JP5360895B2 - Visual evoked potential signal detection system - Google Patents

Description

  The present invention relates to a visual evoked potential signal detection system for detecting a visual evoked potential (VEP) signal induced in a brain by a visual stimulus.

  It is known that when a visual stimulus is given to a human, a visual evoked potential is induced in the brain. In particular, a visual evoked potential induced by a steady visual stimulus of about 5 Hz to 60 Hz is called a Steady-State Visual Evoked Potential (SSVEP). Such visual evoked potentials are observed by all as a signal component included in the electroencephalogram representing the electrical activity of the brain, unless there is an abnormality in the visual conduction path such as the eyeball, optic nerve, cortical visual cortex, etc. It occurs mainly near the visual cortex of the cerebral cortex.

  Such a visual evoked potential is observed as a signal having a time frequency substantially similar to that of a visual stimulus. That is, the visual evoked potential signal (hereinafter referred to as the visual evoked potential signal) is generated when the visual stimulus source is a light source 101 such as an LED (Light Emitting Diode) that blinks at a predetermined frequency, as shown in FIG. Reflects the cellular response of the cerebral cortex visual cortex 102 of the observer OB, and becomes a rhythmic waveform substantially synchronized with the frequency of the blinking drive signal of the light source 101, for example, as shown in FIG. The visual evoked potential signal that is induced when attention is directed to a predetermined visual stimulus has an amplitude value and a degree of phase synchronization with the stimulus that are greater than those induced by other visual stimuli. It is known.

  Conventionally, it has been tried to apply to various fields such as rehabilitation devices and brain computer interfaces by detecting the visual evoked potential signal from the measured electroencephalogram, paying attention to the physiological phenomenon related to such visual evoked potential. Yes.

  For example, in Patent Document 1, gaze stimulation light is identified from components related to the stimulation light signal included in the induced electroencephalogram generated by gazing at any one of a plurality of stimulation lights, and assigned to the identification result. A technique for operating the selected switch is disclosed.

  Further, Patent Document 2 analyzes the frequency distribution of an electroencephalogram signal measured when the blinking light is visually observed, creates movement information of the pointer in the blinking light direction based on the analysis result, and creates the created movement information. A technique for displacing the display position of the pointer displayed on the display based on the above is disclosed.

  Here, the visual evoked potential is observed as a signal component included in the electroencephalogram as described above. However, considering an application form using the visual evoked potential, it is desirable to perform the electroencephalogram measurement non-invasively. As this type of electroencephalogram measurement method, as shown in FIG. 6, there is a method of detecting and recording the electrical activity of nerve cells in the brain using the electrode 103 placed on the scalp of the observer OB. However, since various signals are included in the electroencephalogram measured by such a method, when the visual evoked potential signal is applied to a brain computer interface or the like, the weak visual evoked potential signal is appropriately detected, It is necessary to increase the correct answer rate.

  As one of the results of research conducted on this point of view, Non-Patent Document 1 has a correct answer rate of 87.5% when 48 channels of data with an average data length of 3.8 seconds are input. A brain computer interface is disclosed. The experiment disclosed in Non-Patent Document 1 shows that the blinking frequency of the light source is 0 in a low frequency band (6 Hz to 16 Hz) where the signal-to-noise ratio (hereinafter referred to as SNR) of the electroencephalogram signal is good. .Changed in steps of 2 Hz. In Non-Patent Document 1, as a result of experiments, when the blinking frequency of the light source is in a band of 8 Hz to 10 Hz, it is easily affected by α waves, and the best result is obtained when the blinking frequency is 10 Hz to 12 Hz. It is described that is obtained. In this experiment, the high frequency band (25 Hz to 50 Hz) is not performed because the portion from which the amplitude of the electroencephalogram signal can be derived is small compared to the case of the low frequency band.

  Non-Patent Document 2 discloses a brain computer interface that assumes telephone button input. Specifically, in this non-patent document 2, light sources are assigned to 0-9 numeric keys, enter key, backspace key, and on / off key that constitute a telephone, and the correct answer rate for these 13 channel data is shown. The state of verification is disclosed. In this experiment, as the blinking frequency of the light source, a band that does not overlap with the α-wave band in the low frequency band of 6 Hz to 14 Hz is used, and the average transfer rate is 27.15 bits / minute. .

  Furthermore, Non-Patent Document 3 discloses an experiment in which a character input interface is actually applied to a patient with Amyotrophic Lateral Sclerosis (ALS) using the event-related potential P300. As a result of this experiment, the offline correct answer rate was 81.51%, and the online correct answer rate was 61.98%.

