CN102068246A - Method for enhancing toughness of weak evoked potential (EP) signal - Google Patents

Abstract

The invention provides a method for enhancing toughness of a weak evoked potential (EP) signal. The method comprises the following steps: taking a noisy EP signal as an input signal of a nonlinear dynamical system, wherein a noise signal in the noisy EP signal is described by means of alpha-stable distribution; and adjusting parameters of the nonlinear dynamical system so that the input signal and the nonlinear dynamical system work cooperatively to generate stochastic resonance (SR) effect, and outputting an output signal with the highest signal-to-noise ratio. In the method for enhancing the toughness of the weak EP signal, the noise signal is described by means of alpha-stable distribution, and the signal-to-noise ratio of the EP signal is improved by virtue of the SR effect, thus improving the reliability and accuracy for diagnosis and detection of pathological change of a clinical nervous system injury.

Description

Faint toughness Enhancement Method of bringing out electric potential signal

Technical field

The present invention relates to the medical signals process field, relate in particular to a kind of faint toughness Enhancement Method of bringing out electric potential signal.

Background technology

Bring out current potential (evoked potentials, EP) be meant and give adequate stimulus a certain specific part of nervous system (from the sensor to the cerebral cortex), detect at central nervous system's (comprising peripheral nervous system) corresponding site with stimulation lock is arranged the time relation potential change.

Bring out electric potential signal invariably accompany spontaneous brain electricity signal EEG and other noise signals, promptly bringing out electric potential signal is the signal that band is made an uproar.In the prior art, adopt Gauss distribution to describe EEG signals EEG and other noise signals, yet in operating room or have under the improper situation such as hostility, these noise signals have tangible non-Gaussian pulse.

Generally, it is a lot of a little less than than noise signal to bring out electric potential signal, bring out electric potential signal for use and need improve the signal to noise ratio of bringing out electric potential signal, in the prior art, the method that improves signal to noise ratio has a lot, for example, cumulative mean, wavelet transformation and independent component analysis etc., these methods think that all noise signal is harmful to bringing out electric potential signal, noise signal are suppressed or eliminate, with the raising signal to noise ratio.

Summary of the invention

The object of the present invention is to provide a kind of faint toughness Enhancement Method of bringing out electric potential signal, adopt stable distribution of α to describe noise signal, adopt accidental resonance (stochastic resonance, SR) effect improves the signal to noise ratio of bringing out electric potential signal, has improved the reliability and the accuracy of clinical nervous system injury pathological changes diagnosis and detection.

To achieve the above object, the invention provides a kind of faint toughness Enhancement Method of bringing out electric potential signal, may further comprise the steps: use the band that adopts the stable distribution of α to describe noise signal to make an uproar and bring out the input signal of electric potential signal as nonlinear dynamic system; Adjust the parameter of nonlinear dynamic system, make input signal and nonlinear dynamic system synergism produce accidental resonance effect, the output signal that output signal-to-noise ratio is the highest.

Above-mentioned faint toughness Enhancement Method of bringing out electric potential signal, wherein, described band is made an uproar and is brought out electric potential signal and add that by the purified electric potential signal that brings out the stable noise signal of describing that distributes of α forms.

Above-mentioned faint toughness Enhancement Method of bringing out electric potential signal, wherein, described nonlinear dynamic system adopts following stochastic differential equation to describe:

$\frac{\mathrm{dy}\left(t\right)}{\mathrm{dt}}=-\frac{\∂U(y\left(t\right),t)}{\∂x}+x\left(t\right)$

$=\mathrm{ay}\left(t\right)-{\mathrm{by}}^3\left(t\right)+s\left(t\right)+v\left(t\right)$

Wherein, x (t) represents input signal, it is made up of s (t) and v (t), s (t) represents purified input signal, v (t) expression location parameter is that zero fractional lower-order symmetry α stablizes the distribution additive noise signal, y (t) expression output signal or system mode, U (y (t), t) be four potential function U (y (t), t)=-(a/2) y ²(t)+(b/4) y ⁴(t), wherein a＞0, b＞0 are the parameter of nonlinear dynamic system.

