JP2010051356A - System for detecting motion-related potential signal, headset used in the same and system unit using the same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a system for detecting a motion-related potential signal which detects accurately and in real-time a motion-related potential signal only by using electroencephalogram data acquired from electrodes arranged on the scalp without using an electromyogram, a headset used in the system, and a system unit using the system. <P>SOLUTION: The system includes a correlation value calculating section 1 that carries out frequency analysis of data D1 and D2 of electroencephalograms EEG1 and EEG2 acquired by different deriving methods or by different electrode layouts from a plurality of adjacent positions on the scalp and calculates a correlation value between the electroencephalograms EEG1 and EEG2 for each frequency. The system detects a motion-related potential signal on the basis of the correlation values of a predetermined frequency band. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention detects a motion-related potential signal generated according to a motion command in the brain, for example, a motion-related potential signal detection system widely applicable to a rehabilitation device, a brain computer interface, etc., a headset used therefor, and It relates to a system unit using it.

  When a human performs “exercise” that voluntarily contracts muscles, the “exercise command” that causes the exercise is issued from the cerebrum. This motor command is observed as a motor-related potential (MRCP) in the brain. The motion-related potential is induced when the user actually moves the body or recalls the motion without moving the body, and is generated for each body part in the vicinity of a specific motor area corresponding to the part. Since the generation of this motion-related potential is included as a signal component in the brain wave representing the electrical activity of the brain, the technology for detecting a motion-related potential signal (hereinafter referred to as a motion-related potential signal) from the brain wave is Applications in various fields such as rehabilitation devices and brain computer interfaces are expected.

  Considering its application form, it is preferable that the measurement of the electroencephalogram is non-invasive, and one of them is a method of recording the electrical activity of the nerve cells in the brain from the electrode on the scalp. However, various signals are included in the electroencephalogram measured by this method, and it is extremely difficult to detect a weak motion-related potential signal. Therefore, it is conceivable to use a conventional electroencephalogram analysis method.

For example, Patent Literature 1 discloses a technique for analyzing an electroencephalogram based on the presence or absence of a peak formed in the frequency region of the electroencephalogram using the electroencephalogram power spectrum. It is conceivable that the motion-related potential signal is detected by arranging an electrode in the vicinity of the target motor area and the presence or absence of a peak of the power spectrum.
JP-A-6-296597

Further, for example, Patent Document 2 discloses a technique for detecting an electroencephalogram activity by finding the interaction of each part from the electroencephalogram based on the fact that the activity of each part of the brain interacts. Specifically, the time-frequency region is divided into small regions, the brain activity pattern is statistically estimated from the correlation between the small regions and the correlation in the three-dimensional brain space, and the brain wave activity is detected by pattern matching I do. When applied to the detection of motion-related potential signals, it is possible to statistically process brain wave data during exercise, estimate the brain activity pattern based on the interaction of each part, and detect it by pattern matching with that pattern. Conceivable.
JP 2006-110234 A

Further, for example, Non-Patent Documents 1 to 3 disclose a study in which a coherence function between an electroencephalogram and an electromyogram is examined. Here, the coherence function is a method for examining a proportional relationship between two signals in the frequency domain. It is known that if the autospectrum at the frequency f of the electroencephalogram subjected to electroencephalogram and full wave rectification is set to X (f) and Y (f), respectively, and the coherence function is obtained for each frequency, significant coherence is observed in the β band. It has been. Therefore, it is conceivable to detect a motion-related potential signal by obtaining a coherence function of an electroencephalogram and an electromyogram.
“Electroencephalogram-electromyogram coherence and its application and limitations” Tatsuya Mima, Hiroshi Shibasaki, Clinical EEG January 2003 (Vol. 45, No. 1) p19-25 “EEG-EMG coherence and its clinical application” Tatsuya Mima, Clinical EEG August 2005 (Vol.47, No8) p479-485 “Effects of motor cortex-muscle activity correlations on stability of muscle strength regulation” Junichi Ushiba et al., 26th Biomechanism Academic Conference held at International University of Health and Welfare, October 22-23, 2005 Abstract of the lecture

  However, when the above prior art is used for detection of motion-related potentials, the following problems occur. As described in Patent Document 1, when the detection is performed based on the peak of the power spectrum, the motion-related potential signal is very weak, and the brain wave has various noises (brain signals due to external noise and other brain activities). Therefore, even if a peak is observed, it is not always caused by the target motion-related potential, and the detection accuracy is extremely low.

  In addition, as disclosed in Patent Document 2, the following problem arises when analyzing the correlation of the brain activity of each part in the brain. As described as property 2 in Patent Document 2, the pattern of EEG activity is always the same because it is greatly affected by the internal state of the brain (for example, the degree of attention, familiarity with the task, fatigue, etc.). Not necessarily. For this reason, there is a limit to the number of patterns that can be statistically derived, and the detection accuracy is lowered. Even if a significant number of patterns can be derived, many electroencephalogram data must be preprocessed for that purpose (see paragraph number 0076 of FIG. 10 and FIG. 10). Technologies that mechanically compensate for motor functions such as prosthetic hands, orthotics, and brain computer interfaces should guarantee control linked to motion commands, but this impairs real-time performance and lacks practicality.

