KR100866547B1 - A EEG apparatus with ECG Artifact elimination device and the method thereof - Google Patents

Abstract

The present invention relates to an electroencephalogram test apparatus and method including an electrocardiogram noise removing means, and more specifically, to remove the electrocardiogram noise which is removed from the electroencephalogram by estimating the electrocardiogram mixed in the scalp electroencephalogram using an adaptive digital filter based on least square acceleration. An electroencephalogram test apparatus and method comprising means.

The present invention provides an electroencephalogram test apparatus including an electrocardiogram removal means including at least a signal detector for amplifying, filtering, and converting an electroencephalogram signal detected from an electrode unit into a digital signal, and an arithmetic processing unit for calculating and processing a signal received from the signal detector. An electrocardiogram removal method included in an electroencephalogram (EKG), the method comprising: a signal input step of receiving an electrocardiogram signal including an electrocardiogram signal from the signal detector; A least squares acceleration filtering step of filtering the signal output from the signal input step by the least squares acceleration filter 220; An R wave estimating step of detecting an R wave based on a phase space method and a heartbeat interval of the output signal of the least square acceleration filtering step; An impulse output step of generating an impulse signal synchronized with the R wave detected in the R wave estimating step; An adaptive impulse correlation filtering step of filtering by an adaptive impulse correlation filter using the output signal of the signal input step as an input signal and the output signal of the impulse output step as a reference signal; Characterized in having a.

ECG, ECG, least squares acceleration filter, adaptive impulse correlation filter, R wave

Description

EEG apparatus and method with ECG noise removing means {A EEG apparatus with ECG Artifact elimination device and the method}

1 is an explanatory diagram schematically illustrating a configuration of an electroencephalogram test apparatus including an electrocardiogram noise removing unit according to an exemplary embodiment of the present invention.

FIG. 2 is an explanatory diagram for describing an overall flow of the arithmetic processing unit of FIG. 1.

3 is a flowchart of a least squares acceleration filter of the R-wave estimator of FIG. 1.

4 is a schematic flowchart of an R-wave detector of the R-wave estimator of FIG. 1.

5 is a general flowchart of an R wave estimator of FIG. 1.

6 is a schematic flowchart of the adaptive impulse correlation filter of FIG. 1.

7 is an example of an electrocardiogram signal and an electrocardiogram signal simultaneously measured.

FIG. 8 is an example of an electroencephalogram output from the least squares acceleration filter of FIG. 1.

9 is an example of an output result of the R-wave estimator of FIG. 1.

The present invention relates to an electroencephalogram test apparatus and method including an electrocardiogram noise removing means, and more particularly, to remove the electrocardiogram noise that is estimated from the electrocardiogram mixed with the scalp electroencephalogram by using an adaptive digital filter based on least squares acceleration. An electroencephalogram test apparatus and method comprising means.

In electroencephalogram, non-invasive measurement and evaluation of cerebral function is important not only in brain diseases such as epilepsy but also in sleep medicine.

Electroencephalogram tests are performed in laboratories in hospitals or laboratories according to their purpose, but recently, there is an increasing demand for ambulatory or portable electroencephalogram tests due to various advantages and needs.

The frequency domain analysis of the EEG signal is one of the standard EEG analysis methods along with the time domain analysis, and is based on classifying the frequency of the EEG signal into several bands. However, when the EEG signal includes noise, particularly when the frequency components of the EEG signal and the noise overlap, it is difficult to accurately analyze the EEG signal. In electroencephalogram measurement, the signal flowing through the electrode contains not only the electroencephalogram but also various noises such as EMG, safety, electrocardiogram, etc., and this noise causes distortion in the electroencephalogram signal and increases the error of analysis. . In particular, ECG noise is a periodic and large feature that affects the electroencephalogram signal measured in the scalp by the electrical field of the heart propagates through the body.

