals signal to noise ratio is low, usually contains multiple artefact, as existing diagnostic analysis (comprising frequency-domain analysis) is last, needs experienced doctor to carry out artificial removal's artefact, greatly reduces convenience and the automaticity of brain electrodiagnostic technique.

For above-mentioned limitation, if set up the topography method of reflection non-linear eeg characteristic, resist well intrinsic non-stationary and various interference, the artefact of brain electricity, will be clinically for brain electrodiagnosis provides diagnosis methods more accurately and fast.

Summary of the invention

The object of the present invention is to provide a kind of nonlinear parameter brain electrical activity mapping method, the nonlinear characteristic based on EEG signals effectively reflects cerebral activity state, and can effectively resist the non-stationary property of EEG signals, does not need the pretreatment such as artificial removal's artefact simultaneously.

The object of the present invention is achieved like this:

The EEG signal of leading based on gathering more, through kinematic nonlinearity coding, form multidimensional coding sequence, extract the nonlinear characteristic parameters----information increment entropy of coded sequence, during then in conjunction with collection EEG, electrode is at the putting position of scalp, using characteristic parameter as the scalar function by determining positions, after two-dimensional interpolation, be plotted on the head top view of simulation, realize nonlinear parameter brain electrical activity mapping.Core of the present invention comprises kinematic nonlinearity coding and nonlinear characteristic parameters extraction two large divisions.

Further, the kinematic nonlinearity coding described in the present invention, comprises the following steps:

The multi-lead EEG signal { x collecting for scalp noinvasive _i: 1≤i≤N}, investigates its probability distribution, and calculates first-order difference sequence { Δ x _i: 1≤i≤N-1}, forms new binary code sequence { s according to the encoding symbols of its probability distribution and first-order difference sequence _i: 1≤i≤N-1}.If binary coding word length is m, coding rule is described below:

First to { x _i: after 1≤i≤N} sequence, carry out 2 ^m-1decile, finds 2 ^m-1-1 is waited quantile, is designated as

Then according to difference sequence to { s _ilast position revise,

Thereby obtain non-uniform encoding sequence { s _i.

This coding rule be take probability encoding as basis, broken through the linearity limitation of amplitude coding, can also greatly weaken the disturbance that larger mutation disturbance causes on amplitude territory originally, simultaneously, by introducing difference sequence symbol, the radio-frequency component of primary signal has also been introduced to coding rule, thereby effectively resisted the impact of non-stationary trend term.

Further, the multidimensional information increment entropy described in the present invention calculates, and comprises the following steps:

By binary code sequence { s _ibe converted to decimal sequence { y _i;

To sequence { y _ido respectively 2 dimensions, 3 dimension delay embeddings, obtain vector sequence { B ⁽²⁾(i) }, { B ⁽³⁾(i) };

Calculate respectively under 2 dimension delay embeddings and 3 dimension delay embeddings the probability C that two vectors are identical ⁽²⁾, C ⁽³⁾;

Sequence of calculation information increment entropy

This information increment entropy has reflected under coding rule, a step predictability or the information increment of coded sequence.

Further, the brain electrical activity mapping described in the present invention is drawn, and comprises the following steps:

According to the electrode putting position of brain wave acquisition, by the aforementioned multidimensional information increment entropy obtaining, the two-dimensional function that is configured to be determined by electrode position;

Take electrode putting position as node, the head top view of simulation is carried out to the interpolation of two-dimensional function;

Finally, functional value after interpolation is mapped to color gamut or gray scale territory, draws out the non-linear brain electrical activity mapping of head top view.

In sum, the beneficial effect of the inventive method, it is core that Nonlinear Dynamic coding calculates with information increment entropy, the intrinsic propesties that has realized EEG signal understands, then in conjunction with topography rendering technique, finally draw out the visual non-linear brain electrical activity mapping of reflection cerebration state.

Coding rule in the present invention and calculation of characteristic parameters have all broken through linear domain, simultaneously, can also resist by means of coding the impact of non-stationary trend term and non-stationary mutation disturbance artefact, therefore, this method does not need acquired original data to be carried out to the pretreatment such as filtering or artificial artefact removal, has therefore greatly increased the practicality of this method and the probability of automatization.In addition, because computational process all in this method all only relates to elementary mathematics computing, therefore easily realize.

Accompanying drawing explanation

Fig. 1 is the theory diagram of the non-linear brain electrical activity mapping method of the present invention.

Fig. 2 is the nonlinear parameter brain electrical activity mapping of test data one (Healthy People is done the left hand motion imagination (a) and contrast state (b)).

Fig. 3 is the nonlinear parameter brain electrical activity mapping of test data two (Healthy People is done the leg exercise imagination (a) and contrast state (b)).

Fig. 4 is the nonlinear parameter brain electrical activity mapping of test data three (Healthy People is done the left hand motion imagination (a) and the right hand motion imagination (b)).

The specific embodiment

In order more to understand technology contents of the present invention, especially exemplified by specific embodiment and coordinate appended illustrate as follows.