  In addition, the techniques using the electroencephalogram and electromyogram coherence functions disclosed in Non-Patent Documents 1 to 3 require high electromyograms of the exercise site, although the detection accuracy is high. The detection of motion-related potential signals is expected to be widely used in the field of patients such as paralysis and limb amputation, but it is often difficult or impossible for such patients to collect electromyograms. It becomes difficult.

  Therefore, the present invention is used for a motion-related potential signal detection system capable of detecting a motion-related potential signal with high accuracy and in real time only from electroencephalogram data obtained from electrodes arranged on the scalp without using an electromyogram. An object is to propose a headset and a system unit using the headset.

  In order to facilitate understanding of the present invention, the research by the inventors and the technical idea underlying the present invention will be described first.

  FIG. 1 is an explanatory diagram for explaining the transmission of exercise commands and electroencephalogram / electromyogram. The motion command issued from the human cerebrum a descends the spinal cord b and promotes the contraction of the muscle c. Conventionally, many studies have been conducted on the transmission mechanism of motor commands by examining the electrical activity of brain activity and muscle activity during exercise from the electroencephalogram (EEG) and surface electromyogram (Surface EMG). One of them is a study on coherence between an electroencephalogram and an electromyogram as described in Non-Patent Documents 1 to 3 above.

  The inventors conducted the following experiment on electroencephalogram-electromyogram coherence. Electrodes are placed on the skin of the subject's foot and on the scalp near the motor area of the foot (near Cz in the international 10-20 method), and the electroencephalogram and full-wave rectified myoelectric waves for 180 seconds while moving the foot. Figures were collected. Then, a coherence function for each frequency f between the electroencephalogram and the electromyogram was obtained.

ここで、Ｘ（ｆ）とＹ（ｆ）は、周波数ｆ における信号ｘ（ｔ）とｙ（ｔ）のオートスペクトルを示しており、＊は共役複素数、・は内積、バーはアンサンブル平均である。コヒーレンス関数Ｃｘｙ（ｆ）は、両信号間の線形相関性を周波数領域で表現するものであり、二信号のクロススペクトルを規格化した形式で、その値は０から１の範囲にある。脳波―筋電図コヒーレンスでは、Ｘ（ｆ）を脳波の周波数ｆにおけるスペクトル、Ｙ（ｆ）を筋電図の周波数ｆにおけるスペクトルとする。 The coherence function is generally expressed by the following formula 1.

Here, X (f) and Y (f) indicate the auto spectrum of the signals x (t) and y (t) at the frequency f 1, * is a conjugate complex number, • is an inner product, and bar is an ensemble average. . The coherence function Cxy (f) expresses linear correlation between both signals in the frequency domain, and has a value in a range from 0 to 1 in a form in which a cross spectrum of two signals is normalized. In the electroencephalogram-electromyogram coherence, X (f) is the spectrum at the frequency f of the electroencephalogram, and Y (f) is the spectrum at the frequency f of the electromyogram.

  FIG. 2 is a graph of experimental results by the inventors. Line L is a 95% significant difference line. A significant coherence of 14-30 Hz (β band) was observed in about 1/3 to 1/2 healthy subjects. The same applies to movements of hands and other body parts. Although the mechanism of this phenomenon has not been fully elucidated, it is generally considered that a phenomenon like a transmission circuit is created by a neural circuit called the thalamus or basal ganglia deeper than the cerebral cortex. ing.

  The inventors have obtained an idea from this experiment that it can be applied to various fields by detecting a motion command, that is, a motion-related potential signal with high accuracy only from an electroencephalogram without using an electromyogram. It came to. Furthermore, the inventors have focused on the phenomenon that the motion-related potential is induced not only when actually exercising but also when recalling exercise, and by ensuring real-time characteristics, the motor function of the body is mechanically It is said that it can be widely applied to technologies that compensate for the above, such as prosthetic hands, orthoses, and brain computer interfaces.

  Accordingly, the inventors have conducted intensive research to achieve the above object, and have derived the technical idea that is the basis of the present invention. Hereinafter, for convenience of explanation, an explanation will be given by taking an exercise command of a foot as an example, but the same applies to a hand, a waist, and other body parts. FIG. 3 is an explanatory view illustrating the arrangement of electrodes on the scalp according to the basic principle of the present invention, and FIG. 4 is a partial cross-sectional view of AA ′ of the head. The region (activity source S) responsible for foot movement in the cerebral cortex is located inside the largest groove of the cerebrum. Therefore, the electrode group E is disposed in the vicinity of the activity source S on the scalp of the subject P (in the vicinity of Cz in the international 10-20 method). Then, the electrodes e1, E2, e5 of the electrode group E are selectively used, and the electroencephalogram data D1, D2 of the plurality of electroencephalograms EEG1, EEG2 are collected by a method in which the distribution shape of the spatial sensitivity distribution diagram is different from the adjacent positions. To do. The electroencephalogram data D1 and D2 are digital data representing the electroencephalograms EEG1 and EEG2 in the time domain, and can be collected using a digital electroencephalograph or the like. Here, the spatial sensitivity distribution diagram is a diagram showing the distribution of sensitivity on the scalp when an electrode is used as a sensor in a three-dimensional space, and the distribution shape is a contour line shape representing the sensitivity distribution. In order to make the distribution shapes greatly different, it is effective to use different derivation methods for the electroencephalogram data D1 and D2, or use electrodes arranged in different layouts.