  Several studies have been conducted to remove ECG noise included in EEG signals. Nakamura et al. Proposed a method for removing ECG noise based on an ensemble average assuming non-stationarity of an EEG signal. In this method, the brain conduction signal is divided by synchronizing the electroencephalogram signal to the electrocardiogram measured simultaneously with the electroencephalogram. And we calculated the ensemble mean for the whole divided section, and extracted the shape of ECG included in the EEG signal, and synchronized it to the R-wave of the EKG and subtracted it from the EEG signal.

In addition to Nakamura's method, methods using independent component analysis (ICA) have been proposed.

In the independent component analysis, the ECG signals of the multiple channels including ECG noise were separated into independent components and ECGs were removed.

Conventional methods for removing ECG noise included in an EKG signal have been performed by simultaneously measuring electrocardiograms or using multiple channels of electrocardiograms including ECG noise. However, these methods are disadvantageous in mobile or portable electroencephalography devices that have limited storage capacity and processing capacity because they use multiple biosignals.

Accordingly, the present invention provides an electroencephalogram test apparatus and method capable of processing in real time and including electrocardiogram noise removing means for detecting electrocardiogram noise included in a single channel electroencephalogram. Therefore, the electrocardiogram noise removing means of the present invention is applicable to a mobile or portable electroencephalogram test apparatus.

The present invention also provides an electroencephalogram test apparatus and method including an electrocardiogram noise removing means for effectively removing the electrocardiogram from the electroencephalogram by estimating the electrocardiogram mixed with the scalp electroencephalogram using a least square acceleration-based adaptive digital filter.

SUMMARY OF THE INVENTION The present invention provides an electroencephalogram test apparatus and method including an electrocardiogram noise removing means for estimating electrocardiogram mixed with scalp electroencephalogram using an adaptive digital filter based on least square acceleration, and removing the electrocardiogram noise.

Another technical problem to be achieved by the present invention can be processed in real time, by detecting the electrocardiogram noise included in the single-channel electrocardiogram and remove the included electrocardiogram noise, ECG noise removal means that can be used in a mobile or portable ECG test device It is to provide an electroencephalogram test apparatus and method having a.

Hereinafter, an electroencephalogram test apparatus and method including an electrocardiogram noise removing unit according to an exemplary embodiment of the present invention will be described in detail with reference to the accompanying drawings.

1 is an explanatory diagram schematically illustrating a configuration of an electroencephalogram test apparatus including an electrocardiogram noise removing unit according to an exemplary embodiment of the present invention.

The electroencephalogram test apparatus 10 includes an electrode unit 100, an amplifier 130, a filter unit 170, an A / D converter 190, an operation processor 200, a memory unit 300, and a battery unit 350. ) And an output unit 330.

The electrode unit 100 is composed of electrodes for detecting an electroencephalogram from the scalp. The electrode unit 100 includes three electrodes, an input electrode, a reference electrode, and a ground electrode. Here, the input electrode is attached to the head to measure the signal, the reference electrode is attached to the right earlobe or the left earlobe, and the ground electrode is attached to the non-moving part of the body, for example, forehead, behind the ear, behind the neck, etc. I can attach it.

The amplifier 130 receives and amplifies the output signal of the electrode unit 100.

The filter unit 170 removes noise from the output signal of the amplifier 130. The filter unit 170 may use a bandpass filter of 0.1 Hz to 40 Hz.

The A / D converter 190 converts the output signal of the filter unit 170 into a digital signal.

The calculation processing unit 200 detects the electrocardiogram signal of the electrocardiogram signal from the electroencephalogram signal which is the output signal of the A / D converter 190 and removes the detected electrocardiogram signal from the electroencephalogram signal. The operation processor 200 includes an R-wave estimator 210 and an ECG estimator and remover 260.

