JP2015054221A - Frequency detection device, frequency detection method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a brain-wave frequency detection device capable of highly accurately acquiring an output of BCI (Brain Computer Interface) using SSVEP (Steady-State Visual Evoked Potential) without any need of learning.SOLUTION: The brain-wave frequency detection device includes: a brain wave measurement part that measures a brain wave emitted from a primary visual cortex; a cycle data segmentation part that segments, for each of a plurality of prescribed cycles, cycle data from the brain wave, with each time of the prescribed cycle defined as a start time; an arithmetic average calculation part that adds the cycle data for each prescribed cycle; and an additive-method frequency detection part that detects, in the prescribed cycle, a frequency corresponding to a cycle where a difference between positive/negative peak intensities of data obtained by adding the cycle data for each prescribed cycle is maximized.

Description

  The present invention relates to an electroencephalogram frequency detection apparatus, a frequency detection method, and a program.

  BCI (Brain Computer Interface) is attracting attention in life support for people with extremely limited motor functions. BCI is a technique for analyzing a signal generated by brain activity and using the analysis result as an alternative means for user operation on a computer, machine, or the like. BCI makes it possible to use various computer functions such as typing and web search, and to operate various devices such as wheelchairs, prosthetic hands, and home appliances, even for people who are unable or extremely difficult to move or speak their limbs. Can improve the quality of life.

  One of the BCIs is a technique using SSVEP (Steady-State Visual Evoked Potential). SSVEP is a phenomenon in which a steady visual evoked potential (VEP) having a frequency similar to that of a visual stimulus blinks is generated with respect to a visual stimulus that periodically blinks. Non-Patent Documents 1 and 2 each show a BCI method using SSVEP.

Vialatte F. B. and three others, “Steady-state visually evoked potentials: Focus on essential paradigm and future perspectives”, Progress in Neurobiology 90, 2010, p. 418-438 Itai A. and 1 other, “Spectrum Based Feature Extraction Using Spectrum Intensity Ratio for SSVEP Detection”, 34th Annual International Conference of the IEEE EMBS, 2012, p. 5947-5950

The BCI method using SSVEP described in Non-Patent Document 1 requires learning. For this reason, the BCI method using SSVEP described in Non-Patent Document 1 cannot be used immediately.
Further, the BCI method using SSVEP described in Non-Patent Document 2 has room for improvement in output accuracy.

  The present invention has been made in view of such circumstances, and the purpose thereof is an electroencephalogram frequency detection apparatus capable of obtaining BCI output using SSVEP more accurately without learning. To provide a frequency detection method and program.

  The present invention has been made to solve the above-described problems. An electroencephalogram frequency detection apparatus according to an aspect of the present invention includes an electroencephalogram measurement unit that measures an electroencephalogram emitted from a primary visual cortex, and a plurality of predetermined cycles. For each of the above, from the brain wave, a cycle data cutting unit that cuts out cycle data starting at each time of the predetermined cycle, an addition average calculation unit that adds the cycle data for each predetermined cycle, and the predetermined cycle And an addition method frequency detection unit for detecting a frequency corresponding to a cycle in which the difference between the positive and negative peak intensities of the data obtained by adding the cycle data for each predetermined cycle is maximum.

  An electroencephalogram frequency detection apparatus according to another aspect of the present invention includes an electroencephalogram measurement unit that measures an electroencephalogram emitted from a primary visual cortex, a frequency spectrum acquisition unit that acquires the frequency spectrum of the electroencephalogram, and a plurality of predetermined For each of the frequencies, a maximum value acquisition unit that acquires the maximum value of the intensity of the frequency spectrum of the electroencephalogram in a first predetermined frequency range including the predetermined frequency, and each of the plurality of predetermined frequencies includes the predetermined frequency A median value acquisition unit for acquiring a median value of the intensity of the frequency spectrum of the electroencephalogram in a second predetermined frequency range; and a frequency having a maximum difference between the maximum value and the median value among the plurality of predetermined frequencies. And a spectral method frequency detection unit for detecting.

  An electroencephalogram frequency detection apparatus according to another aspect of the present invention is the electroencephalogram frequency detection apparatus described above, wherein the frequency detected by the spectral method frequency detection unit determines whether or not a predetermined condition is satisfied. When the determination unit and the condition determination unit determine that the frequency does not satisfy the predetermined condition, for each of a plurality of predetermined periods, period data having each time of the predetermined period as a start time from the brain wave And a frequency data extraction unit that detects a frequency corresponding to a period in which the difference between the positive and negative peak intensities of the data obtained by adding the period data for each predetermined period is maximum. And a portion.

  An electroencephalogram frequency detection apparatus according to another aspect of the present invention is the electroencephalogram frequency detection apparatus described above, wherein the electroencephalogram measurement unit detects a plurality of electroencephalograms emitted from the primary visual cortex at different positions. The brain waves at a plurality of positions are measured based on signals from the electrodes.

  A frequency detection method according to another aspect of the present invention is a frequency detection method of an electroencephalogram frequency detection device, comprising an electroencephalogram measurement step for measuring an electroencephalogram emitted from a primary visual cortex, and each of a plurality of predetermined periods. A period data extraction step of extracting period data starting from each time of the predetermined period from the brain wave, an addition average calculation step of adding the period data for each predetermined period, and the predetermined period And an addition frequency detection step of detecting a frequency corresponding to a period in which a difference between positive and negative peak intensities of data obtained by adding the period data for each predetermined period is maximum.

  According to another aspect of the present invention, there is provided a program for controlling an electroencephalogram frequency detecting apparatus, an electroencephalogram measurement step for measuring an electroencephalogram emitted from a primary visual cortex, and the electroencephalogram for each of a plurality of predetermined periods. A cycle data cutting step for cutting out cycle data starting at each time of the predetermined cycle, an addition average calculating step for adding the cycle data for each predetermined cycle, and the predetermined cycle among the predetermined cycles This is a program for executing an addition frequency detection step for detecting a frequency corresponding to a period in which the difference between the positive and negative peak intensities of the data obtained by adding the period data is maximized.

  According to the present invention, BCI output using SSVEP can be obtained more accurately without learning.

It is a schematic block diagram which shows the function structure of the frequency detection apparatus in the 1st Embodiment of this invention. It is explanatory drawing which shows arrangement | positioning of the electrode in the same embodiment. It is explanatory drawing which shows the example of the electroencephalogram detection signal which the electroencephalogram measurement part in the embodiment acquires. It is explanatory drawing which shows the example of the frequency spectrum which the frequency spectrum acquisition part in the embodiment acquires. In the same embodiment, it is a flowchart which shows the example of the process sequence in which the frequency detection apparatus detects the blink frequency of the visual stimulus which the subject is gazing at. It is explanatory drawing which shows the light emission pattern of LED in the verification experiment of the effectiveness of the embodiment. It is explanatory drawing which shows the example of the identification surface in LDA. It is explanatory drawing which shows the example of LDA with respect to the combination of the power value of 6 Hz, and the power value of 7 Hz. It is explanatory drawing which shows the example of LDA with respect to the combination of the power value of 7 Hz, and the power value of 8 Hz. It is explanatory drawing which shows the example of LDA with respect to the combination of the power value of 6 Hz, and the power value of 8 Hz. It is a flowchart which shows the process sequence which acquires an identification surface by learning in LDA method. It is a flowchart which shows the process sequence which determines blink frequency in LDA method. It is explanatory drawing which shows the example of selection of the data for identification in the 4-fold intersection confirmation, and the data for learning. It is explanatory drawing which shows the experimental result about the same embodiment. It is a schematic block diagram which shows the function structure of the frequency detection apparatus in the 2nd Embodiment of this invention. Description showing an example of the addition average of data extracted by the periodic data cutout unit in a time window corresponding to 6 Hz (a time window provided with a 1/6 second period) calculated by the addition average calculation unit in the embodiment FIG. It is explanatory drawing which shows the example of the addition average of the data which the period data extraction part calculated by the addition average calculation part in the embodiment cut out by the time window corresponding to 7 Hz. It is explanatory drawing which shows the example of the addition average of the data which the period average data extraction part computed by the addition average calculation part in the same embodiment cut out with the time window corresponding to 8 Hz. In the same embodiment, it is a flowchart which shows the example of the process sequence in which the frequency detection apparatus detects the blink frequency of the visual stimulus which the subject is gazing at. It is explanatory drawing which shows the experimental result about the same embodiment. It is a schematic block diagram which shows the function structure of the frequency detection apparatus in the 3rd Embodiment of this invention. In the same embodiment, it is a flowchart which shows the example of the process sequence in which the frequency detection apparatus detects the blink frequency of the visual stimulus which the subject is gazing at. It is explanatory drawing which shows the experimental result about the same embodiment.