  Using the brain wave data D1 and D2, the coherence function between the brain waves EEG1 and EEG2 is obtained for each frequency. At a predetermined frequency (in this case, a β band of about 14-30 Hz), it is limited to during exercise or recalling motion. A significant correlation is observed. This is the first event confirmed by the inventors.

  This mechanism is considered as follows. When collecting the electroencephalogram data D1 and D2 by arranging electrodes close to the activity source S, Case 1 has the same derivation method and electrode layout, Case 2 has a different derivation method, Case 3 Assuming that different electrode layouts are used. In the case 1, as shown in FIG. 5A, the electroencephalogram data D1 is collected by the bipolar derivation method using the electrodes e1 and e2 laid out in the vertical direction, and the electroencephalogram data D2 is obtained from the electrodes e3 and e4 having the same layout. Take the case of using the same bipolar derivation method. In the case 2, as shown in FIG. 5B, the electroencephalogram data D1 is collected by the bipolar derivation method using the electrodes e2 and e3 laid out diagonally, and the electroencephalogram data D2 is laid out at the center and the periphery thereof. The case where it collects by the Laplacian derivation method using e1-e5 is taken as an example. As shown in FIG. 5C, the case 3 uses a pair of electrodes e3 and e4 in which the electroencephalogram data D1 is laid out in the horizontal direction using the electrodes e3 and e4 in which the electroencephalogram data D1 is laid out in the vertical direction. Take as an example the case of sampling by the derivation method. In either case, the active source S is positioned immediately below the center of the electrode group.

  In both cases 1, 2, and 3, the motion-related potential signal emitted from the activity source S is included in each brain wave data D 1, D 2 in common. However, in case 1, since the brain wave data D1 and D2 are collected by the same derivation method and electrode layout, the distribution shapes of the spatial sensitivity distribution diagrams are almost the same. That is, since the electroencephalogram data is collected from the adjacent positions with almost the same sensitivity distribution, the electroencephalogram data D1 and D2 are almost the same, and the overall correlation including noise becomes high. On the other hand, in cases 2 and 3, since the brain wave data D1 and D2 are derived by different derivation methods and electrode layouts, the distribution shapes of the spatial sensitivity distribution diagrams are greatly different. Even if the sampling positions are close to each other, the distribution shapes are greatly different, so noise mixed in from various directions is greatly distorted and observed as different waveforms. For this reason, in cases 2 and 3, the correlation of noise is low, and only the motion-related potential signal from the activity source S is high.

  Therefore, the inventors have derived a technical idea that by using this correlation, a motion-related potential signal can be detected with high accuracy and in real time from only brain wave data without using an electromyogram. . The present invention has been completed from the results of such research.

  The motion-related potential signal detection system of the present invention is a system for detecting a specific motion-related potential signal included in an electroencephalogram using the electroencephalogram data collected from an electrode arranged on the scalp, and a plurality of adjacent electrokinetic potential signals. A different derivation method from the position of the above, or frequency analysis of the data of the electroencephalogram collected by different electrode layouts, and a correlation value calculation unit for calculating a correlation value between each electroencephalogram for each frequency, and in the predetermined frequency band A motion-related potential signal is detected based on the correlation value.

  According to the present invention, data of a plurality of electroencephalograms collected by different derivation methods or different electrode layouts from a plurality of adjacent positions are processed. As described above, between the electroencephalograms, the correlation of the motion-related potentials emitted from the activity source near the electrodes is high, and the correlation of noise mixed from many directions is low. Therefore, by using the electroencephalogram data, the correlation value calculation unit calculates the correlation value between the electroencephalograms for each frequency, and the signal of the motion related potential can be detected based on the correlation value.

  Furthermore, the headset of the present invention is a headset for collecting the electroencephalogram data used in the motion-related electroencephalogram potential detection system according to claim 1 in a wearing state, and at the position where the electroencephalogram data is collected. Correspondingly, a plurality of electrodes are arranged close to each other.

  According to the present invention, the electrode is disposed at a position suitable for collecting the electroencephalogram data simply by mounting the mounting tool.

  A system unit according to the present invention includes the motion-related potential signal detection system according to claim 1 and a control target system controlled according to a detection result of the motion-related potential signal detection system. The control target system is preferably an exercise assistance system having a function of assisting body movement or an operation command generation system having a function of generating an operation command.