JP-A-10-97369 Japanese Patent Laid-Open No. 11-73286

Xiaorong Gao, DingfengXu, Ming Cheng, and Shangkai Gao, “A BCI-Based Environmental Controller for the Motion-Disabled”, IEEE Transactions on Neural Systems and Rehabilitation Engineering, June 2003, Vol. 11, No. 2, p. 137-140 Ming Cheng, Xiaorong Gao, Shangkai Gao, and Dingfeng Xu, “Design and Implementation of a Brain-Computer Interface with High Transfer Rates”, IEEE Transactions on Biomedical Engineering, October 2002, Vol. 49, No. 10, p. 1181-1186 F. Nijboer, EWSellers, J. Mellinger, MA Jordan, T. Matuz, A. Furdea, S. Halder, U. Mochty, DJ Krusienski, TM Vaughan, JR Wolpaw, N. Birbaumer, A. Kubler, `` A P300 -based brain-computer interface for people with amyotrophiclateral sclerosis ", Clinical Neurophysiology 119, 2008

  By the way, the visual evoked potential signal generally has a better SNR and a higher correct answer rate as the flashing intensity of the light source is stronger. For this reason, in the conventional techniques including the experiments described in Non-Patent Document 1 to Non-Patent Document 3 described above, a light source having as high a blinking intensity as possible is used.

  However, if the blinking intensity is high, it will be very dazzling for the observer who views the blinking light, and this will place a burden on the observer. For this reason, a light source with high flashing intensity is not suitable for long-time use, and there is a problem that it impedes practical use. On the other hand, when the blinking intensity of the light source is lowered, there is a problem that the response of the visual evoked potential falls, and as a result, the SNR of the visual evoked potential signal observed deteriorates. Photosensitive seizures may also occur when exposed to flashing light stimuli.

  Conventionally, as in the experiments described in Non-Patent Document 1 to Non-Patent Document 3 described above, a light source having a blinking frequency in a low frequency band (about 5 Hz to 16 Hz) with a relatively large response of the visual evoked potential is used. Used. However, assuming a form such as a brain computer interface in which multiple light sources with different flashing frequencies are installed in a narrow area such as a single board, multiple flashing lights with low flashing frequencies will enter the observer's field of view. This leads to a situation where a person excessively cares about flickering of flashing light, and there is a problem that it is not practical.

  In addition, this low frequency band often overlaps with the alpha wave band (about 8 Hz to 13 Hz) observed entirely at rest. Therefore, when a light source having a blinking frequency in such a band is used, an electroencephalogram signal in which a visual evoked potential and an α wave are mixed is observed, and normal power spectrum analysis such as Fourier transform is performed. However, there is a problem that the visual evoked potential signal cannot be properly discriminated and the correct answer rate is lowered. Furthermore, in this case, when a light source having a blinking frequency in a band that does not overlap with the α wave band is used, there is a problem that the blinking frequency becomes an extremely narrow band, making practical application difficult.

  The present invention has been made in view of such circumstances, greatly reducing the burden on the observer by suppressing the glare and flickering of the flashing light, and is extremely suitable for long-time use and excellent practicality. It is an object of the present invention to provide a visual evoked potential signal detection system capable of exhibiting the above.

  The inventor of the present application diligently searched for a method that does not cause the observer to feel dazzling or flickering without reducing the blinking intensity of the light source itself, which causes the deterioration of the SNR of the visual evoked potential signal. More specifically, the brightness when the viewer visually recognizes that the flashing light is continuously lit even though the flashing light is flashing is an average value of luminance. , I thought about using the so-called Talbot-Plateau rule that expresses the characteristics of the human eye.

That is, the visual evoked potential signal detection system according to the present invention that achieves the above-described object is a visual evoked potential signal detection system that detects a specific visual evoked potential signal included in an electroencephalogram, and is an observation in which a light source is provided. An object and a signal processing unit that processes an electroencephalogram signal of an observer who visually recognizes the light source, and the observation object is continuously lit for the observer even though blinking light is blinking. The signal processing unit blinks the light source at a blinking frequency such that it is visually recognized , and performs a cross spectrum analysis to calculate a correlation in the frequency domain between the driving signal of the light source and the electroencephalogram signal, and detecting the visual evoked potential signals based on the analysis result. Further, a visual evoked potential signal detection system for detecting a specific visual evoked potential signal included in an electroencephalogram, which is an observation target provided with a light source and a signal for processing an electroencephalogram signal of an observer viewing the light source and a processing unit, the observation object to blink the light source flicker frequency to the extent that flashing light is visually recognized as lit continuously for to have despite the observer blinks The signal processing unit performs cross spectrum analysis between the driving signal of the light source and the electroencephalogram signal while overlapping the analysis windows and shifting by a predetermined time, and adds a vector on a complex plane representing the cross spectrum. based on the phase and amplitude of the vector after summing the resulting you and detecting the visual evoked potential signals. Further, a visual evoked potential signal detection system for detecting a specific visual evoked potential signal included in an electroencephalogram, which is an observation target provided with a light source and a signal for processing an electroencephalogram signal of an observer viewing the light source and a processing unit, the observation object to blink the light source flicker frequency to the extent that flashing light is visually recognized as lit continuously for to have despite the observer blinks The observation object is provided with a plurality of light sources, each of the plurality of light sources blinks at a different specific blinking frequency, and the signal processing unit includes a drive signal of the light source and an electroencephalogram signal. Cross-spectral analysis is performed to calculate the correlation in the frequency domain, and the visual evoked potential signal is detected based on the analysis result, and the light source watched by the observer corresponding to the detected visual evoked potential signal is determined. It characterized the door.