  Hereinafter, the present invention will be described through embodiments of the invention, but the following embodiments do not limit the invention according to the claims. In addition, not all the combinations of features described in the embodiments are essential for the solving means of the invention.

<First Embodiment>
FIG. 1 is a schematic block diagram showing a functional configuration of a frequency detection device according to the first embodiment of the present invention. In FIG. 1, the frequency detection device 1 includes a frequency detection device main body 100 and an electrode 900. The frequency detection device main body 100 includes a signal input unit 110, a control unit 120, a storage unit 150, and a result output unit 160. The control unit 120 includes an electroencephalogram measurement unit 121, a frequency spectrum acquisition unit 122, a maximum value acquisition unit 123, a median value acquisition unit 124, and a spectrum method frequency detection unit 125.

The frequency detection device 1 analyzes the frequency spectrum of the brain wave of a person (hereinafter referred to as “subject”) who is gazing at the periodically blinking visual stimulus, and detects the blinking frequency of the visual stimulus. More specifically, the frequency detection device 1 determines the frequency of the SSVEP (steady visual evoked potential) by analyzing the frequency spectrum of the electroencephalogram, and selects a frequency corresponding to the frequency of the SSVEP among a plurality of predetermined frequencies. .
The frequency detection apparatus 1 corresponds to an example of an electroencephalogram frequency detection apparatus.

When the subject is gazing at any of the visual stimuli flashing at different frequencies, the frequency detection device 1 can function as BCI by analyzing the brain wave of the subject.
For example, an LED (Light Emitting Diode) that blinks at 6 Hz (Hertz), an LED that blinks at 7 Hz, and an LED that blinks at 8 Hz are placed on a breadboard, and the target person is either Look closely at the LED. The frequency detection device 1 analyzes the subject's brain wave to determine the frequency of the SSVEP, and selects the frequency closest to the determined frequency of the SSVEP among the frequencies of the LEDs. And the frequency detection apparatus 1 outputs the signal which shows the selected frequency to the apparatus of operation object.

For example, when the frequency detection device 1 functions as a BCI for computer operation, the computer performs processing according to a signal from the frequency detection device 1. Further, for example, when the frequency detection device 1 selects 6 Hz, the computer activates a typing application. Moreover, when the frequency detection apparatus 1 selects 7 Hz, the computer starts a web browser. When the frequency detection device 1 selects 8 Hz, the computer displays a menu screen.
In this way, the subject can operate the computer by selecting which LED to watch.

Alternatively, when the frequency detection device 1 functions as a BCI for wheelchair operation, the wheelchair control device operates the wheelchair according to a signal from the frequency detection device 1. For example, when the frequency detection device 1 selects 6 Hz, the wheelchair moves forward. When the frequency detection device 1 selects 7 Hz, the wheelchair turns to the right. Moreover, when the frequency detection apparatus 1 selects 8 Hz, the wheelchair turns to the left.
Thus, the subject can operate the wheelchair by selecting which LED to watch.

Alternatively, when the frequency detection device 1 functions as a BCI for adjusting the magnitude of the output of the device in stages, the magnitude of the output of the device is changed according to the signal from the frequency detection device 1. For example, when the frequency detection device 1 selects 6 Hz, the output is reduced. Further, when the frequency detection device 1 selects 7 Hz, the output is set to a medium level. Further, when the frequency detection device 1 selects 8 Hz, the output is increased.
Thus, the subject can adjust the magnitude of the output of the device by selecting which LED to watch.

In the following description, a case where three visual stimuli blinking at 6 Hz, visual stimuli blinking at 7 Hz, and visual stimuli blinking at 8 Hz will be described as an example. Not exclusively. The number of visual stimuli may be two or more. Moreover, the blinking frequency should just be a frequency which can detect SSVEP, and is different for every visual stimulus.
Hereinafter, a method of detecting the blinking frequency of the visual stimulus by analyzing the frequency spectrum of the electroencephalogram is referred to as “spectrum method”.

In the frequency detection apparatus 1, the electrode 900 detects an electroencephalogram emitted from the primary visual cortex. The frequency detection device 1 includes three electrodes 900. The electrodes 900 are arranged at positions O ₁ , O ₂ , and O _z in the electroencephalogram international 10-10 arrangement method, and an electroencephalogram is provided at each position. Is detected.

FIG. 2 is an explanatory diagram showing the arrangement of the electrodes 900. The figure shows the position of the electrode in the electroencephalogram international 10-10 arrangement method, and the electrode 900 is arranged at O ₁ , O ₂ and O _z among the positions shown in the figure.
However, the arrangement of the electrode 900 is not limited to the positions of O ₁ , O ₂ , and O _z , but may be any position where the visual evoked potential can be detected.

The number of electrodes 900 included in the frequency detection device 1 is not limited to three as shown in FIG. On the other hand, even when the frequency detection device 1 includes a plurality of electrodes 900 and acquires and analyzes brain waves at a plurality of positions, even if the detection accuracy of the brain waves is reduced, such as when noise is externally mixed in any one of the electrodes. Therefore, it is possible to suppress a decrease in frequency detection accuracy.
The electroencephalogram can be detected completely non-invasively, and the electrode 900 may be arranged outside the subject's body. In this respect, the frequency detection device 1 has a small burden on the subject and can be used easily and in the long term.

The frequency detection device main body 100 analyzes the electroencephalogram detected by the electrode 900 and detects the blinking frequency of the visual stimulus that the subject is gazing at. The frequency detection device main body 100 is realized by using a computer such as a PC (Personal Computer).
The signal input unit 110 acquires a signal from the electrode 900. In particular, the signal input unit 110 acquires the brain wave detected by the electrode 900 as an analog brain wave detection signal and converts it into a digital signal. The signal input unit 110 includes, for example, a signal terminal included in the frequency detection device main body 100 and an analog / digital conversion circuit (AD conversion circuit).

The control unit 120 controls each unit of the frequency detection device 1 and executes various functions. In particular, the control unit 120 analyzes the electroencephalogram acquired by the signal input unit 110 and detects the blinking frequency of the visual stimulus that the subject is gazing at.
The control unit 120 is realized, for example, when a CPU (Central Processing Unit) provided in the frequency detection device main body 100 reads out and executes a program from a storage device provided in the frequency detection device main body 100.

  The electroencephalogram measurement unit 121 measures an electroencephalogram emitted from the primary visual cortex using the electrode 900. In particular, the electroencephalogram measurement unit 121 measures the electroencephalogram at a plurality of positions based on signals from the plurality of electrodes 900. Specifically, the electroencephalogram measurement unit 121 acquires an electroencephalogram detection signal detected by each of the electrodes 900 and digitized by the signal input unit 110.

The frequency spectrum acquisition unit 122 acquires the frequency spectrum of the electroencephalogram measured by the electroencephalogram measurement unit 121. Specifically, the frequency spectrum acquisition unit 122 performs Fourier transform on each of the electroencephalogram detection signals for each electrode 900 acquired by the electroencephalogram measurement unit 121.
The frequency spectrum here is data obtained by analyzing the frequency component included in the electroencephalogram by Fourier transforming the electroencephalogram data. Hereinafter, the magnitude of a certain frequency component in the frequency spectrum is referred to as “power value” or “spectral intensity” of the frequency.

The maximum value acquisition unit 123 acquires, for each of a plurality of predetermined frequencies set as the blinking frequency of the visual stimulus, the maximum value of the intensity of the electroencephalogram frequency spectrum in the first predetermined frequency range including the predetermined frequency. . Specifically, the maximum value acquisition unit 123 has a maximum spectrum intensity within a range of plus or minus (±) 0.1 Hz around each blinking frequency of the visual stimulus, with respect to the frequency spectrum acquired by the frequency spectrum acquisition unit 122. Get the value. For example, the maximum value acquisition unit 123 acquires the maximum value of the spectrum intensity in the range of 5.9 Hz to 6.1 Hz with respect to the blinking frequency 6 Hz of the visual stimulus.
The reason why the maximum value acquisition unit 123 acquires the maximum value of the spectrum intensity within the predetermined frequency range is that there may be a deviation between the blinking frequency of the visual stimulus and the frequency of the SSVEP.