  According to the present invention, the control target system such as the exercise assist system or the operation command generation system is controlled according to the detection result of the motion related potential signal detection system. Command).

  In the present invention, since the correlation value of the target motion-related potential signal is high and the correlation value of noise is low, the target motion-related potential signal can be detected with high accuracy. Since an electromyogram is unnecessary, it can be applied to paralyzed patients and patients with disabilities such as limb amputations, and can be widely applied to fields such as rehabilitation, nursing care, diagnosis and treatment. Since detection is based on correlation values in a predetermined frequency band, it can be detected by sequentially processing the collected brain wave data, eliminating the need for preprocessing a large amount of data for pattern analysis as in the prior art Real time can be secured. For example, the present invention is also extremely suitable for techniques for mechanically compensating for motor functions such as prosthetic hands, orthotics, and brain computer interfaces.

(First embodiment)
Hereinafter, specific embodiments to which the present invention is applied will be described in detail with reference to the drawings. The present embodiment is a motion-related potential signal detection system that detects a motion-related potential signal using brain wave data collected from an electrode placed on the scalp, and is highly accurate without using an electromyogram. It can be detected in real time.

(Example 1 of the first embodiment)
This embodiment is a motion-related potential signal detection system 101 for detecting a motion-related potential signal of one motion site. FIG. 6 is a block diagram conceptually illustrating the motion related potential signal detection system 101. The motion related potential signal detection system 101 includes a correlation value calculation unit 1 and a detection unit 2, and is realized by, for example, a microcomputer or a personal computer.

  The motion-related potential signal detection system 101 receives and processes brain wave data D1 and D2 of a plurality of brain waves EEG1 and EEG2. The electroencephalogram data D1 and D2 are digital data collected from an electrode placed on the scalp of the human P using a digital electroencephalograph or the like. The electroencephalograms EEG1 and EEG2 for a predetermined time are digitally converted and displayed in the time domain. It is a thing.

  The electroencephalogram data D1 and D2 are collected by different methods or different electrode layouts from positions close to each other in the vicinity of the target motor area. Here, the vicinity of the target motor area is the vicinity of the activity source S in the brain when a specific movement is performed or when the movement is recalled. For example, in the international 10-20 method, the foot is near Cz, the right hand is near C3, and the left hand is near C4.

  The derivation method is (A-1) a unipolar derivation method using a potential difference between a reference electrode arranged near the inactive site and a related electrode arranged on the scalp, and (A-2) two electrodes on the scalp. Bipolar derivation method using potential difference, (A-3) Laplacian derivation method using difference between potential of one electrode on scalp and surrounding electrode, (A-4) Multiple on scalp There are derivation methods (except for the Laplacian derivation method) using the potential difference between the average potential of the other electrode and the potentials of the other electrodes, and various other derivation methods such as IB2.

  Examples of the electrode layout include various patterns in which the number, arrangement, and shape of the electrodes are different. FIG. 7 is an explanatory diagram illustrating different electrode layouts by way of example. (B-1) is a pattern in which only one electrode is arranged. (B-2) is a plurality of electrodes having the same shape arranged vertically, (B-3) is arranged horizontally, (B-4) is arranged diagonally, and (B-5) is arranged at the vertex of a polygon. It is a pattern. (B-6) is a pattern in which a plurality of electrodes are arranged around the center electrode. (B-7) (B-8) uses a specially shaped electrode, (B-7) is a pattern in which a circular electrode is placed beside a crescent-shaped electrode, and (B-8) is a circle. This is a pattern in which a circular electrode is arranged at the center of an annular electrode. The electrode layout is a channel unit layout on the scalp. In the unipolar derivation method, a reference electrode is separately arranged at an inactive site (such as an earlobe).

FIG. 8 is an explanatory diagram for explaining an example of collection patterns of the electroencephalogram data D1 and D2 by the above derivation method and electrode layout. Each electrode group E1-6 is distribute | arranged to object motor area vicinity. The electrode groups E1 to E4 of (a) to (d) are examples in which different derivation methods are adopted. Here, the electrode layout is inevitably different, but it is important that the derivation method is different. This is because, if the derivation method is different, the distribution shape of the spatial sensitivity distribution diagram is greatly different. (E) The electrode groups E5 and E6 in (f) are the same derivation method, but are examples in which different electrode layouts are adopted. Hereinafter, the derivation methods (A-1) to (A-4) and electrode layouts (B-1) to (B-8) are shown in Table 1.

  In any of the electrode groups E1 to E6, the electroencephalogram data D1 and D2 are collected from close positions, but the distribution shapes of the spatial sensitivity distribution diagrams are different because of different derivation methods or different electrode layouts. As described above, between the electroencephalogram data D1 and D2, the correlation of the motion-related potential signal emitted from the activity source S is high, and the correlation of noise mixed from many directions is low.