  Such a visual evoked potential signal detection system according to the present invention blinks the light source at a blinking frequency high enough to be visually recognized by the observer as though the blinking light is blinking. Therefore, for the observer, the blinking light is visually recognized and fused, and based on the Talbot-Plateau rule, the blinking light is less glare than the case where the blinking frequency belongs to the low frequency band. Flickering can be suppressed and the burden on the observer can be greatly reduced.

  In addition, the signal processing unit in the visual evoked potential signal detection system according to the present invention compensates for the deterioration of the SNR of the visual evoked potential signal due to the increased blink frequency, and the frequency of the blinking drive signal of the light source and the electroencephalogram signal. Cross spectrum analysis for calculating the correlation on the region is performed, and the visual evoked potential signal is detected based on the analysis result.

  As a result, the visual evoked potential signal detection system according to the present invention can detect the visual evoked potential signal with extremely high accuracy even in a case where the analysis accuracy has deteriorated significantly in normal power spectrum analysis such as Fourier transform. Can do.

  In the present invention, a visual evoked potential signal detection system capable of significantly reducing the burden on the observer by suppressing the glare and flickering feeling of flashing light and exhibiting excellent practicality that is extremely suitable for long-time use Can be provided.

It is a figure which shows the structure of the visual evoked potential signal detection system shown as embodiment of this invention. It is a figure which shows the relationship between the blink frequency of a light source, and the power spectrum obtained by analyzing an electroencephalogram signal when a light source is observed. It is a figure for demonstrating the processing content of the signal processing part in the visual evoked potential signal detection system shown as embodiment of this invention. Based on the analysis results obtained by applying the cross spectrum analysis performed by the visual evoked potential signal detection system shown as the embodiment of the present invention to the acquired electroencephalogram signal and the power spectrum analysis similar to the conventional one, respectively. It is a figure which shows the experimental result which determined the light source which the test subject observed. It is a figure shown about the application example to which the visual evoked potential signal detection system shown as embodiment of this invention is applied. It is a figure shown about the general system for measuring a visual evoked potential signal. It is a figure which shows the relationship between the blink drive signal of a light source, and an electroencephalogram signal when the light source is observed.

  Hereinafter, specific embodiments to which the present invention is applied will be described in detail with reference to the drawings.

  This embodiment is a visual evoked potential signal detection system that detects a signal of a visual evoked potential evoked in the brain by a visual stimulus. In particular, this visual evoked potential signal detection system is intended for observation of blinking light that blinks at a blinking frequency in a high-frequency band, where analysis accuracy has been greatly deteriorated in the past.

As shown in FIG. 1, the visual evoked potential signal detection system includes an observation object 10 in which a plurality of light sources 11 _CHn are disposed, and a signal processing unit 20 that processes an electroencephalogram signal of an observer OB that visually _recognizes the light sources 11 _CHn. With.

The observation object 10 is, for example, a board-like object visually recognized by the observer OB, and a plurality of light sources 11 _CHn (n is the number of channels) are arranged on the surface thereof. Here, six light sources 11 _CH1 , 11 _CH2 , 11 _CH3 , 11 _CH4 , 11 _CH5 , and 11 _CH6 are shown. Each of these light sources 11 _CHn is composed of, for example, an LED (Light Emitting Diode).

Such an observation object 10 supplies a drive signal to the light source 11 _CHn for each channel via a predetermined light source control means, and causes each light source 11 _CHn to blink at a unique blinking frequency different from each other. Specifically, the observed object 10 can change the blinking frequency of the light sources 11 _CHn for each channel, for example, such as to vary in 0.5Hz increments, optionally at equal frequency intervals or unequal frequency intervals for each light source 11 _CHn To flash. In addition, the observation object 10 causes each light source 11 _CHn to be lit at a high blinking frequency so that it can be visually recognized in a fused manner so that the observer OB lights continuously even though the blinking light is blinking. Set and control to blink. A specific example of the blinking frequency that is high enough to be visually recognized by the fusion of the blinking light will be described in detail later.