  The median value acquisition unit 124, for each of a plurality of predetermined frequencies set as the blinking frequency of the visual stimulus, in the second predetermined frequency range including the predetermined frequency, the median value of the intensity of the electroencephalogram frequency spectrum (Median) To get. Specifically, the median value acquisition unit 124, for the frequency spectrum acquired by the frequency spectrum acquisition unit 122, is the center of the spectrum intensity within a range of plus or minus (±) 0.3 Hz around each blinking frequency of the visual stimulus. Get the value. For example, the median value acquisition unit 124 acquires the median value of the spectrum intensity in the range of 5.7 Hz to 6.3 Hz with respect to the blinking frequency 6 Hz of the visual stimulus.

  The median here is a value located at the center when data is arranged in ascending order (that is, a value at the same position regardless of whether it is counted from the smallest or the largest). The median value acquisition unit 124 acquires the median value among the spectral intensities for each sampling frequency in the Fourier transform that are included in the above range.

  Spectral method frequency detection unit 125 is a difference between a maximum value acquired by maximum value acquisition unit 123 and a median value acquired by median value acquisition unit 124 among a plurality of predetermined frequencies set as blinking frequencies of visual stimuli. The frequency with the largest size is detected. That is, the spectral method frequency detection unit 125 subtracts the median value acquired by the median value acquisition unit 124 from the maximum value acquired by the maximum value acquisition unit 123, thereby obtaining the spectral intensity of the nearby frequency of the spectrum intensity of the maximum value. The peak size is obtained, and the frequency with the largest peak is selected.

  FIG. 3 is an explanatory diagram showing an example of an electroencephalogram detection signal acquired by the electroencephalogram measurement unit 121. In the figure, the horizontal axis represents time (elapsed time from a predetermined reference time), and the vertical axis represents electroencephalogram amplitude (detected potential (voltage) by the electrode 900). A line L11 indicates an example of an electroencephalogram detection signal acquired by the electroencephalogram measurement unit 121 when the subject is gazing at a visual stimulus blinking at 6 Hz.

FIG. 4 is an explanatory diagram illustrating an example of a frequency spectrum acquired by the frequency spectrum acquisition unit 122. In the figure, the horizontal axis indicates the frequency, and the vertical axis indicates the spectrum intensity. A line L21 shows an example of a frequency spectrum obtained by Fourier transforming the electroencephalogram detection signal of FIG.
Points P21, P22, and P23 indicate examples of the maximum values of the spectrum intensity acquired by the maximum value acquisition unit 123 with respect to the blinking frequencies of visual stimuli 6 Hz, 7 Hz, and 8 Hz, respectively. In addition, the range of the frequency from which the maximum value acquisition unit 123 acquires the maximum value of the spectrum intensity is indicated by a chain line.
Sections A21, A22, and A23 indicate frequency ranges in which the median value acquisition unit 124 acquires median values with respect to blinking frequencies of visual stimuli of 6 Hz, 7 Hz, and 8 Hz, respectively.

When the median value acquisition unit 124 acquires the median value, a difference from the maximum value is more likely to occur than when the average value is acquired.
Here, if the median value acquisition unit 124 acquires an average value instead of the median value, the maximum value acquired by the maximum value acquisition unit 123 is included in the data that is the basis of the average value calculation. In this respect, the average value acquired by the median value acquisition unit 124 is affected by the maximum value, and the difference between the maximum value and the average value becomes small. As a result, the accuracy with which the spectrum method frequency detector 125 selects the frequency with the largest peak may be reduced.

  On the other hand, the frequency peak due to SSVEP shows a high value at a frequency that coincides with the blinking of the visual stimulus as in the example at 6 Hz in FIG. 4, but hardly affects the surrounding frequency band. In this respect, the median value acquired by the median value acquisition unit 124 is not directly affected by the maximum value acquired by the maximum value acquisition unit 123, and the difference between the maximum value and the median value becomes large. Accordingly, it is possible to suppress a decrease in accuracy in which the spectrum method frequency detection unit 125 selects a frequency having the largest peak.

The storage unit 150 stores various data. In particular, the storage unit 150 stores a plurality of predetermined frequencies set as visual stimulation blinking frequencies and programs executed by the control unit 120. The storage unit 150 also functions as a working memory for the control unit 120.
The storage unit 150 includes a storage device included in the frequency detection apparatus main body 100.

The result output unit 160 outputs the detection result of the spectrum method frequency detection unit 125. That is, the result output unit 160 outputs a signal indicating the frequency detected by the spectrum method frequency detection unit 125. The detection result of the spectral frequency detection unit 125 output from the result output unit 160 is used as a command for instructing the operation of the operation target device.
The result output unit 160 includes, for example, a communication circuit included in the frequency detection device main body 100.

Next, the operation of the frequency detection device 1 will be described with reference to FIG.
FIG. 5 is a flowchart illustrating an example of a processing procedure in which the frequency detection device 1 detects the blinking frequency of the visual stimulus that the subject is gazing at.
In the process of drawing, the electroencephalogram measurement section 121 _measures the brain waves with electrodes of _{O 1, O 2, O z} ( step S101). That, O _{1, O 2,} each of the electrodes 900 installed at the position of the O _z outputs a brain wave detection signal by detecting the brain wave, EEG detector 121, the respective brain wave detection signal via the signal input unit 110 get.

Next, the control unit 120 starts a loop L111 for performing processing for each electroencephalogram obtained in step S101 (and therefore for each electrode 900) (step S111). Hereinafter, the electroencephalogram to be processed in the loop L111 is referred to as “the electroencephalogram to be processed”.
In the loop L111, the frequency spectrum acquisition unit 122 performs a Fourier transform on the brain wave to be processed to acquire a frequency spectrum (step S112).

Next, the control unit 120 starts a loop L112 that performs processing for each predetermined frequency set as the blinking frequency of the visual stimulus (step S121). Hereinafter, the frequency to be processed in the loop L112 is referred to as “frequency to be processed”.
In the loop L112, the maximum value acquisition unit 123 sets the maximum value of the frequency spectrum in the range of plus or minus 0.1 Hz centering on the frequency of the processing target (peak intensity of the frequency component) with respect to the frequency spectrum of the brain wave to be processed. ) Is acquired (step S122).

Next, the median value acquisition unit 124 acquires the median value of the intensity of the frequency spectrum in the range of plus or minus 0.3 Hz centering on the frequency of the processing target with respect to the frequency spectrum of the brain wave to be processed (step S123). .
Then, the spectral method frequency detection unit 125 subtracts the median value acquired by the median value acquisition unit 124 in step S123 from the maximum value acquired by the maximum value acquisition unit 123 in step S122 (step S124).

  Next, the control unit 120 performs termination processing of the loop L112 (step S125). Specifically, the control unit 120 determines whether or not the process of the loop L112 has been performed for all of the predetermined frequencies set as the blinking frequency of the visual stimulus. If it is determined that the process of the loop L112 has been performed for all of the predetermined frequencies, the loop L112 is terminated. On the other hand, when it is determined that there is an unprocessed predetermined frequency, the process returns to step S121, and the process of the loop L112 is continuously performed on the unprocessed predetermined frequency.

When the loop L112 is terminated, the spectral method frequency detection unit 125 detects (selects) a predetermined frequency having the maximum difference obtained in step S124 (step S131).
Next, the control unit 120 performs termination processing of the loop L111 (step S132). Specifically, the control unit 120 determines whether or not the process of the loop L111 has been performed on all the electroencephalograms obtained in step S101. If it is determined that the process of the loop L111 has been performed on all the electroencephalograms, the loop L111 is terminated. On the other hand, if it is determined that there is an unprocessed brain wave, the process returns to step S111, and the process of the loop L111 is continuously performed on the unprocessed brain wave.

When the loop L111 is completed, the spectrum method frequency detection unit 125 performs the majority of the frequencies obtained in step S131 based on the brain waves detected by the O ₁ , O ₂ , and _Oz electrodes (step S141).
And the result output part 160 outputs the signal which shows the frequency obtained by step S142 as a detection result of the frequency detection apparatus 1 (step S142).
Thereafter, the process of FIG.
Note that the control unit 120 may perform parallel processing (for example, parallel processing) on the processing of the loop L111, the processing of the loop L112, or both processing.