Above-mentioned faint toughness Enhancement Method of bringing out electric potential signal, wherein, according to the symmetrical coefficient of covariation Corr between the output signal of the input signal of nonlinear dynamic system and nonlinear dynamic system _α,

The parameter a and the b of nonlinear dynamic system are determined in 1＜α≤2.

Above-mentioned faint toughness Enhancement Method of bringing out electric potential signal, wherein, described band is made an uproar and is brought out electric potential signal and adopt interpolation method to reduce after the normalized frequency input signal as described nonlinear dynamic system.

The present invention is faint to be brought out in the toughness Enhancement Method of electric potential signal and adopts stable distribution of α to describe noise signal, truth more can reflect reality, the synergism generation accidental resonance effect of bringing out electric potential signal and nonlinear dynamic system of utilizing band to make an uproar, make the signal to noise ratio of output signal reach maximum, improved the reliability and the accuracy of clinical nervous system injury pathological changes diagnosis and detection, improved the physiological research level of neural point from the angle of signal processing.

Description of drawings

Faint toughness Enhancement Method of bringing out electric potential signal of the present invention is provided by following embodiment and accompanying drawing.

Fig. 1 is the functional-block diagram that the faint toughness Enhancement Method of bringing out electric potential signal of the present invention is applied to detect the prolongation of evoked potential latency.

Fig. 2 is the faint sketch map that brings out the toughness Enhancement Method effect of electric potential signal of diplomatic copy invention.

The specific embodiment

Below with reference to Fig. 1～Fig. 2 faint toughness Enhancement Method of bringing out electric potential signal of the present invention is described in further detail.

The present invention is faint, and the toughness Enhancement Method of bringing out electric potential signal may further comprise the steps:

Use to adopt α stable distribute noise signal described bring out the input signal of electric potential signal as nonlinear dynamic system;

Adjust the parameter of nonlinear dynamic system, make input signal and nonlinear dynamic system synergism produce accidental resonance effect, the output signal that output signal-to-noise ratio is the highest.

The present invention is faint to be brought out in the toughness Enhancement Method of electric potential signal and adopts stable distribution of α to describe noise signal, truth more can reflect reality, the synergism generation accidental resonance effect of bringing out electric potential signal and nonlinear dynamic system of utilizing band to make an uproar, make the signal to noise ratio of output signal reach maximum, improved the reliability and the accuracy of clinical nervous system injury pathological changes diagnosis and detection, improved the physiological research level of neural point from the angle of signal processing.

The notion of accidental resonance was proposed by people such as Benzi in 1981, it has been described one and has had nonlinear bistable system, while input signal and noise under the effect of a little periodic modulation signal, when noise is strengthened to a certain intensity, signal can not reduce than not only, thereby reciprocal system can produce accidental resonance is strengthened output signal significantly, promptly there is a certain best input noise intensity, make system produce the output of highest signal to noise ratio, appeared suddenly out by the signal of noise takeover thereby make originally.

Now should be used for describing in detail faint toughness Enhancement Method of bringing out electric potential signal of the present invention with what bring out electric potential signal:

In the medical science, diagnose neural conducting state whether good by the prolongation that detects evoked potential latency;

Bring out electric potential signal and adopt following model description:

$\left\{\begin{array}{c}{x}_{1k}\left(t\right)=s_k\left(t\right)+v_{1k}\left(t\right)\\ x_{2k}\left(t\right)=s_k(t-D_k)+v_{2k}\left(t\right)\end{array}t=\mathrm{1,2}\mathrm{\ΛN};k=\mathrm{1,2},\mathrm{\ΛK}---\left(1\right)\right.$