  The signal processing unit 20 includes an amplification unit 21 that amplifies the electroencephalogram signal, a signal accumulation unit 22 that accumulates the electroencephalogram signal that has been amplified and A / D converted by the amplification unit 21, and the signal accumulation unit 22 A cross spectrum analysis unit 23 that performs cross spectrum analysis on an electroencephalogram signal, and a signal detection unit 24 that detects a visual evoked potential signal based on an analysis result by the cross spectrum analysis unit 23.

  Of these units, the amplification unit 21 can be configured as a part of a digital electroencephalograph. In addition, the cross spectrum analysis unit 23 and the signal detection unit 24 may be implemented as a program that can be executed using hardware such as a CPU (Central Processing Unit) or a memory in a computer, or may be mounted on an expansion board that can be attached to the computer. It can be mounted using a dedicated processor such as a mounted DSP (Digital Processing Unit). Further, the signal storage unit 22 can be configured using a hard disk or other various storage media that can be built in or externally attached to an apparatus such as a computer that implements the functions of the cross spectrum analysis unit 23 and the signal detection unit 24.

  The amplifying unit 21 inputs the electroencephalogram signal acquired via the electrode 30 installed on the scalp of the observer OB in time series, and amplifies it with a predetermined gain. The analog electroencephalogram signal amplified by the amplifying unit 20 is converted into a digital signal by an A / D converter (not shown) and written in the signal storage unit 22. The electrode 30 is an electroencephalogram acquisition means installed on the scalp to acquire an observer OB electroencephalogram signal. The electrode 30 is composed of, for example, a silver-silver chloride electrode having a diameter of about 1 cm. Usually, a plurality of electrodes 30 are arranged in a predetermined layout on the scalp of the observer OB, particularly on the occipital scalp where the cortical visual cortex is present. Installed.

  The signal storage unit 22 stores the electroencephalogram signal supplied from the amplification unit 21 as time-series digital data. The electroencephalogram signal accumulated in the signal accumulation unit 22 is read out by the cross spectrum analysis unit 23.

The cross spectrum analysis unit 23 reads the electroencephalogram signal as digital data stored in the signal storage unit 22 for a predetermined time at a predetermined sampling period. Note that the sampling period is desirably less than 20 milliseconds in order to ensure real-time processing and to prevent aliasing. Then, the cross spectrum analysis unit 23 calculates the correlation in the frequency domain with the electroencephalogram signal read from the signal storage unit 22, that is, the cross spectrum for each frequency, using the drive signal of each light source 11 _CHn as a reference signal. At this time, the cross spectrum analysis unit 23 calculates in parallel the correlation in the frequency domain between each of the drive signals of the plurality of light sources 11 _CHn and the electroencephalogram signal read from the signal storage unit 22. The cross spectrum analysis unit 23 supplies the calculated analysis result data for each drive signal of each light source 11 _CHn to the signal detection unit 24. The processing in the cross spectrum analysis unit 23 will be described in detail later.

  The signal detection unit 24 stores the analysis result data supplied from the cross spectrum analysis unit 23 in a memory or the like (not shown), and detects a visual evoked potential signal included in the electroencephalogram signal based on the analysis result data. At this time, the signal detection unit 24 detects the visual evoked potential signal according to a predetermined rule using a threshold or the like. The processing in the signal detection unit 24 is performed seamlessly with the processing in the cross spectrum analysis unit 23, and details thereof will be described later together with the processing in the cross spectrum analysis unit 23.

The signal processing unit 20 having such units detects a specific visual evoked potential signal included in the electroencephalogram acquired through the electrode 30. In the visual evoked potential signal detection system, it becomes possible to identify which light source 11 _CHn the blinking light of the observer OB was gazing at based on the detection result by the signal processing unit 20. Information indicating the detection result by the signal processing unit 20 is supplied to a processing unit (not shown) that performs predetermined subsequent processing or an external device. Examples of the external device include a display device that displays various types of information including detection results, a rehabilitation device, and a brain computer interface described later.

In the visual evoked potential signal detection system including the observation object 10 and the signal processing unit 20 as described above, as described above, each light source 11 _CHn is flashed at a high flashing frequency such that the flashing light is visually _recognized. Blink. This is because the so-called Talbot-Plateau rule is used in which the brightness when the blinking light is visually recognized is an average value of luminance.

Here, if the blinking frequency is equal to or higher than the frequency equal to the frame rate of an existing movie or television, the blinking light is visually recognized. That is, in the visual evoked potential signal detection system, each of the light sources 11 _{CHn is set} to a blinking frequency of 24 Hz or more equal to the frame rate of the movie, more preferably a PAL (Phase Alternating Line) method or a SECAM (Sequentiel).
The flashing frequency is 25 Hz or more equal to the frame rate of a couleur a memoire) television, and more preferably, the blinking frequency is 30 Hz or more equal to the frame rate of a NTSC (National Television System Committee) television.