  The correlation value calculation unit 1 performs frequency analysis using the brain wave data D1 and D2, and calculates a correlation value C (f) of the brain waves EEG1 and EEG2 for each frequency f. The correlation value C (f) may be any value that can normalize the correlation between the two signals. For example, a correlation function using coherence, wavelet coherence, or Hilbert transform can be used. Here, the correlation value calculation unit 1 will be described by taking as an example a case where the coherence function Cxy (f) is used. The coherence function Cxy (f) is the same as the above equation 1, X (f) is the spectrum of the electroencephalogram EEG1 at the frequency f, and Y (f) is the spectrum of the electroencephalogram EEG2 at the frequency f.

  FIG. 9 is a block diagram conceptually illustrating the correlation value calculation unit 1. The electroencephalogram data D1 and D2 are electroencephalogram data in the time domain of the electroencephalograms EEG1 and EEG2. When the correlation value calculation unit 1 receives brain wave data D1 and D2 for a predetermined time (for example, 1 s), the segment division unit 1a divides each brain wave data D1 and D2 for each predetermined time (for example, 500 ms). The segment data Dx-t is created. Here, Dx is D1 or D2, and t is a segment number in time series order.

  The created segment data is supplied to the frequency analysis unit 1b. The frequency analysis unit 1b uses FFT or the like, performs frequency analysis on the segment data Dx-t, and creates frequency domain data Dx-t-f. The frequency domain data Dx-t-f is data representing the amplitude and phase of the frequency f in the segment t. Here, f is each frequency f with a predetermined resolution in a predetermined frequency band. In the present embodiment, each frequency f at a resolution of 1 Hz in the frequency band (β band: 14 to 30 Hz) of the motion-related potential signal is used, but any frequency range that includes at least a part of the frequency band of the motion-related potential signal. Is optional.

  Each frequency domain data Dx-t-f is supplied to the calculation unit 1c, and a coherence function Cxy (f) expressed by Equation 1 is obtained for each frequency f. That is, the spectrums s1 and s2 of the brain waves EEG1 and EEG2 are compared in time series for each frequency f. When the amplitude and phase are common over all the segments t (1) to t (m), 1 is irrelevant. In this case, a normalized correlation value C (f) is calculated so as to be 0. Note that the standardization rules are stored in the storage unit, and may be calculated according to the rules. In calculating the correlation value C (f), both the amplitude and phase of the frequency spectrum may be used, or only one of them may be used, or the intensity of the power spectrum may be used.

  The correlation value C (f) of each frequency f calculated in this way is supplied to the detection unit 2. The detection unit 2 detects the motion-related potential signal based on the correlation value C (f) according to a specified rule. Although the rule is arbitrary, it is preferable that it is a threshold determination. For example, when any one of the correlation values C (f) of each frequency exceeds the threshold value Z1, the motion-related potential signal may be present, and the number of correlation values C (f) exceeding the threshold value Z1 is the threshold value Z2. In the case of the above, there may be a motion-related potential signal. In addition, when the correlation value calculation unit 1 calculates a correlation value C in a wide area (for example, 1 Hz to 50 Hz) including the β band, the presence or absence of a peak in the β band may be used.

  Information of the detection result is supplied to a processing unit that performs subsequent processing or an external device. Examples of the external device include a display device that displays a determination result, a prosthetic hand, a wheelchair, a rehabilitation device, and a device that compensates for a motor function such as a brain computer interface. This makes it possible to realize various devices using exercise commands.

(Example 2 of the first embodiment)
The present embodiment is a motion-related potential signal detection system 102 suitable for detecting motion-related potential signals related to motion at a plurality of sites. FIG. 10 is a block diagram conceptually illustrating the motion related potential signal detection system 102. Here, for convenience of explanation, a case will be described as an example where the motion-related potential signals of the left foot (LF), right foot (RF), left hand (LA), and right hand (RA) are detected. In the international 10-20 method, the motor area is located near the position Cz for the right and left legs, the position C3 for the right hand, and the position C4 for the left hand. Therefore, each electrode group E is arranged in the vicinity of each motor area, and from the electrode group E (RF, LF) near the position Cz, the electroencephalogram data D1 (RF) and D2 (RF) related to the right foot and the electroencephalogram data D1 related to the left foot. (LF), D2 (LF), right-hand related electroencephalogram data D1 (RA), D2 (RA) from the electrode group E (RA) near the position C3, and left-hand related electroencephalogram data D1 (LA), from the position C4. D2 (LA) is collected.

  Here, the electroencephalogram data D1 (LF) and D2 (LF) related to the left foot and the electroencephalogram data D1 (RF) and D2 (RF) related to the right foot are collected from the same position Cz, but the electrode e used The combination is different. FIG. 11 shows the combinations. The right foot channel electrode has a combination with high sensitivity near the left region near the position Cz, and the left foot channel electrode has a combination with high sensitivity near the right region near the position Cz. For example, the electroencephalogram data D2 (RF) and D2 (LF) are the difference between the average potential of the electrodes e2 to d5 and the potential of the electrode e1, while the electroencephalogram data D1 (RF) of the right foot is the difference between the electrodes e2 and e3. The electroencephalogram data D1 (LF) of the left foot is collected as a difference between the electrode e4 and the electrode e3.