  As described above, the frequency spectrum acquisition unit 122 acquires the frequency spectrum of the electroencephalogram measured by the electroencephalogram measurement unit 121. The maximum value acquisition unit 123 then adds, for each of the plurality of predetermined frequencies set as the blinking frequency of the visual stimulus, a first predetermined frequency range including the predetermined frequency (in the present embodiment, the predetermined frequency is a plus). The maximum value of the intensity of the electroencephalogram frequency spectrum in the range of minus 0.1 Hz) is acquired. Further, the median value acquisition unit 124, for each of a plurality of predetermined frequencies set as the blinking frequency of the visual stimulus, is added to a second predetermined frequency range including the predetermined frequency (in the present embodiment, the predetermined frequency is a plus). The median of the intensity of the frequency spectrum of the electroencephalogram in the range of minus 0.3 Hz) is acquired. And the maximum value acquisition part 123 detects the frequency with which the magnitude | size of the difference of a maximum value and a median is the largest among several predetermined frequencies set as a blink frequency of visual stimulation.

Thereby, the frequency detection apparatus 1 can output the signal which shows the blink frequency of the visual stimulus which the subject gazes as an output of BCI using SSVEP, without requiring learning.
In particular, since the median value acquisition unit 124 acquires the median value, a difference from the maximum value acquired by the maximum value acquisition unit 123 is easily generated, and the spectrum method frequency detection unit 125 selects the frequency with the largest peak. Can be suppressed. Thereby, the frequency detection apparatus 1 can detect the blink frequency of the visual stimulus which the subject is gazing with high accuracy.
Thus, the frequency detection apparatus 1 can obtain the output of the BCI using SSVEP more accurately without learning.

The electroencephalogram measurement unit 121 measures electroencephalograms at a plurality of positions based on signals from a plurality of electrodes 900 that detect electroencephalograms emitted from the primary visual cortex at different positions.
As a result, the frequency detection device 1 can suppress a decrease in the frequency detection accuracy even when the detection accuracy of the electroencephalogram is reduced, for example, noise is mixed in any of the electrodes.

The inventor of the present application confirmed the effectiveness of the frequency detection method according to the present embodiment through experiments. Hereinafter, the experiment will be described.
Thirty-two people (27 men, 5 women, average age 21.09 years old) who did not feel inconvenience in daily life with respect to vision became subjects (corresponding to the above “subject”). Five of the subjects (4 men and 1 woman) had experienced BCI before, while the other 27 had never received EEG measurements in the past. The subjects were informed of the purpose of the experiment by the experimenter in advance and participated in the experiment with written informed consent.

For electroencephalogram measurement g. An electroencephalogram measurement system (g. BCIsys32USB) manufactured by Tec was used. The electrode was attached _O in _1, O 2, the position of the _{O z} in EEG International 10-10 method arrangement, the ground electrode to the position of _{F z,} fitted with a reference electrode in the right earlobe. The sampling frequency was 256 Hz, and the band pass filter (Band Pass Filter) was set to 0.1 Hz to 30 Hz. The notch filter was set to 50 Hz. In addition, as a visual stimulus, a circuit using a microcomputer (mbed, ARM) and a white LED (light emitting diode) is configured on a breadboard so that it can blink at a set frequency. The on / off time ratio (duty ratio) was 50 percent (%).

FIG. 6 is an explanatory diagram showing a light emission pattern of the LED in the effectiveness verification experiment. As shown in the figure, the subject gazes at two sets of LEDs blinking at 6 Hz, 7 Hz, and 8 Hz for 35 seconds. In addition, as shown in the figure, a 30-second rest (rest with eyes closed) is taken for each enforcement to reduce eye fatigue.
The laboratory was in a dim state with the lights turned off, and the subject looked at the LED about 50 centimeters (cm) away from the LED. Moreover, in order to exclude the influence by the sound stimulus from the outside world, the subject wore earplugs.
The same experiment was conducted for 2 days, and sufficient electroencephalogram data samples were obtained for each subject.

  The inventor of the present application calculated the identification rate (correct frequency detection rate) using numerical analysis software MATLAB (registered trademark) of MathWorks. Data for 5 seconds after the start of the measurement including the time zone for adjusting the electroencephalograph gain was excluded from the calculation of the discrimination rate. Therefore, for each subject, the identification rate was calculated using data for a total of 120 seconds for 30 seconds × 2 sets × 2 days from 5 seconds to 35 seconds for each blinking frequency.

The inventor of the present application obtains the frequency spectrum of the electroencephalogram detected by each of the electrodes O ₁ , O ₂ , and O _z described above, and detects the frequency at which the difference between the maximum value and the intermediate value of the frequency spectrum intensity is maximum. Then, the identification rate of the method of making a majority vote was calculated.
In addition, as a conventional method for comparison, O _z of LDA in brain waves detected by the electrodes (Linear Discriminant Analysis, Linear Discriminant Analysis) by applying the method of detecting the flicker frequency of the visual stimulus (hereinafter, " (Referred to as “LDA method”). With LDA, an identification plane is obtained in advance using learning data divided into two groups, and the group to which newly input data (data to be discriminated) belongs belongs using the identification plane. It is a method of discrimination.

FIG. 7 is an explanatory diagram illustrating an example of an identification surface in the LDA. The horizontal axis and the vertical axis in the figure indicate the index value A and the index value B, respectively, and the learning data of a set of the index value A and the index value B is plotted. In addition, it is known which learning data each of the two groups belongs to.
The identification plane is a straight line that bisects a straight line connecting the centroids of the respective groups (a hyperplane when the feature vector is three-dimensional or more). In the example of FIG. 7, points P1001 and P1002 indicate the centers of gravity of group 1 and group 2, respectively, and line L1001 indicates the identification plane.
In the following, a case where LDA is performed using a graph will be described as an example. However, it is also possible to perform LDA by arithmetic processing without drawing a graph.

In the LDA method, when the blinking frequency is detected, plus / minus 0.1 Hz centered on 6 Hz, plus / minus 0.1 Hz centered on 7 Hz, and maximum value of frequency spectrum for plus / minus 0.1 Hz centered on 8 Hz. To get. Then, using the obtained maximum value as the power value of each frequency band, each combination of two of the three power values as a feature vector, and identifying which of the two frequencies the subject is gazing at Discriminate based on surface.
Since LDA can only discriminate in the two-class discrimination (binary choice) method, discrimination is performed for each of the three combinations of two frequencies, the majority of the discrimination results is determined, and the most frequently selected frequency is determined. The detection result.

FIG. 8 is an explanatory diagram showing an example of LDA for a combination of a power value of 6 Hz and a power value of 7 Hz. In the figure, the horizontal axis indicates a power value of 6 Hz, and the vertical axis indicates a power value of 7 Hz. A line L1011 indicates an identification surface.
The point P1011 to be discriminated is located on the 6 Hz side of the region divided by the identification plane (line L1011), and is determined to be 6 Hz by the LDA.

FIG. 9 is an explanatory diagram showing an example of LDA for a combination of a power value of 7 Hz and a power value of 8 Hz. In the figure, the horizontal axis indicates a power value of 7 Hz, and the vertical axis indicates a power value of 8 Hz. A line L1021 indicates an identification surface.
The point P1021 to be discriminated is located on the 8 Hz side in the region divided by the identification plane (line L1021), and is determined to be 8 Hz by the LDA.

FIG. 10 is an explanatory diagram showing an example of LDA for a combination of a power value of 6 Hz and a power value of 8 Hz. In the figure, the horizontal axis indicates a power value of 6 Hz, and the vertical axis indicates a power value of 8 Hz. A line L1031 indicates an identification surface.
The point P1031 to be discriminated is located on the 6 Hz side in the region divided by the identification plane (line L1031), and is determined to be 6 Hz by the LDA.
A detection result of 6 Hz is obtained by majority decision of the three determination results shown in FIGS.

FIG. 11 is a flowchart showing a processing procedure for acquiring an identification plane by learning in the LDA method. In the process of drawing, first, to measure the brain waves at the location of the electrodes of the _{O z} (step S1101). In learning, the blinking frequency of the visual stimulus is known.
Next, the brain wave obtained in step S1101 is Fourier transformed to obtain a frequency spectrum (step S1102).

Then, the peak intensity of each predetermined frequency set as the blinking frequency of the visual stimulus is detected (step S1103). Specifically, the maximum value of the frequency spectrum is acquired for each of plus / minus 0.1 Hz centered on 6 Hz, plus / minus 0.1 Hz centered on 7 Hz, and plus / minus 0.1 Hz centered on 8 Hz.
Next, like the graphs shown in the examples of FIGS. 8 to 10, the peak intensities obtained in step S1103 are plotted on the respective graphs of combinations of two predetermined frequencies (step S1104).