Further, in the visual evoked potential signal detection system, each light source 11 _CHn is set at a flashing frequency equal to or higher than a so-called critical fusion frequency (CFF), which is a critical frequency for whether or not a human can recognize flashing of flashing light. It is also preferable to make it blink. According to various literatures, this critical fusion frequency is said to be about 37 Hz, but varies depending on the brightness of the light source, the color, the state of the observer, and the like. Actually, when a critical fusion frequency was measured before the experiment for a subject in an experiment described later conducted by the present inventor, a result of 34.3 ± 1.9 Hz was obtained.

  As shown in FIG. 2, in the field related to visual evoked potentials, the frequency equal to the frame rate of movies and televisions and the critical fusion frequency have hitherto been poor in SNR of the visual evoked potential signal component included in the electroencephalogram signal. It is too difficult to analyze, or belongs to a high frequency band where it is said that the visual evoked potential itself is not generated.

In the visual evoked potential signal detection system, the blinking light is fused for the observer OB by blinking each light source 11 _CHn at a blinking frequency belonging to such a high frequency band and at a unique blinking frequency different for each channel. Therefore, based on the Talbot-Plateau rule, it is observed by the observer OB so as to emit light with a luminance lower than the actual blinking intensity of each light source 11 _CHn .

Therefore, in the visual evoked potential signal detection system, the observer OB does not feel glare or flicker of the flashing light while not reducing the flashing intensity of the light source 11 _CHn itself, which causes the deterioration of the SNR of the visual evoked potential signal. And can exhibit excellent practicality without being painful even when used for a long time. Further, in the visual evoked potential signal detection system, the light sources 11 _CHn are caused to blink at a high blinking frequency so that they do not overlap with the α wave band.

Here, in the visual evoked potential signal detection system, although the blinking intensity of the light source 11 _CHn is not reduced, the blinking frequency is increased, so that the SNR of the visual evoked potential signal deteriorates and analysis is difficult. There is a case.

Therefore, in the visual evoked potential signal detection system, in order to compensate for the deterioration of the SNR, as described above, the signal processing unit 20 correlates the drive signal of each light source 11 _CHn and the electroencephalogram signal in the frequency domain. Perform cross-spectral analysis to calculate

  The application of this cross spectrum analysis is based on the knowledge obtained from the following experiment conducted by the present inventor as a preliminary experiment prior to the present application.

  That is, the inventor of the present application conducted an experiment to compare the effectiveness of analysis processing applied to the acquired electroencephalogram signal. The experiment was performed by presenting 6 light sources flashing at a flashing frequency of 7 Hz to 12 Hz to 6 subjects, and recording the electroencephalogram when the designated light source was watched for 5 seconds. At this time, the luminance of the light source was set to two states, a strong state and a weak state. EEG measurements were performed 6 times for one light source, and a total of 72 EEG measurements were performed.

  And when applying the same power spectrum analysis to the EEG signal acquired in this way, and applying the cross spectrum analysis to calculate the correlation in the frequency domain between the EEG signal and the light source drive signal. The case was compared. Specifically, based on each analysis result, the light source that the subject gazes at was determined, the correct answer rate was obtained based on the determination result, and the difference in the correct answer rate by each analysis method was compared.

  As a result, when applying the same power spectrum analysis as before, there was a significant difference in the correct answer rate due to the difference in the brightness of the light source, whereas when applying the cross spectrum analysis, There was no difference in the correct answer rate due to the difference in brightness. This result suggests that by applying the cross spectrum analysis, it is possible to ensure a high correct answer rate even if the brightness of the light source is suppressed, and as an analysis method for a signal having a poor SNR, the cross spectrum analysis is performed. This suggests that the analysis is effective.

  Here, this preliminary experiment is directed to a light source that blinks at a blinking frequency in a low frequency band of 7 Hz to 12 Hz. It was considered that cross spectrum analysis was effective even for signals. Therefore, in the visual evoked potential signal detection system, the signal processing unit 20 performs processing using the following cross spectrum analysis.

Specifically, in the visual evoked potential signal detection system, information about which light source is blinking at which blinking frequency among the plurality of light sources 11 _CNn is known in advance. Therefore, as shown in FIG. 3, the cross spectrum analysis unit 23 uses the drive signal of each light source 11 _CNn as the reference signal y _k (t), and the electroencephalogram signal x for each frequency with respect to the reference signal y _k (t). A correlation on the frequency domain indicating how much _k (t) matches, that is, a cross spectrum X _k (f) · Y _k ^* (f) is calculated. At this time, the cross spectrum analyzer 23 calculates the cross spectrum X _k (f) · Y _k ^* (f) for each of the drive signals y _k (t) of the plurality of light sources 11 _CHn in parallel.

