JP2001061801A - Device and method of wink act analysis - Google Patents

Abstract

PROBLEM TO BE SOLVED: To analyze a wink act of a subject's eye from brain wave signals. SOLUTION: The time series original (raw) data of brain waves measured by an electroencephalograph are taken in (S100). The data are wavelet-converted using a spline 4 function for the mother wavelet (S102), and resolved into wavelets to the level (-4) with the data themselves as the level 0 (S104). Then the low frequency component f as the result of the resolution to the level (-4) is extracted as a wink waveform (S106), and whether the component satisfies a wink judgement condition or not is judged (S108). Based on the result of the judgement, the occurrence, frequency, etc., of wink acts are found and displayed together with the waveform of brain waves.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus and a method for detecting and analyzing a blink action from an electroencephalogram signal.

[0002]

2. Description of the Related Art As a technique for measuring the degree of concentration of a person, for example, there is a "concentration degree estimating apparatus" described in Japanese Patent Application Laid-Open No. 9-262216. In this apparatus, a plurality of biological information are measured from the nerve activity of the subject, and the measurement information and the concentration setting rule information are used to estimate the concentration.
In addition, in the `` characteristic brain electromagnetic wave detection device '' described in JP-A-11-137530, a pulse wave, eye movement, a brain electromagnetic wave in which characteristic waves generated due to blinks and the like are mixed by wavelet analysis,
The component of the characteristic wave is determined by comparing the decomposed brain electromagnetic wave data with the biological information time-series data. This device aims to reduce the burden on EEG readers, doctors, and other people who read EEG, and provides detection means other than EEG to detect pulse waves, eye movements, and blinks. The characteristic wave is determined based on the detection signal of the means. Further, in Japanese Unexamined Patent Publication No. Hei 11-65422, "Method of evaluating physical and mental state of worker, method of controlling work content of work using device and work content control system" use detection signals of skin impedance sensor and blink sensor. , The mental and physical condition of the worker is determined. The blink sensor measures the blink frequency.

[0003]

When working toward a display such as a personal computer, it is said that the number of blinks of the worker decreases as the degree of concentration and interest increases. Thus, the information on the blinking action is an important factor in knowing the mental activity of a human.

As a technique for observing human brain activity, brain wave measurement is widely used. In order to detect the blinking information at the same time as this EEG measurement and to perform complex analysis with EEG and other biological information, it is necessary to attach a sensor such as a blink sensor to the subject in addition to the EEG meter. ,
Subject burden was heavy. Further, there is a possibility that the electroencephalograph and the blink sensor may interfere with each other, so that the reliability of the measurement data may be reduced.

In the technique described in Japanese Patent Application Laid-Open No. H11-137530, another sensor detects biological information such as a pulse wave and blinks simultaneously with an electroencephalogram. This biological information is only used to specify characteristic waves in brain waves, and this biological information is not analyzed to analyze biological phenomena such as blinks.

The present invention has been made in view of such a situation, and an object of the present invention is to provide an apparatus and a method for obtaining information about blinks simultaneously with brain waves without increasing the burden on a subject.

[0007]

In order to solve the above problem, in the present invention, time-series data of brain waves are decomposed by wavelet analysis and separated into blink waveform data and normal brain wave data. The blink confirmation information is obtained by using the separated blink waveform data.

According to the present invention, there is provided a blink motion analyzing apparatus, comprising: an electroencephalogram detecting means; a decomposing means for performing a wavelet decomposition of the time series data of the electroencephalogram detected by the electroencephalogram detecting means to a predetermined decomposition level; Blink analysis means for obtaining information on blink operation from a low-frequency component in the decomposition result of the predetermined decomposition level.

For example, in a preferred aspect, the slope of the low frequency component in the decomposition result of the predetermined decomposition level is negative, and the value of the low frequency component is α times the average of the negative values of the brain wave time series data (α is The presence / absence of a blinking operation is determined based on the fact that the value becomes smaller than a constant).

For example, in wavelet decomposition, a spline 4 function is used as a mother wavelet, and the predetermined decomposition level is decomposed to level (−4).