Next, it is determined whether or not an appropriate number of data has been obtained (step S1105). The number of data is preferably at least 6 or more. In addition, in order to expect more accuracy, the inventor of the present application conducted an experiment using about 10 times (about 60) data.
If it is determined in step S1105 that an appropriate number of data has not been obtained (step S1105: NO), the process returns to step S1101.
On the other hand, when it is determined that an appropriate number of data has been obtained (step S1105: YES), the center of gravity of each group is calculated for each graph (step S1111). Then, an identification plane is calculated for each graph (step S1112).
Then, the process of FIG. 11 is complete | finished.

FIG. 12 is a flowchart showing a processing procedure for determining the blinking frequency in the LDA method. In the process of drawing, first, to measure the brain waves at the location of the electrodes of the _{O z} (step S1201). Then, the brain wave obtained in step S1201 is Fourier transformed to obtain a frequency spectrum (step S1202).

Next, the peak intensity of each predetermined frequency set as the blinking frequency of the visual stimulus is detected (step S1203). Specifically, the maximum value of the frequency spectrum is acquired for each of plus / minus 0.1 Hz centered on 6 Hz, plus / minus 0.1 Hz centered on 7 Hz, and plus / minus 0.1 Hz centered on 8 Hz.
Then, as in the graphs shown in the examples of FIGS. 8 to 10, the peak intensity obtained in step S1203 is plotted on each graph of the combination of two predetermined frequencies (step S1204).

Next, as in the examples of FIGS. 8 to 10, in each graph, it is determined based on the identification plane which group the determination target data is included in (step S1205).
Then, the majority of the determination results for each graph is performed, and the most frequent determination result is set as the detection result of the blinking frequency of the visual stimulus (step S1206).
Thereafter, the process of FIG.

When determining the identification rate, in order to examine how many seconds the correct identification is performed, the data was divided every 3 seconds to determine the average of the identification rates. Similarly, for 4 seconds and 5 seconds, the data was divided and the average of the identification rates was calculated.
Moreover, in order to investigate the discrimination performance by LDA, 4-fold intersection confirmation was used. In n-fold (n is a positive integer), the data is divided into n, one of them is used as identification data, and the rest is used as learning data to determine the correct answer rate. In this procedure, the correct rate when each of the n-divided data is used as identification data is obtained in order, and these are averaged to calculate the discrimination rate.
In the 4-fold intersection confirmation used in this experiment, the identification rate was calculated by dividing the data into four.

FIG. 13 is an explanatory diagram illustrating an example of selecting identification data and learning data in 4-fold intersection confirmation.
As shown in the figure, in the 4-fold intersection confirmation, one of the four divided data is sequentially used as identification data, and the other data is used as learning data to calculate an identification rate. Then, the average of the obtained identification rates is calculated to obtain the estimation accuracy (in this experiment, the detection accuracy of the blinking frequency of the visual stimulus).

In addition, the inventor of the present application uses the numerical analysis software MATLAB (registered trademark) of The MathWorks and the statistical analysis software SPSS (registered trademark) of IBM Japan to detect at each electrode of O ₁ , O ₂ , and O _z. The frequency spectrum of the electroencephalogram is calculated, and the method of voting by detecting the frequency where the difference between the maximum value and the intermediate value of the frequency spectrum intensity is the maximum is determined, and the existence of a significant difference between the LDA method is tested. It was. The risk rate in the test was 5%, and if it was less than this, the null hypothesis was rejected and there was a statistically significant difference between the two groups.

  First, whether or not the distribution of spectral intensity data follows a normal distribution by the Shapiro-Wilk test is used. When the distribution follows a normal distribution (both p ≧ 0.05), a corresponding t-test is used for comparison between the two groups. . On the other hand, when the normal distribution was not followed (at least one was p <0.05), Wilcoxon signed rank test was used.

FIG. 14 is an explanatory diagram showing experimental results for the first embodiment. In the figure, the visual recognition blinking frequencies of 6 Hz, 7 Hz, 8 Hz, and the average of these experimental results, the average value of the discrimination rate and the value of the risk factor p are shown. Column "spectrum _{method", O 1, O 2, O} obtains a frequency spectrum of the EEG detected by each electrode of _z, detecting a frequency difference between the maximum value and the intermediate value of the frequency spectral intensity is maximized Thus, the average value of the identification rate by the method of making a majority vote is shown. The column “LDA” indicates the identification rate by the LDA method. This figure shows the experimental results for each case where the data is divided every 3 seconds, divided every 4 seconds, or divided every 5 seconds.

In the experimental results shown in FIG. 14, the risk rate is 5% or less in many cases. As a result, the frequency spectrum of the electroencephalogram detected at each of the electrodes O ₁ , O ₂ , and O _z is obtained, and the majority is determined by detecting the frequency at which the difference between the maximum value and the intermediate value of the frequency spectrum intensity is maximum. The superiority of the method is shown.
In particular, when the data is divided every 4 seconds and when divided every 5 seconds, the risk rate is 5 percent or less (0 percent) for all of the visual stimulus blinking frequencies 6 Hz, 7 Hz, and 8 Hz. A strong advantage is shown.

In addition, as described above, many of the subjects have experienced brain wave measurement for the first time. At this point, the frequency spectrum of the electroencephalogram detected at each of the electrodes O ₁ , O ₂ , and O _z is obtained, and the majority is determined by detecting the frequency at which the difference between the maximum value and the intermediate value of the frequency spectrum intensity is maximum. The method performed does not require prior training and can be said to be immediately applicable.

Note that the width of the range (first predetermined frequency range) in which the maximum value acquisition unit 123 acquires the maximum value is not limited to the above-described plus or minus 0.1 Hz. The width of the range (second predetermined frequency range) in which the median value acquisition unit 124 acquires the median value is not limited to the above-described plus or minus 0.3 Hz.
For example, the width of the range (first predetermined frequency range) in which the maximum value acquisition unit 123 acquires the maximum value may be a range of plus or minus 0.5 Hz, and the median value acquisition unit 124 acquires the median value. The width of the range (second predetermined frequency range) may be a range of plus or minus 3 Hz. In particular, the combination of plus / minus 0.1 Hz and plus / minus 0.3 Hz described above is suitable for the combination of the width of the first predetermined frequency range and the width of the second predetermined frequency range.

<Second Embodiment>
FIG. 15 is a schematic block diagram illustrating a functional configuration of a frequency detection device according to the second embodiment of the present invention. In the figure, the frequency detection device 2 includes a frequency detection device main body 200 and an electrode 900. The frequency detection device main body 200 includes a signal input unit 110, a control unit 220, a storage unit 150, and a result output unit 160. The control unit 220 includes an electroencephalogram measurement unit 121, a periodic data extraction unit 227, an addition average calculation unit 228, and an addition method frequency detection unit 229.
15, parts having the same functions corresponding to the respective parts in FIG. 1 are denoted by the same reference numerals (110, 121, 150, 160, 900), and description thereof is omitted.

The frequency detection device 2 detects the blinking frequency of the visual stimulus based on the data obtained by adding the data extracted from the brain wave by the time window for each predetermined period set as the cycle corresponding to the blinking frequency of the visual stimulus. . The period corresponding to the frequency here is specifically a period as the reciprocal of the frequency.
As in the case of the frequency detection device 1 (FIG. 1), the frequency detection device 2 can function as a BCI.
The frequency detection device 2 corresponds to an example of an electroencephalogram frequency detection device.

The frequency detection device main body 200 extracts data from the electroencephalogram detected by the electrode 900 using a time window for each predetermined period set as a period corresponding to the blinking frequency of the visual stimulus, and based on data obtained by adding the extracted data. The blinking frequency of the visual stimulus is detected. The frequency detection device main body 200 is realized by using a computer such as a PC (Personal Computer).
Hereinafter, a method of detecting the blinking frequency of the visual stimulus based on the data obtained by adding the data extracted from the electroencephalogram data in the time window corresponding to the cycle corresponding to the blinking frequency of the visual stimulus is referred to as “addition method”. .

The control unit 220 controls each unit of the frequency detection device 2 and executes various functions. In particular, the control unit 220 analyzes the electroencephalogram acquired by the signal input unit 110 and detects the blinking frequency of the visual stimulus that the subject is gazing at.
The control unit 220 is realized by, for example, a CPU (Central Processing Unit) included in the frequency detection device main body 200 reading and executing a program from a storage device included in the frequency detection device main body 200.