X _k (f) is an auto spectrum of the electroencephalogram signal x _k (t) at the frequency f, and Y _k (f) is an auto spectrum of the reference signal x _k (t) at the frequency f. Y _k ^* (f) is a conjugate complex number of Y _k (f), k is an analysis window number, and is an inner product. Then, X _k (f) and Y _k (f) respectively represent the amplitude A _kx of the electroencephalogram signal and the phases θ _kx and θ _{kx of the} electroencephalogram signal and the reference signal, respectively, as shown in the following equation (1). Is represented by

When reading the electroencephalogram signal from the signal storage unit 22, the cross spectrum analysis unit 23 calculates a cross spectrum for each drive signal of all the light sources 11 _CHn with an analysis window length of a predetermined time such as 1 second, for example. Then, the cross spectrum analysis unit 23 shifts the analysis window by a slight predetermined time so as to overlap the current analysis window by 90%, for example, and similarly drives each light source 11 _CHn. A cross spectrum is calculated for the signal. The cross spectrum analysis unit 23 repeatedly performs such processing, generates a spectrum distribution composed of a plurality of cross spectra for each drive signal (reference signal) of each light source 11 _CHn , and stores this in the memory of the signal detection unit 24. Store.

Here, as can be seen from the above equation (1), by obtaining the cross spectrum, the time difference between the electroencephalogram signal and the reference signal, that is, the phase difference θ _kx −θ _ky in the frequency domain, and the amplitude A of the electroencephalogram signal _kx is obtained. The right side of FIG. 3 shows how the cross spectrum is expressed as a vector on the complex plane. The phase of this vector indicates how much the electroencephalogram signal is delayed with respect to the flashing light from the light source 11 _{CHn. The} longer the vector length, the larger the inner product of the amplitudes of the electroencephalogram signal and the reference signal. It is big.

Therefore, if the correlation between the electroencephalogram signal and the reference signal is always high, vectors having substantially the same length in substantially the same phase direction are obtained for a plurality of cross spectra obtained by shifting the analysis window. That is, since a specific visual evoked potential signal included in a certain electroencephalogram signal and the corresponding driving signal of the light source 11 _CHn are synchronized, the phase difference in the latency is substantially constant and represents a cross spectrum. Even if the analysis window is shifted, the vector is added and expanded in a substantially constant phase direction. On the other hand, since other components included in a certain electroencephalogram signal are not synchronized with the drive signal of the light source 11 _CHn , the phase difference becomes random, and the cross spectrum cancels out in each analysis window. Therefore, the addition result of the vector representing the cross spectrum is larger in the specific visual evoked potential signal synchronized with the driving signal of the light source 11 _CHn , and the observer OB is watching by detecting this. The light source 11 _CHn can be specified.

Therefore, as shown in the following equation (2), the signal detection unit 24 adds a vector representing a cross spectrum for each drive signal (reference signal) of each light source 11 _CHn , and calculates the phase and amplitude of the vector after the addition. calculate. Then, the signal detection unit 24 determines that a specific visual evoked potential signal is included in the electroencephalogram signal when the phase of the vector after addition is within the predetermined phase range and the vector length reaches the predetermined length. Then, it is determined that the light source 11 _CHn corresponding to the reference signal used for _obtaining the vector is the light source that the observer OB is gazing at.

  Specifically, the signal detection unit 24 sets a threshold in advance for the phase of the vector after the addition, as indicated by a broken line in the complex plane shown on the right side of FIG. 3, and the phase defined by this threshold Only the added vectors that occur within the range are processed. This is because the length from the human eye to the visual cortex and the transmission speed of the nerve are almost the same, although there are some individual differences depending on the human, so the phase of the vector, that is, the reference signal and the electroencephalogram signal This is because it is possible to grasp how much the phase difference from the specific visual evoked potential signal included is. In FIG. 3, a threshold is provided for a predetermined phase in the first quadrant, but this phase range changes according to the frequency of the reference signal. That is, the signal detection unit 24 calculates the average phase difference and standard deviation at the frequency based on, for example, a phase difference histogram obtained experimentally in advance according to the frequency of the reference signal, and sets the threshold based on this. It should be set. Thereby, the signal detection part 24 can filter appropriately the signal component larger than the case where the phase difference of a noise component is equal, or a visual evoked potential signal.

  Then, the signal detection unit 24, when the length of the vector after addition reaches the radius of the circle in the complex plane shown on the right side of FIG. 3, that is, the sector-shaped radius defined by the threshold value of the vector phase It is determined that there is a specific visual evoked potential signal. That is, the signal detection unit 24 sets a second threshold for the length of the vector after the addition, and continues the process until the length of the vector after the addition reaches this threshold.

  By performing such threshold processing, the signal detection unit 24 increases the length of a vector representing each noise component, for example, when a large amount of noise components are mixed not from a specific brain wave but from various directions. Since the phase direction of the vector is different for each analysis window, the vector length after the addition is not increased, and the possibility of erroneous determination that there is a specific visual evoked potential signal can be reduced. On the other hand, even if the amplitude of a specific electroencephalogram signal is small and the vector length thereof is short, the signal detection unit 24 can perform the addition after the addition if the vector phase direction for each analysis window matches. Although it takes time until the length of the vector reaches the second threshold, it can be reliably determined that there is a specific visual evoked potential signal.