Fig. 1 is the theory diagram of a kind of non-linear brain electrical activity mapping method of the present invention.

A non-linear brain electrical activity mapping method, step comprises:

1), for the EEG signal of leading collecting, through kinematic nonlinearity coding, form multidimensional coding sequence more;

2) the multidimensional coding sequence obtaining in step 1, extracts nonlinear characteristic parameters----information increment entropy;

3) nonlinear characteristic parameters obtaining in step 2, as the scalar function by determining positions, and take electrode position as node carries out two-dimensional interpolation, is finally plotted on the head top view two dimensional surface of simulation, realizes nonlinear parameter brain electrical activity mapping.Every electrode all has data to provide coded sequence.

Described step 1) in, kinematic nonlinearity coding, 3 the word length binary codings of take are example, specifically comprise:

1.1) to the original series { x that respectively leads _iby ascending order (or descending), arrange and obtain sequence { u _i: 1≤i≤N}, obtains sequence { u successively _iin 3 quartering point values (i.e. 25%, 50%, 75% quantile), be designated as successively from small to large t ₁, t ₂, t ₃; Each is led and independently encodes respectively and calculate.For the symbol of simplifying below represents, the mark subscript of leading not, so, while narrating here with " respectively leading ".

1.2) according to formula (1) to the original series that respectively leads before N-1 data carry out the binary coding of 3 word lengths,

$s_i=\left\{\begin{array}{c}000:x_i\&le;t_1\\ 010:t_1<x_i\&le;t_2\\ 100:t_2<x_i\&le;t_3\\ 110:t_3<x_i\end{array}\right.,(1\&le;i\&le;N-1)---\left(1\right)$

1.3) increment sequence { the Δ x that reentries _i: 1≤i≤N-1}, wherein

Δx _i＝x _i+1-x _i (2)

Symbol correction step 1.2 with increment sequence) lowest order of coding in,

$s_i=\left\{\begin{array}{cc}**0& (\mathrm{\&Delta;}{x}_i\&le;0)\\ **1& (\mathrm{\&Delta;}{x}_i>0)\end{array}\right.,(1\&le;i\&le;N-1)---\left(3\right)$

Thereby form 3 binary code sequence { s _i: 1≤i≤N-1}.

Described step 2), in, nonlinear parameter---the extraction of-information increment entropy specifically comprises:

2.1) by binary code sequence { s _i: 1≤i≤N} is converted into decimal coded sequence { y _j: 1≤j≤N-1}

2.2) to sequence { y _jdo respectively 2 dimensions, 3 dimension delay embeddings, obtain vector sequence:

B ⁽²⁾(i)＝(y _i,y _i+1)，1≤i≤N-2 (4)

B ⁽³⁾(i)＝(y _i,y _i+1,y _i+2)，1≤i≤N-3 (5)

2.3) definition be respectively and B ⁽²⁾(i), B ⁽³⁾(i) identical vector number, 2 dimensions, the vector of 3 dimensions under delay embeddings between two identical probability be respectively:

$C^{\left(2\right)}=\frac{1}{N-2}\underset{i=1}{\overset{N-2}{\mathrm{\&Sigma;}}}\frac{n_i^{\left(2\right)}}{N-2}---\left(6\right)$

$C^{\left(3\right)}=\frac{1}{N-3}\underset{i=1}{\overset{N-3}{\mathrm{\&Sigma;}}}\frac{n_i^{\left(3\right)}}{N-3}---\left(7\right)$

2.4) information increment entropy is

$S=-\ln \frac{C^{\left(3\right)}}{C^{\left(2\right)}}---\left(6\right)$

Described step 3) in, the information increment entropy result that each calculating in step 2 led, as the interpolation knot of the two-dimensional function relevant with electrode position, carries out two-dimensional interpolation, then be mapped to color gamut, finally draw out nonlinear parameter brain electrical activity mapping (employing prior art).

With this method, at the EEG to actual acquisition, apply below, the invention will be further described by reference to the accompanying drawings.

With reference to figure 2, be the EEG that a Healthy People is collected, utilize our method, the non-linear brain electrical activity mapping of drafting.Wherein subgraph (a) is done the state of the left hand motion imagination for it, and subgraph (b) is the contrast state of the imagination of not taking exercises.As seen from the figure, during the motion imagination, when the information increment entropy of both sides raises, there is reduction in the information increment entropy of middle part, top and occipitalia.

With reference to figure 3, be the EEG that another Healthy People is collected, utilize our method, the non-linear brain electrical activity mapping of drafting.Wherein subgraph (a) is done the state of the leg exercise imagination for it, and subgraph (b) is the contrast state of the imagination of not taking exercises.As seen from the figure, when imagination motion, the information increment entropy of occipitalia is significantly lower than the information increment entropy of contrast state occipitalia.