The R-wave estimator 210 detects the R-wave by highlighting the spike-shaped R-wave from the signal output from the A / D converter 190 and detects an impulse synchronized with the R-wave from the R-wave estimator 210. . The R wave estimator 210 includes a least squares acceleration (LSA) filter 220 and an R wave detector 230. That is, the R wave estimator 210 highlights the spike-shaped R wave using the least square acceleration filter 220. Then, the R wave detector 230 detects an R wave based on a phase space method and an R-R interval, and detects an impulse synchronized with the R wave therefrom.

The ECG estimator 260 estimates and removes an ECG signal using the signal output from the A / D converter 190 and the output signal of the R-wave estimator 210. The ECG estimator 260 includes an adaptive impulse correlated filter (AICF) 280.

The memory unit 300 stores the processed electroencephalogram signal received from the operation processor 200.

The output unit 330 displays an output signal of the operation processor 200 and outputs the same to a printer.

The battery unit 350 includes a battery for driving the electroencephalogram test device 10.

FIG. 2 is an explanatory diagram for describing an overall flow of the arithmetic processing unit of FIG. 1.

In the signal input step, an electrocardiogram signal including an electrocardiogram signal is received from the A / D converter (S10).

As a least square acceleration filtering step, the smallest square acceleration filter 220 filters (S20).

In the R-wave detection step, the R-wave detector 230 detects the R-wave based on the phase space method and the heartbeat interval (S30).

In the impulse output step, an impulse signal synchronized with the detected R wave is output (S40). That is, signals synchronized with the R wave are 1 and the rest are 0.

 In the adaptive impulse correlation filtering step, the output signal of the signal input step S10 is used as an input signal, and the output signal of the impulse output step S30 is used as a reference signal, and then filtered by the adaptive impulse correlation filter 280 (S50).

The EKG signal from which the ECG signal is removed from the adaptive impulse correlation filter 280 is output (S60).

Next, the procedure of the calculation processing unit 200 will be described in more detail.

The calculation processor 200 includes an R-wave estimator 210 and an ECG estimator and remover 260, among which the R-wave estimator 210 includes a least square acceleration filter 220 and an R-wave detector 230. Is made of.

First, the least square acceleration filter 220 of the R-wave estimator 210 will be described.

Prior to the description of the least square acceleration filter 220, a general least square acceleration filter will be described.

Least Squares Acceleration Filters are filters that are used to accurately measure sharpness in situations where signal complexity and magnitude change or noise is present. Because it can be implemented in real time, it has a strong advantage in noise. That is, the least square acceleration filter is a filter in the form of a finite impulse-response (FIR) and is a method for estimating the derivative or sharpness of a discrete signal. Least squares acceleration filters are well known and will be described schematically here.

The least squares acceleration filter approximates a discrete signal consisting of p samples to an nth order polynomial that minimizes mean squared error (MSE) and then calculates the nth derivative of the polynomial. In this case, since the n-order differential is used for the n-th order polynomial, the coefficient of the n-th order polynomial can be used as it is without complex calculation. The coefficient of the least square acceleration filter is calculated in the closed form. The second derivative is called the least square acceleration filter because it has the same meaning as the estimation of the acceleration of the signal.

Time series signal {(x _i , y _i ), i = 1,... … , p} is the data to be interpolated by the second order polynomial P (x) = a ₂ x ² + a ₁ x + a ₀ , the mean square error

라는 식을 이용할 수 있다. To find the best coefficients a ₂ , a ₁ , a ₀ to minimize

Can be used.

X _ij = X _{i3 -j} and Y = [y ₁ ... y _p ] 'for the interval 1≤i≤p, 1≤j≤3. X 'represents that X is transposed and (X'X) ⁺ represents the Moore-Penrose pseudoinverse of the symmetric matrix X'X. If A = X'X, then the elements of matrix A

It is obtained through the equation

The equation for obtaining the optimal a ₂ prediction was obtained by obtaining, multiplying, and simplifying the symbol pseudoinverse of matrix A. Since the equation for a ₂ is linear for Y (e.g., the data used to obtain the acceleration prediction), this prediction may be obtained from the output of the FIR filter with some delay.