Wherein, k represents sweep signal, and t represents discrete-time variable, x _1k(t) expression is with the reference of making an uproar to bring out electric potential signal, x _2k(t) electric potential signal that brings out to be measured of making an uproar, s are with in expression _k(t) the purified reference of expression is brought out electric potential signal, s _k(t-D _k) the purified electric potential signal that brings out to be measured of expression, D _kBe the delay of evoked potential latency, v _1k(t), v _2k(t) all represent EEG signals EEG and other noise signal, suppose that the two is incoherent, v _1k(t), v _2k(t) describe with stable distribution of α;

Figure 1 shows that the faint toughness Enhancement Method of bringing out electric potential signal of present embodiment is applied to detect the functional-block diagram of the prolongation of evoked potential latency, as shown in Figure 1, SR ₁, SR ₂All are the nonlinear dynamic systems that adopt optimized parameter, the parameter of this nonlinear dynamic system can make band make an uproar to bring out generation accidental resonance effect between electric potential signal and this nonlinear dynamic system;

Nonlinear dynamic system SR ₁Be used for the reference that reinforcing band makes an uproar and bring out electric potential signal x _1k(t) toughness promptly improves and is with the reference of making an uproar to bring out electric potential signal X _1k(t) signal to noise ratio, nonlinear dynamic system SR ₂Be used for the electric potential signal x that brings out to be measured that reinforcing band is made an uproar _2k(t) toughness promptly improves the electric potential signal x that brings out to be measured that band is made an uproar _2k(t) signal to noise ratio;

Nonlinear dynamic system SR ₁, SR ₂The available random differential equation:

$\frac{\mathrm{dy}\left(t\right)}{\mathrm{dt}}=-\frac{\∂U(y\left(t\right),t)}{\∂x}+x\left(t\right)---\left(2\right)$

$=\mathrm{ay}\left(t\right)-{\mathrm{by}}^3\left(t\right)+s\left(t\right)+v\left(t\right)$

Wherein, x (t) represents input signal, it is made up of s (t) and v (t), s (t) represents purified input signal, v (t) expression location parameter is stable (FLOA S α S) additive noise signal that distributes of fractional lower-order symmetry α of zero, y (t) expression output signal or system mode, U (y (t), t) be four potential function U (y (t), t)=-(a/2) y ²(t)+(b/4) y ⁴(t), wherein a＞0, b＞0 are the parameter of nonlinear dynamic system, four potential functions have two stable smallest point y (t)=± (a/b) ^1/2With the barrier height between one understable some y (t)=0, two smallest point be Δ U=a ²/ (4b);

Under adiabatic approximation condition, the numerical solution of above-mentioned stochastic differential equation is approached analytic solutions, still, and the primary electric potential signal x that brings out _1k(t), x _2k(t) normalized frequency is Tai Gao and do not satisfy adiabatic approximation condition often, and that uses promptly that formula (1) describes brings out electric potential signal x _1k(t), x _2k(t) normalized frequency is too high not to satisfy adiabatic approximation condition, therefore, brings out electric potential signal x with what formula (1) was described _1k(t), x _2k(t) can not be directly as nonlinear dynamic system SR ₁, SR ₂Input signal, as shown in Figure 1, adopt interpolation method to reduce the primary electric potential signal x that brings out in the present embodiment _1k(t), x _2k(t) normalized frequency, thereby, input nonlinear dynamic system SR ₁, SR ₂Signal be to bring out electric potential signal x through what interpolation method was handled _1k' (t), x _2k' (t);

Under adiabatic approximation condition, adopt Euler's method to obtain the numerical solution of above-mentioned stochastic differential equation:

y(t+1)＝y(t)+ΔT[ay(t)-by ³(t)+s(t)+v(t)] (3)

Wherein, Δ T is the time step of nonlinear dynamic system;

Under the symmetrical coefficient of covariation criterion of maximum, adjust parameter a, the b of nonlinear dynamic system, band is made an uproar bring out between electric potential signal and this nonlinear dynamic system generation accidental resonance effect, determine that the method for nonlinear dynamic system parameter a, b is: the symmetrical coefficient of covariation of establishing between the output signal of the input signal of each nonlinear dynamic system and nonlinear dynamic system is Corr _α,