  The motion-related potential signal detection system 102 processes the electroencephalogram data D1 and D2 independently and in parallel for each part of the right foot (RF), left foot (LF), right hand (RA), and left hand (LA). That is, each detection target includes processing units 3, 3, and 3 (correlation value calculation unit 1 and detection unit 2) having substantially the same functions as those of the motion-related potential signal detection system 101 of the above embodiment. Thereby, it becomes possible to distinguish and detect the motion-related potential signals of different parts (RF, LF, RA, LA) in parallel. In particular, the left foot and the right foot can be detected by distinguishing the right foot and the left foot even if they are collected from the same position Cz using the common electrode group E (RF, LF).

(Second Embodiment)
This embodiment is a headset 200 that is used in combination with the motion-related potential signal detection system 102. FIG. 12 is an explanatory diagram conceptually illustrating the headset 200 according to the present embodiment. The headset 200 is used by being worn on the head, and an electrode group E is arranged corresponding to a position where desired brain wave data D1 and D2 are collected in the worn state. In this example, an electrode group E (RF, LF), an electrode group E (LA), and an electrode group E (RA) corresponding to the extremities (RF, LF, RA, LA) are arranged. In each electrode group E, a plurality of electrodes e,. As shown in FIG. 8, the electrodes e,... E of each electrode group E need only be arranged so that the derivation methods of the electroencephalogram data D1 and D2 are different or the electrode layouts are different. The electrode e0 in the vicinity of the left and right earlobe is a reference electrode when unipolar derivation is adopted.

  The headset 200 is used by being connected to the electroencephalogram measurement unit 10. The electroencephalogram measurement unit 10 is a digital electroencephalogram measurement device or the like, which measures the electroencephalograms EGG1 and EGG2 for each part using the potentials of the electrodes e,. It has the function to generate. The electroencephalograms EEG1 and EEG2 are amplified by the amplification unit 10a, converted into digital data by the A / D conversion unit 10b, and stored and stored in the storage unit 10c as electroencephalogram data.

  The electroencephalogram data of each part (RF, LF, RA, LA) accumulated and stored in the electroencephalogram measurement unit 10 is read out for a predetermined time at a predetermined sampling period, and motion-related potentials are obtained as electroencephalogram data D1, D2 in the time domain. Input to the system 102. Note that the sampling period is preferably 10 ms to 20 ms in order to ensure real-time performance.

  FIG. 13 shows an example of the headset 200, where (a) is a perspective view and (b) is a top view. The headset 200 has an electrode group E (LF, RF) at a position corresponding to the vicinity of the position Cz on the scalp, an electrode group E (RA) at a position corresponding to the vicinity of the position C3, and an electrode group at a position corresponding to the vicinity of the position C4. E (LA) is arranged. In any of the electrode groups E, the electrodes e are arranged close to each other in the layout shown in the electrode group E1 in FIG. 8A, and the electroencephalogram can be derived by the Laplacian derivation method and the bipolar derivation method.

  According to this headset 200, simply by putting the headset 200 on the head, electrodes are arranged at a desired position, and the electroencephalogram data D1 and D2 suitable for the processing of the motion-related potential signal detection system 102 are easily collected. be able to. In addition, when there is one target part, one electrode group E is used and combined with the motion related potential detection system 101.

(Third embodiment)
The present embodiment is a system unit that includes a motion-related potential signal detection system. The motion-related potential detection systems 101 and 102 are used in combination with a control target system to be controlled, and constitute a system unit corresponding to a desired purpose. Examples of the operation target system include an exercise assisting device, an arithmetic processing device, a prosthetic hand, a wheelchair, a game device, and a display device. Hereinafter, as an example of a control target system, an explanation will be given of an exercise assisting device that assists body movement and an operation command generation system having a function of generating an operation command.

(Example 1 of the third embodiment)
The system unit 301 of the present embodiment is a system that assists body movement. In the present embodiment, for convenience of explanation, a system unit 301 suitable for rehabilitation of a left hand paralyzed patient will be described. In a hand paralyzed patient, muscle activity predominating in the flexor group is recognized in the wrist joint. Therefore, in this system unit 301, rehabilitation is performed in which the wrist joint is bent and extended in conjunction with a motion command of the hand.

  FIG. 14 is an explanatory diagram for explaining the system unit 301. The system unit 301 includes an exercise-related potential signal detection system 101 and an exercise assistance system 400. The same constituent elements as those in the above embodiment are denoted by the same reference numerals, and description thereof is omitted.