[0011]

BEST MODE FOR CARRYING OUT THE INVENTION As a preliminary experiment for the development of the device of the present embodiment, the inventor installed a blink sensor electrode around a subject's eye to detect blink behavior, and In parallel, the electroencephalogram was measured with an electroencephalograph. Then, by comparing the detection result of the electroencephalograph and the detection result of the blink sensor when the blink operation was performed, it was examined how the brain wave time-series data was changed by the blink operation. At this time, the subject was asked to perform the blinking operation at various time intervals, so that the electroencephalogram time-series data changed by various types of blinking operation was recorded. From this preliminary experiment, the characteristics of blink-derived components mixed in the EEG waveform were captured.

In this experiment, wavelet transform was performed on the electroencephalogram time-series data in which it was found that waveforms caused by blinks were mixed. The wavelet transform itself is a known technique,
For example, detailed information on Mathematical Science Wavelet Beginner's Guide, Susumu Sakakibara, Tokyo Denki University Press, 1995, etc. is available. Hereinafter, the wavelet transform processing used in the present embodiment will be briefly described.

The time series data of the electroencephalogram measured at time t is expressed as f
(t), the expansion coefficient of the wavelet transform for the waveform f (t) is

(Equation 1) Becomes Where Ψ is the mother wavelet
Wavelet) function, Haar, Daubesy (Dau
Various things such as bechies) and Mexican hats (MexicanHat) are known. In this example, a wavelet transform is performed using a Spline4 function, which is said to have a high frequency resolution, as a mother wavelet function. Equation (1)
In a, a is a parameter related to expansion and contraction of the mother wavelet function in the time axis direction, and corresponds to information on frequency components. Also, b is a parameter relating to the parallel translation of the mother wavelet function in the time axis direction. To apply this transformation to discretely sampled data, parameters a and b are

(Equation 2) Is defined. Then, equation (1) becomes

(Equation 3) And this inverse transformation is

(Equation 4) Can be expressed as Next, equation (4) is

(Equation 5) By using the EEG time series data f (t) as the original data f ₀ (t)

(Equation 6) I can write here,

(Equation 7) Where j represents the decomposition level, and the smaller the value of j, the longer the sampling interval becomes, so that low frequency components can be detected. As is clear from equation (7), f _j (t) can be decomposed as follows.

[0014]

(Equation 8) This is the wavelet decomposition. The second term on the right side of this equation (8) represents a wavelet expansion coefficient, that is, a high-frequency component, and the first term on the right side is a low-frequency component obtained by removing the high-frequency component from the original signal (left side). Represents The same equation (8) is applied again to the low frequency component f _j-1 obtained by this decomposition operation, and this is repeated to proceed with the decomposition.

The inventors have prepared a large number of electroencephalogram time-series data, which have been confirmed to contain waveform components due to the blinking action, and used the above wavelet decomposition by using a spline 4 function as a mother wavelet. operation alms to examine a low frequency signal f _j and the high-frequency components g _j at each decomposition level. As a result, it has been found that in most cases, the waveform component due to the blinking operation can be extracted by decomposing the level to the level (−4).

That is,

(Equation 9) It was confirmed by experiments that the brain wave component due to the blink action can be captured in f- ₄ by decomposing.

[0017] FIG. 1 shows the original waveform of the brain wave signal f ₀ used in the experiment. This waveform is in the range shown by reference numeral 1,
In addition to a normal brain wave component due to activity in the brain, a brain wave component due to blinking action is superimposed. FIG. 2 shows the signal shown in FIG.
low frequency component f ₋₁ of level (−1) obtained by decomposing f ₀ ,
FIG. 3 shows a waveform of the high-frequency component g _-1 and FIG. 3 shows a waveform of the low-frequency component f _-2 and the high-frequency component g _-2 of the level (-2) obtained by decomposing f _-1 . Similarly, FIG. 4 shows the waveform of each component (that is, f _-3 , g _-3 , f _-4 , g _-4 ) at the level (-3), and FIG. 5 shows the waveform at the level (-4). The waveform of the low-frequency component f- ₄ of the level (-4) shown in FIG. 5A has a sufficiently smooth portion corresponding to the range of code 1 of the original signal in FIG.
This represents the waveform of the brain wave component due to the blinking action.