${\mathrm{Corr}}_{\mathrm{\α}}(X,Y)={\mathrm{\λ}}_{X,Y}{\mathrm{\λ}}_{Y,X}=\frac{{[X,Y]}_{\mathrm{\α}}{[Y,X]}_{\mathrm{\α}}}{{[X,X]}_{\mathrm{\α}}{[Y,Y]}_{\mathrm{\α}}}$ 1＜α≤2 (4)

Wherein, X, Y represent the stable two-way stochastic signal that distributes and describe with α respectively, α represents the stable characteristic index that distributes of this two paths of signals α, under the condition of parameter b=1, ordering parameter a, a changes to 2 with the interval of step-length 0.01 from 0.01, with b=1, a gets different value substitution formula (4) respectively, and calculating symmetrical coefficient of covariation is Corr _α, symmetrical coefficient of covariation Corr _αThe a value of maximum correspondence is exactly the optimized parameter value that nonlinear dynamic system produces the accidental resonance effect, and promptly its parameter b of the nonlinear dynamic system among Fig. 1=1, parameter a get Corr _αA value (the nonlinear dynamic system SR of maximum correspondence ₁, SR ₂All adopt the definite optimized parameter separately of said method);

Have the output signal of the nonlinear dynamic system output signal-to-noise ratio maximum of optimized parameter, the promptly faint toughness of bringing out electric potential signal is enhanced, for next step prolongation that detects evoked potential latency provides convenience;

Electric potential signal x is brought out in the reference that band is made an uproar _1k(t) by nonlinear dynamic system SR ₁Obtain the high output signal y of signal to noise ratio _1k' (t), the electric potential signal x that brings out to be measured that band is made an uproar _2k(t) by nonlinear dynamic system SR ₂Obtain the high output signal y of signal to noise ratio _2k' (t);

Down-sampled processing makes the normalized frequency of the output signal of nonlinear dynamic system get back to original normalized frequency of bringing out electric potential signal;

Nonlinear dynamic system SR ₁Output signal y _1k' (t) after down-sampled processing, obtain signal y _1k(t), y _1k(t) normalized frequency equals the original electric potential signal x that brings out _1k(t) normalized frequency, nonlinear dynamic system SR ₂Output signal y _2k' (t) after down-sampled processing, obtain signal y _2k(t), y _2k(t) normalized frequency equals the original electric potential signal x that brings out _2k(t) normalized frequency;

Signal y _1k(t) and signal y _2k(t) obtain co-variation correlation function signal R through the covariant function processing _c(m), detect co-variation correlation function signal R _c(m) peak just can obtain the delay D of evoked potential latency _k, the delay D of this evoked potential latency _kWhether diagnosable neural conducting state is good.

With the faint toughness Enhancement Method of bringing out electric potential signal of computer simulating, verifying present embodiment right bring out the enhanced effect of electric potential signal, the computer program operation platform adopts Matlab:

Bring out electric potential signal according to what formula (1) structural belt was made an uproar, electric potential signal s is brought out in purified reference _k(t) the electric potential signal multiple averaging of normally bringing out by clinical acquisition obtains, and electric potential signal x is brought out in the reference that band is made an uproar _1k(t) bring out electric potential signal s by purified reference _k(t) add S α S noise signal v _1k(t) form the electric potential signal x that brings out to be measured that band is made an uproar _2k(t) by s _k(t) the signal s in D sampling interval of delay _k(t-D _k) add S α S noise signal v _2k(t) form v _2k(t) simulated the situation of evoked potential latency extension under the non-normal conditions such as anoxia asphyxia;