  As described above, in the visual evoked potential signal detection system, the visual evoked potential signal is detected by applying the cross spectrum analysis, so that the analysis accuracy of the normal power spectrum analysis such as Fourier transform is markedly deteriorated. However, since the correlation of the specific visual evoked potential signal is high and the correlation of the noise component is calculated low, the specific visual evoked potential signal can be detected with extremely high accuracy. In the visual evoked potential signal detection system, since the visual evoked signal is detected based on the correlation in a predetermined frequency band, it can be detected by sequentially processing the acquired electroencephalogram signal, and pattern analysis is performed as in the past. Therefore, it is not necessary to pre-process a large amount of data, and real-time processing can be ensured.

  The present inventor conducted the following experiment in order to verify the effectiveness of the present invention. In this experiment, as shown in FIG. 1, as the light source, six LEDs arranged in 2 rows × 3 columns at a 3.7 cm interval on a predetermined board surface were used. The LED is a red LED (TLRE180AP) manufactured by Toshiba. The blinking frequency of each light source was 37 Hz to 42 Hz, and was set in increments of 1 Hz for the six LEDs. Then, the brain wave was recorded when the subject faced a position 1 m away from the board and the designated light source was observed. There were 17 subjects in total, including 9 healthy subjects (21-24 years old, 7 men, 2 women), 8 ALS patients (47-84 years old, 4 men, 4 women). Name). In these subjects, the action of gazing at one LED for 10 seconds was sequentially performed from the first LED to the sixth LED, and this was regarded as one task, and an experiment of 10 tasks per subject was conducted. Acquired. The electroencephalogram was collected from the occipital region of the subject using electrodes arranged according to the International 10-20 method.

  Then, based on the analysis results obtained by applying the power spectrum analysis similar to the conventional one and the above-described cross spectrum analysis to the electroencephalogram signal thus obtained, the light source that the subject gazes at is determined. The correct answer rate was obtained based on the determination result.

  As a result, as shown in FIG. 4, a higher accuracy rate is obtained when the cross spectrum analysis of the present invention is applied than when the conventional analysis is applied, and an electroencephalogram signal of about 5 seconds or more is processed. A correct answer rate of 75% or more was obtained. The correct answer rate is a numerical value obtained by dividing the correct answer rate of all subjects by the number of people. In addition, there was no difference in the correct answer rate between healthy subjects and ALS patients.

As described above, in the visual evoked potential signal detection system, even if each light source 11 _CHn is blinked at a high blinking frequency such that the blinking lights are fused and visually recognized, the visual evoked potential signal is very accurately detected. Can be detected.

As described above, in the visual evoked potential signal detection system shown as the embodiment of the present invention, it is visually recognized as being continuously lit by the observer OB even though the blinking light is blinking. By flashing each light source 11 _CHn at a high flashing frequency, it is possible to prevent the observer OB from feeling dazzling or flickering of the flashing light, and it is excellent without causing pain for long-time use. Practicality can be demonstrated. Further, in the visual evoked potential signal detection system, in order to compensate for the deterioration of the SNR of the signal caused by blinking each light source 11 _CHn at a high blinking frequency, on the frequency domain between the drive signal and the electroencephalogram signal of each light source 11 _CHn. By performing the cross spectrum analysis for calculating the correlation at, the visual evoked potential signal can be detected with extremely high accuracy.

  Such a visual evoked potential signal detection system is extremely suitable when applied to a technology that mechanically compensates for motor functions, such as a brain computer interface, and realizes various devices that perform visual device control commands. Can do.

  For example, as shown in FIG. 5, a control in which the observation object 10 and the signal processing unit 20 in the visual evoked potential signal detection system described above are made into one box in the vicinity of the electric bed 51 on which the observer OB as an operator sits. The device 50 is disposed. At this time, the light source provided in the control device 50 includes, for example, angle adjustment of the electric bed 51, opening / closing of the blind 52, power on / off of the air conditioner 53, temperature setting, and power on / off of the television 54, respectively. The control contents of various devices to be controlled by the control device 50, such as channel switching, keyboard operation of the personal computer 55, key operation of the telephone 56 and calling / incoming calls, are assigned in advance. The control device 50 controls various devices by transmitting a control signal indicating the control contents assigned to the light source to the various devices in response to the observer OB directing his / her line of sight to a desired light source. can do.