  The motion-related potential signal detection system 101 is almost the same as the above embodiment. The electrode group E (LA) is arranged in the vicinity of the motor area C4 of the left hand. The electroencephalogram data D <b> 1 and D <b> 2 are processed almost simultaneously with the brain activity, and information on detection results is output from the motion-related potential signal detection system 101 in real time. Information on the detection result is converted by the signal conversion unit 20 into a control signal for controlling the exercise assisting system 400 based on a predetermined conversion rule. The control signal is, for example, a command signal for rotating the motor in the reverse direction.

  The exercise assisting system 400 is a device that assists the exercise of the target body part. In the present embodiment, the exercise assisting system 400 includes a bending / extending unit 4a that performs a bending / extending operation by rotation of a built-in motor. When the motor assisting system 400 rotates the motor according to the control signal from the signal converting unit 20, the bending / extending unit 4a connected via the rotating shaft 4b swings and bends or extends.

  The system unit 301 preferably further includes a headset 201. The left-hand electrode group E (LA) is arranged in the headset 201, and the collected electroencephalogram data D 1 and D 2 are input to the motion related potential signal detection system 101 via the electroencephalogram measurement unit 10.

  During rehabilitation, the vicinity of the wrist joint of the patient P is fixed to the bending / extending portion 4a by a fixing portion such as a belt. Thereby, according to the motion command of the hand in the brain of the patient P, the flexion / extension motion of the wrist joint is assisted substantially in real time, and rehabilitation linked to the motion command in the brain is performed.

  In addition, as an exercise assistance system, it is not limited to the above, and may assist any exercise for any purpose, such as an assisting exercise of a foot.

(Example 2 of the third embodiment)
The system unit 302 of the present embodiment is an example applied to a so-called brain computer interface, and in particular, controls an arithmetic processing system mounted on a game device or a personal computer to operate image display on a display device. is there.

  FIG. 15 is an explanatory diagram for explaining the system unit 302. The system unit 302 includes an exercise-related potential signal detection system 102 and an operation command generation system 500. The same constituent elements as those in the above embodiment are denoted by the same reference numerals, and description thereof is omitted.

  The motion-related potential signal detection system 102 is substantially the same as the above embodiment. The electroencephalogram data D1x, D2x (x = RA, LA, RF, LF) of each part of the limb are processed in parallel, and the detection result for each part is calculated in real time. Information on the detection result is supplied to the operation command generation system 500.

  The operation command generation system 500 has a function of generating a predetermined operation command according to the determination result, and is realized by, for example, an arithmetic processing unit or a storage unit built in a game device or a personal computer. The storage unit stores a lookup table in which the electroencephalogram data D1x and D2x (x = RA, LA, RF, and LF) of each motion part are associated with the operation command (scan code). For example, the lookup table defines the association between the detection result of each limb motor potential signal and the operation command for operating the limb of the character in the game or virtual space. When the operation command generation system 500 receives the detection result information, the operation command generation system 500 refers to the lookup table, and performs an operation corresponding to the electroencephalogram data D1x, D2x (x = RA, LA, RF, LF) in which the motion-related potential signal is detected. Generate a command. The operation command is supplied to a processing unit that performs predetermined processing according to the operation command, and an image according to the operation command is displayed on the display unit 30.

  When using this system unit 302, the operator actually exercises or recalls the exercise. For example, when moving the right hand of the character Ch, if the right hand is moved or recalled, the motion-related potential signal detection system 102 detects the motion-related potential signal from the electroencephalogram data D1 (RA), D2 (RA), Based on the detection result, the operation command generation system 500 generates an operation command, and the display unit 30 moves the right hand of the character Ch. Therefore, the character Ch can be operated in conjunction with the motion command in the operator's brain.

(Example 3 of the third embodiment)
The system unit 303 of this embodiment is obtained by adding a mode switching function to the system unit 302 described above. Myoelectric signals generated by movements near the head, such as conscious blinks, teething, facial muscle activity, and neck tilt, are also observed on the electrodes on the scalp. Since the myoelectric signal has characteristics different from those of the electroencephalogram, it can be easily distinguished from the electroencephalogram. Therefore, the control mode is switched using the myoelectric signal.

  For example, a conscious blink myoelectric signal is mixed as large noise (a large amplitude in the vicinity of about 3 Hz) into the electroencephalogram data in a wide range on the scalp. Therefore, the system unit 303 according to the present embodiment uses the brain wave data D1x and D2x (x = RA, LA, RF, and LF) of the extremities to generate a predetermined frequency band (here, all brain waves). If an amplitude equal to or greater than the threshold value Z3 is observed at 3 Hz), the operation mode is switched.

  FIG. 16 is an explanatory diagram for conceptually explaining the system unit 303. A monitoring unit 40 that monitors generation of myoelectric signals due to conscious blinks and a mode switching unit 50 that performs mode switching are provided. The monitoring unit 40 monitors the electroencephalogram data D1x and D2x from the electroencephalogram measurement unit 10. When the amplitude of the predetermined frequency (3 Hz) exceeds the threshold value Z3 in the frequency domain, the supply of the electroencephalogram data D1x and D2x to the motion-related potential signal detection system 102 is temporarily stopped, and the mode switching unit 50 performs mode switching. Command signal is output.