As described above, in the preliminary experiment, the brain wave time series data measured from the electrode attached to an arbitrary position of the frontal pole was decomposed to the level (-4) using the spline 4 function for the mother wavelet. By doing so, it was found that the electroencephalogram component generated by the blinking action can be extracted. Blink components tend to appear in the signal of the frontal electrode.

FIG. 6 is a diagram schematically showing a waveform portion caused by the blinking operation in the waveform of f- ₄ shown in FIG. 5A. As shown in FIG. 6, generally, at the time of blinking, f _-4
The component generally rises sharply in the positive direction, then falls in the negative direction, and then draws a trajectory that rises in the positive direction and then gradually calms down. In the experiment, it was confirmed that there was only one negative drop in each of the blinks for the various EEGs used. Therefore, it is conceivable that blink information is obtained by detecting a portion that falls in the negative direction indicated by reference numeral 2 in FIG. In the experiment, if the blinking action is continued for a short period of time, the shape of the blinking waveform may be distorted from the typical waveform as shown in Fig. 6, but there is only one blink per negative direction. Has not collapsed.

Experiments have also confirmed that the amplitude of brain wave components due to blinks is larger than the amplitude of brain waves due to normal brain activity (other than blinks).

That is, as shown in FIG. 7, the portion 5 corresponding to one blink in the f- ₄ component always has one downgraded portion 4, and the f- ₄ component swings negatively. At the point, the absolute value of the peak is larger than the absolute value of the average level 3 of the negative value of the electroencephalogram signal.

From these, in this embodiment, the level (-
When the slope of the low-frequency component f- ₄ of the result of the wavelet decomposition in 4) is negative and the absolute value of f- ₄ becomes larger than the average of the brain wave amplitude by a predetermined ratio α or more, it is determined that the blinking operation has occurred. I did it. The reason why the ratio is set to the predetermined ratio or more is because margin for preventing erroneous determination is considered.

For this reason, the average of the value X satisfying X (i) <0 is first obtained from the time series data X (i) (i is time) of the electroencephalogram detected by the electrode attached to the frontal pole.

[0024]

(Equation 10) Here, N is the number of data extracted to obtain the average value.

Then, 2 of the following equations (11) and (12)
When the two conditional expressions were satisfied at the same time, it was determined that the blinking operation was performed.

[0026]

[Equation 11]

(Equation 12) The above is called a blink operation algorithm. This means that when the slope of f- ₄ is negative and the value becomes α times the average of the amplitude of the electroencephalogram when the eye is closed, it is determined that the waveform of the blinking action is captured.

The differential on the right side of the equation (12) is expressed by the level (-
Assuming that the sampling interval in 4) is Δt,

(Equation 13) Ask for.

Here, the coefficient α can be selected by the user while viewing the state of the displayed waveform. Experiments have shown that it is desirable to limit the selection range of α to 1.4 ≦ α ≦ 2.0. In this way, the selection range of α has a wide range because there are individual differences in the amplitude of the electroencephalogram waveform when the eyes are closed, and there is also an individual difference in the amplitude of the blinking waveform. If this happens, it may be applicable to one person but not to another.

The lower limit 1.4 and the upper limit 2.0 of the selection range of α were determined by experiments. That is, from the analysis of many cases, when α is set to a value smaller than 1.4, the brain wave waveform due to normal brain activity that is not blinking motion is
It has been found that there is a possibility that it may be erroneously determined to be a blinking motion. When α is set to a value greater than 2.0, it has been found that there is a possibility that detection omission may occur when the blink waveform is small. From the above, it is determined that the blinking operation has occurred when both equations (11) and (12) are satisfied, and the time at which both equations are satisfied is determined as the blink occurrence time.

FIG. 8 shows a schematic configuration of the apparatus of this embodiment. This apparatus includes an electroencephalograph 10, an electroencephalogram analysis display processing unit 20,
And the display device 30. The electroencephalograph 10 is a conventional general electroencephalograph. The electroencephalogram analysis display processing unit 20 performs an analysis on the blinking action on the electroencephalogram signal (time-series data) obtained by the electroencephalograph 10, and generates display image information indicating the analysis result and the electroencephalogram waveform. The display device 30 is a display device such as a CRT or a liquid crystal display that displays the display image information generated by the electroencephalogram analysis display processing unit 20.