  The cycle data cutout unit 227 uses each time of the predetermined cycle as the start time from the electroencephalogram measured by the electroencephalogram measurement unit 121 for each of a plurality of predetermined cycles set as a cycle corresponding to the blinking frequency of the visual stimulus. Cut out the periodic data to be used. That is, the cycle data cutout unit 227 cuts out data from the electroencephalogram measured by the electroencephalogram measurement unit 121 using a time window for each predetermined cycle set as a cycle corresponding to the blinking frequency of the visual stimulus.

The addition average calculation unit 228 adds the cycle data cut out by the cycle data cutout unit 227 for each predetermined cycle set as a cycle corresponding to the blinking frequency of the visual stimulus. Thereby, the addition average calculation unit 228 periodically superimposes the data cut out from the electroencephalogram.
Specifically, the addition average calculation unit 228 calculates the addition average of the period data extracted by the period data extraction unit 227 for each predetermined period set as a period corresponding to the blinking frequency of the visual stimulus.

  The addition average is a conventionally known method, but it is not common to use the addition average for SSVEP. In particular, for steady-state evoked responses, it is common to perceive that the brain waves are moving constantly at a certain frequency, and processing such as cutting out and adding steady-state evoked response data is unnecessary and cumbersome. This is considered to be normal and not done.

  Also, in the EEG measurement method that generally performs averaging processing, a stimulus (trigger) that changes the EEG is given, and the response latency and the amplitude at each latency until the EEG changes actually appear. In order to measure the change of the value, a channel for inputting the timing at which the stimulus is applied is necessary. On the other hand, in addition averaging in the present invention, it is only necessary to use amplitude intensity information, so there is no need to consider reaction latency, and no channel for inputting a trigger is required.

Furthermore, in a method of measuring an electroencephalogram that generally performs an averaging process, it is difficult to obtain a preferable waveform unless a baseline process is performed. On the other hand, in the case of the addition average in the present invention, the difference between the maximum value and the minimum value of the amplitude intensity is used. Therefore, the effect of the present invention can be achieved without performing the baseline processing.
In addition, the process in which the addition average calculation unit 228 periodically superimposes the data cut out from the electroencephalogram may be a process of adding the periodic data, and is not limited to the addition average. For example, the addition average calculation unit 228 may perform only addition of periodic data and not perform averaging.

FIG. 16 illustrates an example of the addition average of data extracted by the periodic data extraction unit 227 in a time window corresponding to 6 Hz (a time window provided with a 1/6 second period) calculated by the addition average calculation unit 228. It is explanatory drawing shown. The horizontal axis of the figure shows the time in the time window (time from the time window start time). The vertical axis represents the amplitude of the electroencephalogram (detected potential by the electrode 900).
A line L31 indicates an example of data obtained by addition averaging performed by the addition average calculation unit 228 when the subject is gazing at a visual stimulus having a blinking frequency of 6 Hz.

FIG. 17 is an explanatory diagram illustrating an example of the addition average of data calculated by the addition average calculation unit 228 and extracted by the periodic data extraction unit 227 using a time window corresponding to 7 Hz. The horizontal axis of the figure shows the time in the time window. The vertical axis represents the amplitude of the electroencephalogram (detected potential by the electrode 900).
A line L41 indicates an example of data obtained by addition averaging performed by the addition average calculation unit 228 when the subject is gazing at a visual stimulus having a blinking frequency of 6 Hz.

FIG. 18 is an explanatory diagram illustrating an example of the addition average of data calculated by the addition average calculation unit 228 and extracted by the periodic data extraction unit 227 using a time window corresponding to 8 Hz. The horizontal axis of the figure shows the time in the time window. The vertical axis represents the amplitude of the electroencephalogram (detected potential by the electrode 900).
A line L51 indicates an example of data obtained by addition averaging performed by the addition average calculation unit 228 when the subject is gazing at a visual stimulus having a blinking frequency of 6 Hz.

  Referring to FIGS. 16 to 18, in the example of FIG. 16 in which the period corresponding to the blinking frequency of the visual stimulus and the period in which the time window is provided match, the change in the amplitude of the electroencephalogram is large. On the other hand, in the example of FIG. 17 and the example of FIG. 18 in which the period corresponding to the blinking frequency of the visual stimulus and the period in which the time window is provided do not match, the change in the amplitude of the electroencephalogram is small.

  Here, as main components of VEP (visual evoked potential), components of potential change appearing after about 75 milliseconds, 100 milliseconds, and 145 milliseconds after receiving a visual stimulus are known. It is called the component, the P100 component, and the N145 component. When the visual stimulus blinks periodically, these visual evoked potentials also appear periodically corresponding to the cycle.

  Therefore, when brain wave data is cut out with a period corresponding to the period corresponding to the blinking frequency of the visual stimulus and the averaging is performed, a clear visual evoked potential appears in the obtained waveform. On the other hand, when brain wave data is cut out with a different period from the period corresponding to the blinking frequency of the visual stimulus and the addition averaging is performed, the appearance position of the visual evoked potential differs for each piece of the cut out data, so that the addition average is obtained. A clear visual evoked potential does not appear in the waveform.

  The addition method frequency detection unit 229 generates a positive / negative sign of data obtained by adding the cycle data for each predetermined cycle among the predetermined cycles set as a cycle corresponding to the blinking frequency of the visual stimulus generated by the addition average calculation unit 228. A frequency corresponding to a period in which the difference in peak intensity is maximum is detected. Specifically, the addition frequency detection unit 229 selects data having the largest difference between the maximum value and the minimum value from the data generated by the addition average calculation unit 228 for each predetermined period, and corresponds to the data. A predetermined frequency is acquired as a detection frequency.

As described above, when electroencephalogram data is cut out with a period corresponding to the period corresponding to the blinking frequency of the visual stimulus and the averaging is performed, a clear visual evoked potential appears in the obtained waveform. Thereby, in the data obtained by cutting out the electroencephalogram data at a period corresponding to the period corresponding to the blinking frequency of the visual stimulus and performing the averaging, the difference between the maximum value and the minimum value becomes large.
On the other hand, when electroencephalogram data is cut out with a period different from the period corresponding to the blinking frequency of the visual stimulus and the averaging is performed, a clear visual evoked potential does not appear in the waveform obtained by the averaging. As a result, the difference between the maximum value and the minimum value is small in the data obtained by cutting out the electroencephalogram data with a period different from the period corresponding to the blinking frequency of the visual stimulus and performing the averaging.

  Thus, the addition method frequency detection unit 229 selects the data having the largest difference between the maximum value and the minimum value, and thereby obtains the electroencephalogram data cut out at a cycle corresponding to the cycle corresponding to the blinking frequency of the visual stimulus. Data (and thus data associated with a period corresponding to the blinking frequency of the visual stimulus) can be selected. Thereby, the addition method frequency detection unit 229 can detect the blinking frequency of the visual stimulus.

Next, the operation of the frequency detection device 2 will be described with reference to FIG.
FIG. 19 is a flowchart illustrating an example of a processing procedure in which the frequency detection device 2 detects the blinking frequency of the visual stimulus that the subject is gazing at.
Steps S201, S211, and S221 in the figure are the same as steps S101, S111, and S121 in FIG. 5, respectively.

In the loop L122, the cycle data cutout unit 227 cuts out data from the electroencephalogram measured by the electroencephalogram measurement unit 121 using a time window for each predetermined cycle set as a cycle corresponding to the blinking frequency of the visual stimulus (step S222). .
And the addition average calculation part 228 calculates the addition average of the period data which the period data extraction part 227 cut out by step S222 for every predetermined period set as a period corresponding to the blinking frequency of a visual stimulus (step). S223).

Next, the addition method frequency detection unit 229 detects the maximum value for each data obtained in step S223 (step S224).
Moreover, the addition method frequency detection unit 229 detects the minimum value for each data obtained in step S223 (step S225).
Furthermore, the addition method frequency detection unit 229 calculates, for each data obtained in step S223, the difference between the maximum value obtained in step S224 and the minimum value obtained in step S225 (step S226).

Step S227 is the same as step S125 of FIG.
When the loop L122 ends, the addition method frequency detection unit 229 selects a frequency having the maximum difference obtained in step S226 (step S231).
Step S232 is the same as step S132 of FIG.

Upon completion of the loop L121, the addition method the frequency detecting unit 229 _performs the _{O 1,} O _2, O _z frequency majority obtained in step S231 based on the brain wave detected by each electrode (step S241).
And the result output part 160 outputs the signal which shows the frequency obtained by step S241 as a detection result of the frequency detection apparatus 2 (step S242).
Then, the process of FIG. 19 is complete | finished.
Note that the control unit 220 may perform parallel processing (for example, parallel processing) on the processing of the loop L121, the processing of the loop L122, or both processing.