Acceleration predictions are

Is given by

는 Z보다 작거나 같은 최대의 상수를 나타낸다.Where {y _k , k = 1,2, ...} is the signal to analyze, p is the number of points (order of FIR filter) used for the parabolic fit,

Denotes the largest constant less than or equal to Z.

The coefficient F (p, dt) = {Fj (p, dt), j = 1, ..., p} of the filter depends only on the order p and the time step dt of the signal to be analyzed. For a fixed value of p these coefficients can be calculated only once and stored for later application to the filter. LSA filter coefficients

 It can be calculated from the equation

In other words, the output of the least squares acceleration filter is an acceleration prediction value obtained by equation (3), and the coefficient of the filter in equation (3) is obtained by equation (4).

Here, the procedure of the least squares acceleration filter applying the p point, n th order LSA filter to the M data is as follows.

First, the coefficients (a (i), i = 1, 2, ..., p) of the LSA filter are obtained as functions of p and n.

Second, read data b (k), (where k = 1, 2, ..., M).

Third, find the output of the LSA filter.

Where p = 3, the output of the LSA filter is

 LSA (1) = b (1) * a (1) + b (2) * a (2) + b (3) * a (3)

 LSA (2) = b (2) * a (1) + b (3) * a (2) + b (4) * a (3)

 LSA (3) = b (3) * a (1) + b (4) * a (2) + b (5) * a (3)

 LSA (4) = b (4) * a (1) + b (5) * a (2) + b (6) * a (3)

 LSA (5) = b (5) * a (1) + b (6) * a (2) + b (7) * a (3)

   . . . .

   . . . .

 LSA (M-2) = b (M-2) * a (1) + b (M-1) * a (2) + b (M) * a (3)

   . . . .

   . . . .

Fourth, listing the output of the LSA filter thus obtained, that is, LSA (1), LSA (2), ..., LSA (M-2), ..., ... LSA filter output.

3 is a flowchart of a least squares acceleration filter of the R-wave estimator of FIG. 1.

In FIG. 3, v (n) is an electroencephalogram signal including an electrocardiogram, u (n) is the output of the least square acceleration filter, and b1 to b9 are the least square acceleration filter coefficients.

Here, b1 = 0.0606, b2 = 0.0152, b3 = -0.0173, b4 = -0.0368, b5 = -0.0433, b6 = -0.0368, b7 = -0.0173, b8 = 0.0152, b9 = 0.0606.

That is, the sum of multiplying the above coefficients by a total of 9 samples including the current sample of the electrocardiogram signal including the electrocardiogram and up to 8 points, becomes the output value of the least square acceleration filter of the current sample.

In the present invention, the order of the least squares acceleration filter may be any order such as 3rd, 9th, 12th order. In the present invention, the description was made using the third or ninth order for convenience of description.

Next, the R wave detector 230 of the R wave estimator 210 will be described.

The R-wave detector 230 detects the R-wave at the output of the least square acceleration filter 220 by using a phase space method and a heartbeat interval. That is, the R wave detector 230 uses a phase space method and a heartbeat interval to detect the R wave from the first differential signal of the estimated EEG output from the least square acceleration filter 220.

4 is a schematic flowchart of an R-wave detector of the R-wave estimator of FIG. 1. That is, FIG. 4 is a procedure of detecting the R wave in the R wave detector 230.

The output of the least squares acceleration filter 220 is received (S100).

In the threshold comparison step, points that are less than or equal to the threshold (th, th) in the phase space of the output of the least squares acceleration filter 220 are detected as first order candidates (S110). That is, in the phase space of the least square acceleration filter output (the x-axis is the filter output signal and the y-axis is the signal delayed by one point), the points below the threshold (0, 0) are detected and are called primary candidates.