FIG. 9 shows an electroencephalogram analysis display processing unit 2 of this apparatus.
It is a flowchart which shows the procedure of the blink operation analysis of 0.
Hereinafter, the configuration and operation of the device of the present embodiment will be described with reference to FIGS.

The electrodes of the electroencephalograph 10 are, for example, Fp on the frontal pole.
1, Fp2, frontal F3, Fz, F4, lower frontal F7, F8, central head
C3, Cz, C4, P3, Pz, P4 on parietal, O1, O2 on occiput, temporal
Attach T5 and T6 to the posterior temporal region of T3 and T4, and attach A1 and A2 to the earlobe, respectively. This method is the 10/20 method recommended as a standard method by the International EEG Society. However, this electrode mounting method is merely an example, and the method of the present embodiment does not basically depend on the electroencephalogram electrode mounting method. For example, if a new system appears in the future, the electrodes may be mounted in that system.

The raw electroencephalogram signal detected by the electroencephalograph 10 is amplified and A / D converted into time-series data, which is input to the display image generation section 26 and the wavelet decomposition section 22 of the electroencephalogram analysis display processing section 20. (S100 in FIG. 9).

The wavelet decomposition section 22 performs a wavelet transformation on the input brain wave time series data as a mother wavelet using a spline 4 function (S102), performs wavelet decomposition to a level (-4), and uses the blink. The brain wave component is separated from the brain wave component of the normal brain activity (S104). This disassembly result is obtained by the blink analysis unit 2
4 The blink analyzing unit 24 extracts a low-frequency component f- ₄ of level (-4) from the received signal group of the wavelet decomposition result (S106), and the extracted result is expressed by equations (11) and (12). It is determined whether or not the indicated blink determination condition is satisfied (S108). If the result of this determination is Yes, it is determined that there is a blinking operation, and if No, it is determined that there is no blinking operation.

In this procedure, the time at which the blink determination condition is satisfied is defined as the time at which the blink operation occurs. In addition, the blink analysis unit 2
In step 4, this determination is continuously performed on the time-series data of the series of brain wave measurement results, and the number of times satisfying the blink determination criterion is counted. Can be requested. Further, the number of blink operations per predetermined unit time interval may be counted. When the unit time interval can be set by the user, an analysis result according to the purpose of the user is obtained.

The information on the blink occurrence time, the total number of occurrences, and the number of occurrences per unit time obtained by the blink analysis unit 24 in this manner is passed to the display image generation unit 26. The display image generation unit 26 combines the information on the blink occurrence time and the number of times received with the waveform display image of the electroencephalogram signal data received from the electroencephalograph 10 and supplies the resultant to the display device 30. It is preferable to combine the blink analysis result and the display result of the electroencephalogram signal so as to match the time phases according to the processing delay in the blink analyzing unit 24.

With such a configuration, the user can obtain information on the occurrence of the blinking action and the number of times of the blinking action, along with the brain wave waveform that changes from moment to moment, on the display of the display device 30. Moreover, in this device, information on the blinking operation can be obtained simply by attaching the electrodes of the electroencephalograph 10 to the subject. Therefore, it is not necessary to attach a dedicated detection device for detecting the blinking action to the subject, so that the burden on the subject is reduced.

In the above example, the electrodes of the electroencephalograph are attached to the entire head, but only the electrodes of the frontal pole and the electrodes of the reference potential may be used for blink detection.

The blink operation analysis of the present embodiment described above is as follows.
It can be applied in real time to an electroencephalogram signal detected every moment by the electroencephalograph 10, or it can be obtained by extracting the electroencephalogram time-series data measured in the past by the electroencephalograph 10 and stored in the storage device and applied to it. Can also. In addition,
In the case of real-time processing, if a long time is required for the blink analysis processing, the real-time property of the waveform displayed on the display device 30 may be impaired. However, if a high-speed personal computer is used, there is a practical problem. The analysis can be performed in a short time.