  As described above, the cycle data cutout unit 227 has each of the predetermined cycles from the electroencephalogram acquired by the electroencephalogram measurement unit 121 for each of a plurality of predetermined cycles set as a cycle corresponding to the blinking frequency of the visual stimulus. Cut out periodic data starting from time. Then, the addition method frequency detector 229 has a maximum difference in positive and negative peak intensities of data obtained by adding the period data for each predetermined period among the predetermined periods set as the period corresponding to the blinking frequency of the visual stimulus. A frequency corresponding to a certain cycle is detected.

Thereby, the frequency detection apparatus 2 can output the signal which shows the blink frequency of the visual stimulus which the subject gazes as an output of BCI using SSVEP, without requiring learning.
Further, the addition average calculation unit 228 performs addition of the periodic data (addition average in the present embodiment), so that the influence of noise can be reduced even when the brain wave data includes noise.
As described above, the frequency detection device 2 can obtain the output of the BCI using the SSVEP more accurately without learning.

The inventor of the present application confirmed the effectiveness of the frequency detection method according to the present embodiment through experiments. Specifically, the inventor of the present application, for each predetermined period set as a period corresponding to the blinking frequency of the visual stimulus, from the brain waves detected by the electrodes of O ₁ , O ₂ , and O _z described above. The discrimination rate of the method in which the majority is determined by detecting the frequency with the maximum difference between the maximum value and the minimum value in the average of the data cut out in the time window was calculated.
The electroencephalogram data acquired in the experiment described in the first embodiment was used, and a comparison with the LDA method was performed as in the experiment described in the first embodiment.

FIG. 20 is an explanatory diagram showing experimental results for the second embodiment. In the figure, the visual recognition blinking frequencies of 6 Hz, 7 Hz, 8 Hz, and the average of these experimental results, the average value of the discrimination rate and the value of the risk factor p are shown. Column "addition method" _was cut out in _O 1, O _2, O from the brain wave detected by each electrode of _z, the time window for each predetermined period which is set as the period corresponding to the flicker frequency of the visual stimulus The average value of the discrimination rate by the method of detecting the frequency with the largest difference between the maximum value and the minimum value in the addition average of data and performing a majority decision is shown. The column “LDA” indicates the identification rate by the LDA method. This figure shows the experimental results for each case where the data is divided every 3 seconds, divided every 4 seconds, or divided every 5 seconds.

In the experimental results shown in FIG. 20, the risk rate is 5% or less in many cases. Thus, from the brain wave detected by each electrode of _{O 1, O 2, O z} , in averaging the data cut out in a time window of a predetermined cycle which is set as the period corresponding to the flicker frequency of the visual stimulus It shows the superiority of the method of voting by detecting the frequency where the difference between the maximum value and the minimum value is maximum.
In particular, when the data is divided every 5 seconds, the risk factor is 5 percent or less (0 percent) for all of the visual stimulus blinking frequencies of 6 Hz, 7 Hz, and 8 Hz, indicating a strong advantage.
Further, as described for the first embodiment, O _{1, O 2,} from the brain wave detected by each electrode of the O _z, every predetermined period that is set as the period corresponding to the flicker frequency of the visual stimulus It can be said that the method of detecting the frequency with the maximum difference between the maximum value and the minimum value in the average of the data cut out in the time window and performing the majority decision does not require prior training and can be applied immediately.

Note that the start timing of the time window set by the periodic data extraction unit 227 for extraction of data from the electroencephalogram does not have to coincide with the lighting start timing or lighting end timing of the visual stimulus. Therefore, it is not necessary to synchronize the visual stimulus with the frequency detection device 2.
On the other hand, when the start timing of the time window set by the periodic data cutout unit 227 to cut out data from the electroencephalogram coincides with the lighting start timing or lighting end timing of the visual stimulus, N75, P100, N145, etc. In addition, it is easy to confirm a visual evoked potential with a known relationship with the timing of the start of lighting of the visual stimulus.

<Third Embodiment>
FIG. 21 is a schematic block diagram illustrating a functional configuration of a frequency detection device according to the third embodiment of the present invention. In the figure, the frequency detection device 3 includes a frequency detection device main body 300 and an electrode 900. The frequency detection device main body 300 includes a signal input unit 110, a control unit 320, a storage unit 150, and a result output unit 160. The control unit 320 includes an electroencephalogram measurement unit 121, a frequency spectrum acquisition unit 122, a maximum value acquisition unit 123, a median value acquisition unit 124, a spectrum method frequency detection unit 125, a condition determination unit 326, and periodic data extraction. Unit 227, addition average calculation unit 228, and addition method frequency detection unit 229.
In FIG. 21, the same reference numerals (110, 121 to 125, 150, 160, 227 to 229, 900) are given to portions having similar functions corresponding to the respective portions in FIG. 1 or FIG. .

The frequency detection device 3 detects the blinking frequency of the visual stimulus by a combination of the spectrum method described in the first embodiment and the addition method described in the second embodiment. As in the case of the frequency detection device 1 (FIG. 1), the frequency detection device 3 can function as a BCI.
The frequency detection device main body 300 detects the blinking frequency of the visual stimulus based on the brain wave detected by the electrode 900 by a combination of the spectrum method and the addition method. The frequency detection apparatus main body 300 is realized by using a computer such as a PC (Personal Computer).
The frequency detection device 3 corresponds to an example of an electroencephalogram frequency detection device.

The control unit 320 controls each unit of the frequency detection device 3 to execute various functions. In particular, the control unit 320 detects the blinking frequency of the visual stimulus based on the brain wave acquired by the signal input unit 110 using a combination of the spectrum method and the addition method.
The control unit 320 is realized, for example, when a CPU (Central Processing Unit) included in the frequency detection device main body 300 reads out and executes a program from a storage device included in the frequency detection device main body 300.

  The condition determination unit 326 determines whether the frequency detected by the spectrum method frequency detection unit 125 satisfies a predetermined condition. Specifically, the condition determination unit 326 has a difference calculated by the spectrum method frequency detection unit 125 that is at least three times the difference at other frequencies at the frequency at which the difference is the maximum, and is similar for all electrodes. It is determined whether or not a determination result is obtained. In this way, the condition determination unit 326 evaluates the size of the peak in the frequency spectrum of the electroencephalogram.

When it is determined that the frequency detected by the spectrum method frequency detection unit 125 satisfies the predetermined condition, the condition determination unit 326 outputs data indicating the frequency detected by the spectrum method frequency detection unit 125 to the result output unit 160. Then, the result output unit 160 outputs (transmits) the data to the outside of the frequency detection device main body 300 as a detection result of the frequency detection device main body 300.
On the other hand, when it is determined that the frequency detected by the spectrum method frequency detection unit 125 does not satisfy the predetermined condition, the condition determination unit 326 sends the periodic data extraction unit 227, the addition average calculation unit 228, and the addition method frequency detection unit 229 to each other. The processing described in the second embodiment is performed. In this case, the addition method frequency detection unit 229 outputs data indicating the detected frequency to the result output unit 160. And the result output part 160 outputs the said data as a detection result of the frequency detection apparatus main body 300. FIG.

Next, the operation of the frequency detection device 3 will be described with reference to FIG.
FIG. 22 is a flowchart illustrating an example of a processing procedure in which the frequency detection device 3 detects the blinking frequency of the visual stimulus that the subject is gazing at.
Step S301 in the figure is the same as step S101 in FIG.
In step S302, the control unit 320 (frequency spectrum acquisition unit 122, maximum value acquisition unit 123, median value acquisition unit 124, and spectrum method frequency detection unit 125) performs the same processing as steps S111 to S141 in FIG.

After step S302, the condition determination unit 326 determines whether or not the predetermined condition described above is satisfied (step S303). When it is determined that the condition is satisfied (step S303: YES), the condition determination unit 326 outputs a signal indicating the frequency detected by the spectrum method frequency detection unit 125 to the result output unit 160, and the result output unit 160 Is output as a detection result of the frequency detection device main body 300 (step S311).
Thereafter, the process of FIG. 22 is terminated.