As the minimum point detection step, the smallest point is selected from consecutive primary candidates, and this is called a secondary candidate (S120).

In the second candidate interval determination step, the interval between the secondary candidates is calculated, and R waves are estimated by selecting points that are 70% or more of the averages of five consecutive secondary candidate intervals (S130). Here, the average of five consecutive secondary candidate intervals refers to the average of the previous five secondary candidate intervals of the current sample.

As an R-wave synchronous impulse configuration step, an impulse synchronized to the estimated R wave is configured (S140). That is, it outputs 1 from the R pine sample and 0 from all the samples other than the R wave.

5 is a general flowchart of an R wave estimator of FIG. 1. That is, a flowchart of estimating the R wave through the least square acceleration filter and the R wave detector, in which the procedures of FIGS. 3 and 4 are combined.

 As a least square acceleration filtering step, the ninth order least square acceleration filter is filtered (S105).

In the threshold comparison step, the threshold (0, 0) in the phase space of the least square acceleration filter output (x-axis is the ninth order least square acceleration filter output signal, y-axis is a signal delayed by one point at the ninth order least square acceleration filter output). If it is determined below (S115), if it is less than the threshold (0, 0), it returns to the least square acceleration filtering step (S105). In other words, it detects the points below the threshold (0, 0) in the phase space of the least square acceleration filter output (x-axis is the ninth-order least square acceleration filter output signal, and y-axis is one delayed signal from the ninth-order least square acceleration filter output). It is called a candidate for tea.

In the threshold comparison step, if the phase space of the least square acceleration filter output (x-axis is the filter output signal, y-axis is one point delayed signal) or less than the threshold (0, 0), it is detected as a first order candidate (S120).

In the storing step, the primary candidate is stored (S125).

 As the minimum point selecting step, the smallest point is selected from consecutive primary candidates and this is called a secondary candidate (S130).

In the second candidate interval determination step, the secondary candidate calculates the interval between the previous minimum points, that is, the previous secondary candidates, and determines whether the interval is 70% or more of the average of five consecutive second candidate candidate intervals. If less than 70%, the process returns to the least square acceleration filtering step S105. In other words, the interval between the secondary candidates is calculated and the R wave is estimated by selecting the points that are 70% or more of the mean of the five consecutive candidate secondary intervals.

As an R-wave synchronous impulse configuration step, in the second candidate interval determination step, if 70% or more, an impulse synchronized to the estimated R wave is configured (S140).

Next, the calculation processing unit 200 will be described with respect to the ECG estimation and removal unit 260.

The ECG estimator 260 estimates and removes an ECG signal using the signal output from the A / D converter 190 and the output signal of the R-wave estimator 210. The ECG estimation and removal unit 260 includes an adaptive impulse correlation filter 280.

Therefore, the adaptive impulse correlation filter 280 will be described here.

6 is a schematic flowchart of the adaptive impulse correlation filter of FIG. 1.

In the signal input step, an impulse signal u (n) synchronized to the estimated R wave is received from the R wave detector 230, and an electroencephalogram signal d (n) including an electrocardiogram is also received from the A / D converter. Receive (S200).

In the weight calculation step, the weight w (n) is obtained (S205). That is, the weight is a value for ECG estimation from an impulse signal synchronized to the estimated R wave.

The weight w (n) is obtained by the following equation.

Where w (n) is a weight, u (n) is a reference input, i.e. an impulse signal synchronized to the estimated R wave, e (n) is an error signal, and the received pure EKG signal from which the ECG signal has been removed to be.

In the weighting step, the impulse signal synchronized with the estimated R wave is multiplied by the weight w (n) (S210). That is, the signal y (n) obtained by multiplying the impulse signal synchronized with the estimated R wave by the weight w (n) is an ECG signal.