In the emulation, signal length is 3 scan periods, each scan period 128 points of sampling, i.e. K=3 in the formula (1), N=128; The delay D of evoked potential latency to be measured _kBe set at 10 sampling intervals; Based on the symmetrical coefficient of covariation of maximum is Corr _αThe parameter of selected nonlinear dynamic system is a=0.01, b=1; Time step Δ T in the formula (3) is 0.001, initial value y (0)=0, nonlinear dynamic system SR ₁, SR ₂The down-sampled rate of output signal is 1000: 1;

Figure 2 shows that the faint sketch map that brings out the toughness Enhancement Method effect of electric potential signal of diplomatic copy invention, Fig. 2 (a) brings out electric potential signal s (t) for the purified reference through cumulative mean, Fig. 2 (d) is for through the electric potential signal s (t-D) that brings out purified to be measured of cumulative mean, and Fig. 2 (b) brings out electric potential signal x for the reference that S α S (α=1.6) noise signal, the band of signal to noise ratio MSNR=0 make an uproar that contains of emulation ₁(t), the electric potential signal x that brings out to be measured that makes an uproar for the band that contains S α S (α=1.6) noise signal, signal to noise ratio MSNR=0 of emulation of Fig. 2 (e) ₂(t), Fig. 2 (c) is nonlinear dynamic system SR ₁Output signal y ₁(t), Fig. 2 (f) is nonlinear dynamic system SR ₂Output signal y ₂(t), as can be seen from Figure 2, being submerged in pure reference in the high power pulse S α S noise signal brings out electric potential signal s (t) and shows especially out through the accidental resonance effect, the electric potential signal s (t-D) that brings out pure to be measured that is submerged in the high power pulse S α S noise signal shows especially out through the accidental resonance effect, illustrate that the signal to noise ratio MSNR of output signal is improved by the accidental resonance effect.

Claims (5)

1. a faint toughness Enhancement Method of bringing out electric potential signal is characterized in that, may further comprise the steps:

Use the band that adopts the stable distribution of α to describe noise signal to make an uproar and bring out the input signal of electric potential signal as nonlinear dynamic system;

Adjust the parameter of nonlinear dynamic system, make input signal and nonlinear dynamic system synergism produce accidental resonance effect, the output signal that output signal-to-noise ratio is the highest.

2. faint toughness Enhancement Method of bringing out electric potential signal as claimed in claim 1 is characterized in that, described band is made an uproar and brought out electric potential signal and add that by the purified electric potential signal that brings out the stable noise signal of describing that distributes of α forms.

3. faint toughness Enhancement Method of bringing out electric potential signal as claimed in claim 1 is characterized in that, described nonlinear dynamic system adopts following stochastic differential equation to describe:

$\frac{\mathrm{dy}\left(t\right)}{\mathrm{dt}}=-\frac{\∂U(y\left(t\right),t)}{\∂x}+x\left(t\right)$

$=\mathrm{ay}\left(t\right)-{\mathrm{by}}^3\left(t\right)+s\left(t\right)+v\left(t\right)$

Wherein, x (t) represents input signal, it is made up of s (t) and v (t), s (t) represents purified input signal, v (t) expression location parameter is that zero fractional lower-order symmetry α stablizes the distribution additive noise signal, y (t) expression output signal or system mode, U (y (t), t) be four potential function U (y (t), t)=-(a/2) y ²(t)+(b/4) y ⁴(t), wherein a＞0, b＞0 are the parameter of nonlinear dynamic system.

4. faint toughness Enhancement Method of bringing out electric potential signal as claimed in claim 3 is characterized in that, according to the symmetrical coefficient of covariation Corr between the output signal of the input signal of nonlinear dynamic system and nonlinear dynamic system _α,

The parameter a and the b of nonlinear dynamic system are determined in 1＜α≤2.

5. as claim 1 or 3 described faint toughness Enhancement Method of bringing out electric potential signal, it is characterized in that described band is made an uproar and brought out electric potential signal and adopt interpolation method to reduce after the normalized frequency input signal as described nonlinear dynamic system.

Images