The electroencephalogram analysis display processing section 20 can be implemented in software using a computer system such as a personal computer as a platform. In this case, a program describing the above-described processing procedure may be executed by the computer system.
This program can be provided by the vendor in the form of a portable recording medium such as a CD-ROM. The user can configure the electroencephalogram analysis display processing unit 20 by installing the program on the recording medium into his / her computer system.

The preferred embodiment of the present invention has been described above, but this is merely an example, and various modifications are included in the scope of the present invention.

For example, in the above example, the spline 4 function was used for the mother wavelet of the wavelet decomposition, but another mother wavelet function can be used. When the spline 4 function is used, the level (-
Although the electroencephalogram component due to the blinking action can be extracted by the decomposition up to 4), when another function is used, it is necessary to determine the number of decomposition levels according to the function. It is sufficient to determine to what decomposition level the decomposition should be performed by an experiment or the like, and to change the setting of the number of levels of the wavelet decomposition of the wavelet decomposition unit 22 accordingly.

Further, the illustrated blink determination condition is merely an example, and other determination conditions may be used.

[0044]

As described above, according to the present invention,
Since information about the blinking action can be extracted from the electroencephalogram measurement result, just by attaching the electroencephalogram detection device to the subject, it is possible to obtain information on the blinking action as well as the brainwave,
The burden on the subject can be reduced.

[Brief description of the drawings]

FIG. 1 is a diagram showing a waveform of original brain wave data including a component due to a blink.

FIG. 2 is a diagram showing waveforms of a low-frequency component f-1 and a high-frequency component g-1 at level (-1) obtained by subjecting original data to wavelet decomposition.

FIG. 3 is a diagram showing a waveform of a decomposition result of level (-2).

FIG. 4 is a diagram showing a waveform of a decomposition result of level (−3).

FIG. 5 is a diagram showing a waveform of a decomposition result of level (-4).

FIG. 6 is a diagram illustrating characteristics of a blink waveform.

FIG. 7 is a diagram for explaining conditions for determining a blinking waveform.

FIG. 8 is a diagram showing a device configuration of the embodiment.

FIG. 9 is a flowchart illustrating a procedure of a blink analysis process.

[Explanation of symbols]

10 electroencephalograph, 20 electroencephalogram analysis display processing section, 22 wavelet decomposition section, 24 blink analysis section, 26 display image generation section, 30 display device.

Claims (7)

[Claims] 1. An electroencephalogram detecting means, a decomposing means for performing wavelet decomposition of time-series data of an electroencephalogram detected by the electroencephalogram detecting means to a predetermined decomposing level, and a low decomposing result of the predetermined decomposing level obtained by the decomposing means. A blink analysis means for obtaining blink information from a frequency component. 2. The blink analysis means according to claim 1, wherein the low-frequency component has a negative gradient in the decomposition result of the predetermined decomposition level, and the value of the low-frequency component is α times an average of negative values of the electroencephalogram time-series data. 2. The blink operation analysis apparatus according to claim 1, wherein the presence or absence of a blink operation is determined by using a value smaller than (α is a constant) as a blink determination condition. 3. The blinking motion analyzing apparatus according to claim 2, further comprising means for receiving a setting of the constant α from a user. 4. The apparatus according to claim 2, further comprising means for analyzing the electroencephalogram time-series data and counting the number of times the blink determination condition is satisfied, thereby obtaining the number of blink operations. Blink motion analysis device. 5. A means for analyzing the electroencephalogram time series data and counting the number of times the blink determination condition is satisfied within a predetermined unit time, thereby obtaining a blink operation frequency. The blinking motion analysis device according to claim 2, wherein: 6. The method according to claim 1, wherein the decomposition unit performs a wavelet decomposition to a level (−4) as the predetermined decomposition level using a spline 4 function as a mother wavelet. Blinking motion analysis device as described. 7. A step of detecting an electroencephalogram, a step of performing a wavelet decomposition of the time-series data of the detected electroencephalogram to a predetermined decomposition level, and a low-frequency component in the decomposition result of the predetermined decomposition level of the electroencephalogram time-series data, A step of obtaining information on a blinking operation;