In the pure signal detection step, in the output signal of the A / D converter, i.e., the electroencephalogram signal including the electrocardiogram, the impulse signal synchronized with the estimated R wave multiplied by the weight (w (n)) (y (n)) Subtracting to obtain a pure signal (e (n)) that does not contain an electrocardiogram (S220). In addition, the pure signal e (n) not including the obtained ECG is stored to obtain the next weight w (n).

This is expressed as a formula as follows.

e (n) = d (n) -y (n)

Here, e (n) is an error signal, a received pure electroencephalogram signal from which an ECG signal is removed, d (n) is an injection force, an electroencephalogram signal including an electrocardiogram, and y (n) is an estimated ECG signal. , That is, the reference input × weight.

The output e (n) of the pure signal detection step is output as a pure electroencephalogram signal from which the ECG is removed (S240).

7 is an example of an electrocardiogram signal and an electrocardiogram signal simultaneously measured.

FIG. 7A is an electroencephalogram signal in which ECG noise is introduced and measured in a general manner, that is, an output signal of the A / D converter of the present invention. FIG. 7B is measured simultaneously with FIG. 7A and is an ECG signal measured through a general limb induction method.

The hatched portions in FIGS. 7A and 7B indicate a time point at which an R wave of an electrocardiogram is generated, and a blue circle is a peak point of an R wave. It can be seen that the ECG signal is introduced into the EKG signal measured in the general manner in FIG. 7.

FIG. 8 is an example of an electroencephalogram output from the least squares acceleration filter of FIG. 1. FIG. 8 compares an output by driving a general electrocardiograph at the same time as driving the least square acceleration filter 220 of the present invention, and compares the electroencephalogram and the phase space of the electrocardiogram after the least square acceleration filter.

8A shows an electroencephalogram signal of one channel into which ECG noise is introduced (the dotted line at the top of FIG. 8A), and the measured ECG signal (the dotted line at the bottom of FIG. 8A) at the same time; The output signal of the ninth-order least squares acceleration filter 220 (the solid line in the middle in FIG. 8A) is shown. FIG. 8B shows the phase space of the ECG signal of FIG. 8A, and FIG. 8C shows the phase space of the ninth-order least squares acceleration filter 220 output result.

As shown in FIG. 8, it can be seen that the QRS signal including the R wave of the electrocardiogram shows a very large trajectory in phase space. In the general ECG signal, the R wave is mainly located in the first quadrant, and in the case of the least square acceleration filter signal used in the present invention, the R wave is mainly located in the third quadrant. Therefore, in the present invention, the R wave is detected by using a phase space method having a two-dimensional mapping dimension and a delay time of one sample. FIG. 8 is a diagram showing a phase space method, where the x-axis is an electroencephalogram signal and the y-axis is a signal delayed by one sample. That is, it is assumed in the present invention that the point (secondary candidate) is located at the furthest distance among the points (primary candidate) located in the third quadrant with respect to the origin, and the interval between the secondary candidates is calculated and continuous Points that are 70% or more of the average of the five second candidate intervals are selected and estimated by the R wave, and an impulse is generated in synchronization with the estimated R wave.

9 is an example of an output result of the R-wave estimator of FIG. 1.

9A is an output signal of the A / D converter 190 and is an electroencephalogram signal into which an ECG signal is introduced. FIG. 9B shows the result of the least squares acceleration filter 220 and the first-order candidates (parts indicated by ○ in FIG. 9B). FIG. 9C shows the results of the least squares acceleration filter 220 and the second-order candidates (parts indicated by ○ in FIG. 9C). FIG. 9 (d) shows the result of the least squares acceleration filter 220 and the R wave estimation result (part of (d) of FIG. 9). FIG. 9E shows that an impulse (part shown by □ in FIG. 9E) is configured in accordance with the R wave estimation result. According to the impulse (square) of FIG. 9E, the part estimated by R wave in FIG. 9A (part shown by (square) of FIG. 9A) is shown.