On the other hand, when it is determined in step S303 that the condition is not satisfied (step S303: NO), the control unit 320 (period data extraction unit 227, addition average calculation unit 228, and addition method frequency detection unit 229) performs step of FIG. Processing similar to S211 to S241 is performed (step S321).
And the addition method frequency detection part 229 outputs the signal which shows the detected frequency to the result output part 160, and the result output part 160 outputs the said signal as a detection result of the frequency detection apparatus main body 300 (step S322).
Thereafter, the process of FIG. 22 is terminated.

  As described above, the condition determination unit 326 determines whether the frequency detected by the spectrum method frequency detection unit 125 satisfies a predetermined condition. When the condition determination unit 326 determines that the frequency detected by the spectrum method frequency detection unit 125 does not satisfy the predetermined condition, the period data extraction unit 227 is a plurality of periods set as a period corresponding to the blinking frequency of the visual stimulus. For each of the predetermined periods, period data starting from each time of the predetermined period is cut out from the electroencephalogram. And the addition method frequency detection part 229 detects the frequency of the peak of an electroencephalogram based on the data which added period data for every predetermined period set as a period corresponding to the blinking frequency of a visual stimulus.

As described above, the condition determination unit 326 performs the condition determination on the frequency detected by the spectrum method frequency detection unit 125, so that the frequency detection device 3 switches between the spectrum method and the addition method, and the subject gazes. The blinking frequency of the visual stimulus being performed can be accurately detected.
Further, as described above, neither the spectral method nor the addition method requires learning.
As described above, the frequency detection device 3 can obtain the output of the BCI using the SSVEP with higher accuracy without learning.

  In particular, the frequency detection device 3 can detect the blinking frequency of the visual stimulus more accurately using the spectral method when a peak is clearly shown in the spectral frequency. On the other hand, if the peak is not clearly shown in the spectrum frequency due to noise mixing or individual differences in the electroencephalogram detection signal, the frequency detection device 3 uses a noise-resistant addition method to more accurately detect the visual stimulus. The blinking frequency can be detected.

The inventor of the present application confirmed the effectiveness of the frequency detection method according to the present embodiment through experiments. Specifically, the inventor of the present application calculated the identification rate of the method combining the spectrum method and the addition method described above.
The electroencephalogram data acquired in the experiment described in the first embodiment was used, and a comparison with the LDA method was performed as in the experiment described in the first embodiment.

  FIG. 23 is an explanatory diagram showing experimental results for the third embodiment. In the figure, the visual recognition blinking frequencies of 6 Hz, 7 Hz, 8 Hz, and the average of these experimental results, the average value of the discrimination rate and the value of the risk factor p are shown. The “combination” column shows the average value of the discrimination rates by a method combining the spectrum method and the addition method. The column “LDA” indicates the identification rate by the LDA method. This figure shows the experimental results for each case where the data is divided every 3 seconds, divided every 4 seconds, or divided every 5 seconds.

In the experimental results shown in FIG. 23, the risk rate is 5% or less in many cases. Thereby, the superiority of the method combining the spectrum method and the addition method is shown.
In particular, when the data is divided every 5 seconds, the risk factor is 5% or less for all of the blinking frequencies of visual stimuli of 6 Hz, 7 Hz, and 8 Hz, indicating a strong advantage.

Further, in the third embodiment, the ratio of the result of the addition method (the ratio of S303 in FIG. 21 is No) than the ratio of the result of the spectrum method (the ratio of S303 in FIG. 21 is YES). It was confirmed that the method using the addition method is superior in the above-described predetermined conditions. On the other hand, even if the identification method has a low identification rate due to the effects of noise, etc., there may be cases where a good identification rate has been obtained by the spectrum method. The method combining the spectrum method and the addition method can be supplemented to both. Therefore, it is possible to detect the frequency with higher accuracy.
Further, as described in the first embodiment, it can be said that the method combining the spectrum method and the addition method does not require prior training and can be applied immediately.

Note that a program for realizing all or part of the functions of the control units 120, 220, and 320 is recorded on a computer-readable recording medium, and the program recorded on the recording medium is read into a computer system and executed. By doing so, you may process each part. Here, the “computer system” includes an OS and hardware such as peripheral devices.
Further, the “computer system” includes a homepage providing environment (or display environment) if a WWW system is used.
The “computer-readable recording medium” refers to a storage device such as a flexible medium, a magneto-optical disk, a portable medium such as a ROM and a CD-ROM, and a hard disk incorporated in a computer system. Furthermore, the “computer-readable recording medium” dynamically holds a program for a short time like a communication line when transmitting a program via a network such as the Internet or a communication line such as a telephone line. In this case, a volatile memory in a computer system serving as a server or a client in that case, and a program that holds a program for a certain period of time are also included. The program may be a program for realizing a part of the functions described above, and may be a program capable of realizing the functions described above in combination with a program already recorded in a computer system.

  The embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to this embodiment, and includes design changes and the like without departing from the gist of the present invention.

1, 2, 3 Frequency detection device 100, 200, 300 Frequency detection device main body 110 Signal input unit 120, 220, 320 Control unit 121 EEG measurement unit 122 Frequency spectrum acquisition unit 123 Maximum value acquisition unit 124 Median value acquisition unit 125 Spectral method Frequency detection unit 150 Storage unit 160 Result output unit 227 Periodic data cutout unit 228 Addition average calculation unit 229 Addition method frequency detection unit 326 Condition determination unit 900 Electrode

Claims (6)

An electroencephalogram measurement unit for measuring an electroencephalogram emitted from the primary visual cortex;
For each of a plurality of predetermined cycles, a cycle data cutting unit that cuts out cycle data starting from each time of the predetermined cycle from the brain wave,
An addition average calculating unit that adds the period data for each predetermined period;
Among the predetermined periods, an addition method frequency detector that detects a frequency corresponding to a period in which the difference between the positive and negative peak intensities of the data obtained by adding the period data for each predetermined period is maximum;
An electroencephalogram frequency detection apparatus comprising: An electroencephalogram measurement unit for measuring an electroencephalogram emitted from the primary visual cortex;
A frequency spectrum acquisition unit for acquiring a frequency spectrum of the brain wave;
For each of a plurality of predetermined frequencies, a maximum value acquisition unit that acquires a maximum value of the intensity of the frequency spectrum of the brain wave in a first predetermined frequency range including the predetermined frequency;
For each of the plurality of predetermined frequencies, a median value acquisition unit that acquires a median value of the intensity of the frequency spectrum of the electroencephalogram in a second predetermined frequency range including the predetermined frequency;
A spectral method frequency detector for detecting a frequency having a maximum difference between the maximum value and the median value among the plurality of predetermined frequencies;
An electroencephalogram frequency detection apparatus comprising: A condition determination unit that determines whether the frequency detected by the spectrum method frequency detection unit satisfies a predetermined condition;
When the condition determining unit determines that the frequency does not satisfy the predetermined condition, for each of a plurality of predetermined periods, period data for cutting out period data starting from each time of the predetermined period from the electroencephalogram A cut-out section;
Among the predetermined periods, an addition method frequency detector that detects a frequency corresponding to a period in which the difference between the positive and negative peak intensities of the data obtained by adding the period data for each predetermined period is maximum;
The brain wave frequency detection device according to claim 2, further comprising:   The electroencephalogram measurement unit measures the electroencephalogram at a plurality of positions based on signals from a plurality of electrodes that detect electroencephalograms emitted from the primary visual cortex at different positions. 4. The electroencephalogram frequency detection apparatus according to any one of items 3 to 4. A method for detecting a frequency of an electroencephalogram frequency detection device, comprising:
An electroencephalogram measurement step for measuring an electroencephalogram emitted from the primary visual cortex;
For each of a plurality of predetermined cycles, a cycle data cutting step of cutting out cycle data starting from each time of the predetermined cycle from the brain wave,
An addition average calculating step of adding the period data for each predetermined period;
An addition method frequency detection step for detecting a frequency corresponding to a cycle in which a difference between positive and negative peak intensities of data obtained by adding the cycle data for each predetermined cycle among the predetermined cycles is maximized;
A frequency detection method comprising: To the computer that controls the EEG frequency detector,
An electroencephalogram measurement step for measuring an electroencephalogram emitted from the primary visual cortex;
For each of a plurality of predetermined cycles, a cycle data cutting step of cutting out cycle data starting from each time of the predetermined cycle from the brain wave,
An addition average calculating step of adding the period data for each predetermined period;
An addition method frequency detection step for detecting a frequency corresponding to a cycle in which a difference between positive and negative peak intensities of data obtained by adding the cycle data for each predetermined cycle among the predetermined cycles is maximized;
A program for running

Images