  The visual evoked potential signal detection system is extremely suitable when applied to such a brain computer interface, not only supporting the life of a physically handicapped person but also controlling the desired equipment by a person with a full hand. You can expect applications such as soaking. The visual evoked potential signal detection system can control a wide variety of devices by increasing the number of light sources, that is, the number of channels, and can also be applied to a virtual keyboard or the like. Furthermore, the visual evoked potential signal detection system may assign a plurality of control contents to one light source. For example, in the visual evoked potential signal detection system, the device is turned on / off or set according to the time when the predetermined light source is watched, or the device is set according to the combination of the light sources to be watched. It is also possible to do so.

  Thus, the visual evoked potential signal detection system is expected to be widely applied in various fields. In addition, this visual evoked potential signal detection system is superior in that the observer does not have to worry about causing a photosensitivity attack because it uses a light source with a blinking frequency that cannot be confirmed by the human naked eye.

  The present invention is not limited to the embodiment described above.

  For example, in the above-described embodiment, the observation object on which the light source is arranged is described as a board. However, the present invention can be applied to any observation object as long as the observer can visually recognize each light source. Can be of any shape. Similarly, the present invention is not limited to the arrangement of light sources.

  In the above-described embodiment, the electroencephalogram signal is acquired using the electrode. However, the electrode is not necessarily used as long as the electroencephalogram signal can be acquired. However, considering an application form such as a brain computer interface in the present invention, it is desirable to perform electroencephalogram measurement non-invasively, and therefore it is preferable to acquire an electroencephalogram signal using an electrode. In this case, it is possible to easily acquire an electroencephalogram signal by preparing a headset in which electrodes are arranged and attaching the headset to the observer's head.

  Further, in the above-described embodiment, the contents of the analysis window used at the time of analyzing the electroencephalogram signal are not particularly mentioned, but the present invention can change the analysis window length according to the frequency of the signal, for example, Arbitrary analysis windows can be applied, such as processing only a part of the signals contained in the window and making the data in the remaining analysis windows void values, depending on the analysis window used and the analysis window length. It can be expected to improve the analysis accuracy.

  Thus, it goes without saying that the present invention can be modified as appropriate without departing from the spirit of the present invention.

10 observed object _{11 CH1, 11 CH2, 11 CH3} , 11 CH4, 11 CH5, 11 CH6, 11 CHn light source 20 signal processing section 21 amplifying section 22 signal storage section 23 cross spectral analyzer 24 signal detection unit 30 electrode 50 controller 51 Electric bed 52 Blind 53 Air conditioner 54 Television 55 Personal computer 56 Telephone OB Observer

Claims (5)

A visual evoked potential signal detection system for detecting a specific visual evoked potential signal contained in an electroencephalogram,
An observation object provided with a light source, and a signal processing unit for processing an electroencephalogram signal of an observer who visually recognizes the light source,
The observation object blinks the light source at a blinking frequency such that the observation object is visually recognized as being continuously lit even though blinking light is blinking ,
Wherein the signal processing unit, visual, characterized in that perform cross spectrum analysis to calculate a correlation in the frequency domain on the drive signal and the electroencephalogram signal of the light source, for detecting the visual evoked potential signals based on the analysis results Evoked potential signal detection system. A visual evoked potential signal detection system for detecting a specific visual evoked potential signal contained in an electroencephalogram,
An observation object provided with a light source, and a signal processing unit for processing an electroencephalogram signal of an observer who visually recognizes the light source,
The observation object blinks the light source at a blinking frequency such that the observation object is visually recognized as being continuously lit even though blinking light is blinking ,
The signal processing unit overlaps the analysis window and shifts by a predetermined time while performing a cross spectrum analysis of the driving signal of the light source and the electroencephalogram signal, and adds a vector on a complex plane representing the cross spectrum. based on the phase and amplitude of the vector after addition to be a visual evoked potential signals detected system that comprises detecting a visual evoked potential signals. The signal processing unit sets a threshold value for at least a phase range of vectors after addition, and sets only a vector after addition that occurs within a phase range defined by the threshold value as a processing target. visual evoked potential signal detection system 2 according. A visual evoked potential signal detection system for detecting a specific visual evoked potential signal contained in an electroencephalogram,
An observation object provided with a light source, and a signal processing unit for processing an electroencephalogram signal of an observer who visually recognizes the light source,
The observation object blinks the light source at a blinking frequency such that the observation object is visually recognized as being continuously lit even though blinking light is blinking ,
The observation object is provided with a plurality of light sources,
Each of the plurality of light sources blinks at a unique blinking frequency different from each other,
The signal processing unit performs a cross spectrum analysis for calculating a correlation in the frequency domain between the driving signal of the light source and the electroencephalogram signal, detects a visual evoked potential signal based on the analysis result, and detects the detected visual evoked potential visual evoked potential signal detection system the observer corresponding to the signal you and judging gaze light sources. The visual evoked potential signal detection system according to any one of claims 1 to 4 , wherein a blinking frequency of the light source is equal to or higher than a critical fusion frequency.