That is, Figure 9 shows the result of generating the impulse synchronized with the R wave by detecting the R wave of the electrocardiogram included in the electroencephalogram by the method according to the present invention, the R wave estimator 210 properly estimates the R wave It can be seen.

The present invention is not limited to the above described and illustrated in the drawings, and of course, more modifications and variations are possible to those skilled in the art within the scope of the following claims.

As described above, the electroencephalogram test apparatus and method of the present invention includes electrocardiogram noise removing means for estimating the electrocardiogram mixed in the scalp electroencephalogram using an adaptive digital filter based on least square acceleration and removing the electrocardiogram from the electroencephalogram. In addition, the electroencephalogram test apparatus and method of the present invention can be processed in real time, by detecting the electrocardiogram noise included in the single-channel electrocardiogram and remove the included electrocardiogram noise, ECG noise removing means that can be used in a mobile or portable electrocardiogram test apparatus It is provided.

That is, the present invention does not need to simultaneously measure the electrocardiogram or the electrocardiogram of multiple channels including the electrocardiogram noise in order to know the information of the electrocardiogram noise as in the conventional method. It can be used as an advantage in the portable ECG device, and can be processed in real time by the mobile or portable ECG device, and can effectively remove ECG noise based on the method of detecting ECG noise included in the single channel ECG and the adaptive digital filter. Provide a new way.

Claims (16)