With reference to figure 4, be the EEG that another Healthy People is collected, utilize our method, the non-linear brain electrical activity mapping of drafting.Wherein subgraph (a) is done the state of the left hand motion imagination for it, and subgraph (b) is for doing the state of the right hand motion imagination.As seen from the figure, during the different hands movement imagination, between top and occipitalia, the magnitude relationship of information increment entropy in the left and right sides presents to be put upside down.

Above-mentioned diagram presentation of results, a kind of nonlinear parameter brain electrical activity mapping that the present invention proposes, can reflect the different active states of brain effectively.

In the present invention, core is dynamic non-uniform encoding and nonlinear parameter---the calculating of-information increment entropy.EEG signals essence is nonlinear, has considerable non-stationaryly, has numerous artefacts and interference simultaneously.Therefore,, by the reflection cerebral activity of brain electricity and state, need to adopt the effectively non-stationary nonlinear method of antagonism.In the present invention, kinematic nonlinearity coding has been broken through the linear restriction of traditional amplitude coding, adopts probability encoding can also realize variable-resolution, fact proved it is a kind of effective non-uniform encoding.In the present invention, also include the instant dynamic change of EEG in coding rule, thereby retained the radio-frequency component of initial data.Inventor thinks, the kinematic nonlinearity coding in the present invention can solve non-stationary and artefact, the interference problem existing in current electroencephalogramsignal signal analyzing well, realizes effective reservation of brain dynamical system intrinsic propesties.In the present invention, nonlinear parameter----information increment entropy, under aforementioned coding prerequisite, has reflected information increment when system is done one-step prediction, is portraying of nonlinear characteristic, thereby has realized effective statement of nonlinear dynamic characteristic.

Calculating because this method relates to, does not all need stationarity supposed premise, does not need to carry out the pretreatment such as artificial artefact removal yet, meanwhile, because computational process all in this method all only relates to elementary mathematics computing, easily realize, therefore, can easily realize and be beneficial to the automatic processing of data.

Although the present invention discloses as above with preferred embodiment, so it is not in order to limit the present invention.Persond having ordinary knowledge in the technical field of the present invention, without departing from the spirit and scope of the present invention, when being used for a variety of modifications and variations.Therefore, protection scope of the present invention is when being as the criterion depending on claims person of defining.

Claims (5)

1. a method that obtains nonlinear parameter brain electrical activity mapping, is characterized in that, comprises following three steps:

1) for the multi-lead electroencephalogram arriving in scalp synchronous acquisition, by kinematic nonlinearity coding rule, encode, form multidimensional coding sequence;

2) to step 1) in the multidimensional coding sequence that obtains, calculate multidimensional information increment entropy;

3) to step 2) in the multidimensional information increment entropy that obtains, according to the electrode position of brain wave acquisition, be mapped to the head top view of simulation, and carry out two-dimensional interpolation, finally draw out nonlinear parameter brain electrical activity mapping.

2. nonlinear parameter brain electrical activity mapping method according to claim 1, is characterized in that described step 1) in, kinematic nonlinearity coding, comprising:

Original series { x for the electroencephalogram that respectively leads _i: 1≤i≤N}, investigates its probability distribution, and calculates first-order difference sequence { Δ x _i: 1≤i≤N-1}, forms new coded sequence { s according to the encoding symbols of its probability distribution and first-order difference sequence _i: 1≤i≤N-1}; If binary coding word length is m, coding rule is described below:

First original series { the x to the electroencephalogram that respectively leads _i: after 1≤i≤N} sequence, carry out 2 ^m-1decile, finds 2 ^m-1-1 is waited quantile, is designated as new coded sequence { s _i}

Then according to difference sequence to { s _ilast position revise after,

Thereby obtain final non-uniform encoding sequence { s _i.

3. nonlinear parameter brain electrical activity mapping method according to claim 1, is characterized in that described step 2) in, calculate multidimensional information increment entropy, comprise the following steps:

3.1) by binary system non-uniform encoding sequence { s _iconvert decimal sequence { y to _i;

3.2) to sequence { y _ido respectively 2 dimensions, 3 dimension delay embeddings, obtain vector sequence { B ⁽²⁾(i) }, { B ⁽³⁾(i) };

3.3) calculate respectively under 2 dimension delay embeddings and 3 dimension delay embeddings, vector is identical probability C between two ⁽²⁾, C ⁽³⁾;

3.4) computing information increment entropy

4. nonlinear parameter brain electrical activity mapping method according to claim 1, is characterized in that, this information increment entropy has reflected under coding rule, a step predictability or the information increment of coded sequence.

5. nonlinear parameter brain electrical activity mapping method according to claim 1, is characterized in that, according to the electrode putting position of brain wave acquisition, by the aforementioned multidimensional information increment entropy obtaining, the two-dimensional function that is configured to be determined by electrode position;

Take electrode putting position as node, the head top view of simulation is carried out to the interpolation of two-dimensional function;

Finally, the functional value after interpolation is mapped to color gamut or gray scale territory, draws out the non-linear brain electrical activity mapping of head top view.