An electrode unit having electrodes for detecting an electroencephalogram, a signal detector for amplifying, filtering, and converting an electroencephalogram signal detected from the electrode unit into a digital signal, an arithmetic processing unit for arithmetic processing a signal received from the signal detecting unit, and the arithmetic processing unit An electroencephalogram test device comprising electrocardiogram noise removing means having an output portion for displaying an output signal of The operation processing unit An arithmetic processing including least squares acceleration filtering is performed from the signal received from the signal detection unit to detect spike-shaped R waves (R waves of electrocardiogram), and impulses synchronized to the R waves (that is, An R-wave estimator for outputting 1 from the R-pine sample and outputting 0 from all samples other than the R-wave); An electrocardiogram estimator and estimator for estimating and removing an electrocardiogram signal using an electroencephalogram signal received from the signal detector and an output signal of the R wave estimator; Electroencephalogram test device having an electrocardiogram noise removing means characterized in that it comprises a. The method of claim 1, The R wave estimating unit A least squares acceleration filter that highlights spike-shaped R waves in the signal received from the signal detector; An R wave detector for detecting an R wave from a signal received from the least square acceleration filter using a phase space method and an R-R interval, and outputting an impulse synchronized with the R wave; Electroencephalogram test device having an electrocardiogram noise removing means characterized in that it comprises a. The method of claim 1, And an electrocardiogram estimating and removing unit including an adaptive impulse correlated filter (AICF). The method of claim 1, The electrode unit is an electroencephalogram test device having an ECG noise removing means, characterized in that it comprises three electrodes of an input electrode, a reference electrode, and a ground electrode. The method of claim 1, The EEG test apparatus having an ECG noise removing means, characterized in that the signal detection unit comprises a bandpass filter of 0.1Hz to 40Hz. The method of claim 1, And an electrocardiogram noise removing means (100) for storing the processed electroencephalogram signal received from the arithmetic processing unit. EKG in the electroencephalogram test apparatus having an electrocardiogram removal means comprising at least a signal detector for amplifying, filtering and converting the electroencephalogram signal detected from the electrode into a digital signal; In the included ECG removal method, A signal input step of receiving an electroencephalogram signal including an electrocardiogram signal from the signal detection unit; A least squares acceleration filtering step of filtering the signal output from the signal input step by the least squares acceleration filter 220; An R wave estimating step of detecting an R wave based on a phase space method and a heartbeat interval of the output signal of the least square acceleration filtering step; An impulse output step of generating an impulse signal synchronized with the R wave detected in the R wave estimating step; An adaptive impulse correlation filtering step of filtering by an adaptive impulse correlation filter using the output signal of the signal input step as an input signal and the output signal of the impulse output step as a reference signal; Electrocardiogram removal method included in the electroencephalogram characterized in that it comprises a. The method of claim 7, wherein And the output of the adaptive impulse correlation filtering step is an electrocardiogram signal from which an electrocardiogram signal has been removed. EKG in the electroencephalogram test apparatus having an electrocardiogram removal means comprising at least a signal detector for amplifying, filtering and converting the electroencephalogram signal detected from the electrode into a digital signal; In the included ECG removal method, Performing an arithmetic processing including least squares acceleration filtering from the signal received from the signal detection unit to detect spike-shaped R waves (R-wave of electrocardiogram), and to detect impulses synchronized to the R waves. R wave detection step; An electrocardiogram estimating and removing step of estimating and removing an electrocardiogram signal using the output signal of the R-wave detecting step and the signal received from the signal detector; Electrocardiogram removal method included in the electroencephalogram characterized in that it comprises a. The method of claim 9, The R wave detection step A least squares acceleration filtering step of highlighting a spike-shaped R wave by least squares acceleration filtering the signal received from the signal detection unit; An R wave estimating step of detecting an R wave using a phase space method and an R-R interval in a signal received from the least squares acceleration filtering step; An impulse output step of generating an impulse signal synchronized with the R wave detected in the R wave estimating step; Electrocardiogram removal method included in the electroencephalogram characterized in that it comprises a. The method of claim 9, The ECG estimation and removal step An electrocardiogram included in an electrocardiogram comprising an adaptive impulse correlation filtering step of filtering the signal received from the signal detector as an input signal and filtering the adaptive signal using an output signal of the R-wave estimating step as a reference signal. How to remove. The method according to any one of claims 7 to 10,  The R wave estimating step The phase space of the output of the least square acceleration filtering step (x-axis is the output signal of the least square acceleration filtering step, the y axis is a signal delayed by one point in the output signal of the least square acceleration filtering step) is less than or equal to the threshold (0, 0) A threshold comparison step of detecting points and making them primary candidates; Selecting a minimum point in successive primary candidates at the output of the threshold comparison step and detecting the minimum point as a secondary candidate; A second candidate interval determination step of calculating intervals between secondary candidates and selecting points that are 70% or more of the averages of five consecutive secondary candidate intervals and estimating them as R waves; Electrocardiogram removal method included in the electroencephalogram characterized in that it comprises a. The method according to any one of claims 7 to 10, The impulse output step of the ECG removal method included in the electroencephalogram characterized in that to output a 1 for the R-pine sample, and to output all 0 for the sample other than the R-wave. The method according to any one of claims 7 to 11, The adaptive impulse correlation filtering step A signal input step of receiving an impulse signal u (n) synchronized to the R wave from an R wave detector and receiving an electroencephalogram signal d (n) including an electrocardiogram from the signal detector; A weight calculation step of calculating the weight w (n); A weighting step of estimating a signal y (n) obtained by multiplying the impulse signal synchronized with the R wave by the weight w (n) as an electrocardiogram signal; From the output signal d (n) of the signal detector, a signal obtained by subtracting a signal y (n) obtained by multiplying the impulse signal synchronized with the R wave by a weight w (n), is pure EEG without an electrocardiogram. A pure signal detection step of obtaining as a signal e (n); Electrocardiogram removal method included in the electroencephalogram characterized in that it comprises a. The method of claim 14, The electroencephalogram signal e (n), which does not include the electrocardiogram obtained in the pure signal detection step, is stored to obtain the next weight w (n). The method of claim 14, The weight w (n) is

Where w (n) is a weight, u (n) is a reference input, i.e., an impulse signal synchronized to the estimated R wave, and e (n) is an error signal. Pure EEG signal) Electrocardiogram removal method included in the electroencephalogram, characterized in that obtained by.

Images