  The mode switching unit 50 switches modes according to the command signal from the monitoring unit 40. Specifically, lookup tables T1 and T2 of different modes are stored in the storage unit, and the lookup tables T1 and T2 referred to by the operation command generation system 500 are switched. For example, the look-up table T1 describes an operation command for moving the limbs of the character Ch corresponding to the detection result, and the look-up table T2 describes an operation command for moving the background vertically and horizontally.

  When the mode switching by the mode switching unit 50 is completed, the monitoring unit 40 cancels the supply stop of the electroencephalogram data D1x and D2x. Thereby, subsequent processing is performed in a state where the mode is switched.

  In the conventional brain computer interface, it is necessary to collect myoelectric signals by arranging electrodes on the skin of the body when using body motion together. According to this mode switching function, since electrodes on the scalp arranged for brain waves are used, there is no need to separately arrange electrodes on the skin of the body, and operation variations can be increased only with electrodes on the scalp. it can. Note that mode switching is not limited to this, and may be, for example, movement of a cursor or power ON / OFF.

Example 1
The following experiments 1 to 4 were performed, and the effectiveness of the present invention was confirmed in comparison with a method using a conventional electroencephalogram-electromyogram coherence. There are two subjects.

  Experiment 1: A conventional electroencephalogram-electromyogram coherence experiment was performed. The coherence function between the electroencephalogram and the electromyogram when the left foot was moved was determined by placing electrodes on the skin of the left foot and Cz near the motor area of the foot on the scalp.

Experiments 2-3: An electrode group E was placed on the scalp of each subject. The location is the vicinity Cz of the motor area of the foot. With the layout of the electrode group E shown in FIG. 3, the electroencephalogram data D1 was collected by the bipolar derivation method, and the electroencephalogram data D2 was collected by the Laplacian derivation method. The bipolar derivation method is the difference between the potentials of the electrodes e3 and e4, and the Laplacian derivation method is the difference between the potential of the electrode e1 and the average potential of the electrodes e2 to e5. The subject was put into the states of the following experiments 2 to 4, and the coherence function in the frequency domain of the brain waves D1 and D2 was obtained.
Experiment 2: The left foot was moved.
Experiment 3: Recollected without moving the left foot.
Experiment 4: There was no movement or recall, and the patient was resting (control experiment).

  FIG. 17 is a graph of the results of Experiments 1 to 4. Subject 1 and 2 are data of two different subjects, (1-a) and (2-a) are Experiment 1, (1-b) and (2-b) are Experiment 2, and (1-c) (2- c) is the result of Experiment 3, and (1-d) and (2-d) are the results of Experiment 4. In each experiment, a plurality of trials EXfoot 1 to 6 were performed under the same conditions, and a coherence function was plotted for each trial on the graph. The plot shown in green is the average of the coherence function for each trial.

  In Experiment 1 (1-a) (1-b), a clear peak is observed in the frequency band of 14 to 30 Hz, and it is confirmed that the frequency of the motion-related brain potential signal when the foot is moved is in the β band. It was done. In Experiment 2 (1-b) (2-b) and Experiment 3 (1-c) (2-c), a clear peak was observed in a frequency band approximately similar to Experiment 1, so Experiments 2 and 3 It was confirmed that the motion-related potential signal can also be detected by this method. Further, it has been known in the conventional research that the method using the electroencephalogram-electromyogram coherence in Experiment 1 can detect the motion-related potential signal with high accuracy. In the methods of Experiments 2 and 3, it was confirmed that detection can be performed with high accuracy as much as that of the method of Experiment 1 without using an electromyogram.

  In the above embodiment, the case of Laplacian derivation and bipolar derivation has been illustrated, but the effectiveness is confirmed in the same manner even when other derivation methods are used, and different electrode layouts can be obtained even with the same derivation method. The same effectiveness was confirmed by doing (for example, FIGS. 8E and 8F).

(Example 2)
Experiments by the following analysis methods 1 and 2 were performed, and the effectiveness of the present invention was confirmed in comparison with a method using a conventional power spectrum. There is one test subject.

  Electrode group E was placed on the scalp of the subject. The location is the vicinity Cz of the motor area of the foot. As shown in FIG. 8 (a), the electrode has an E1 layout, and the subject was allowed to repeatedly perform a trial of 4 seconds of motion image and 9 seconds of rest on the left and right feet 30 times alternately (Phasic task). . For each trial, brain wave data D1 (RF) and D2 (RF) for the right foot and brain wave data D1 (LF) and D2 (LF) for the left foot were collected. As shown in FIG. 11, the electroencephalogram data D1 (RF) for the right foot is the difference between the potentials of the electrodes e2 and e3 (bipolar derivation method), and the electroencephalogram data D1 (LF) for the left foot is the electrode e4. The potential difference of the electrode e3 (bipolar derivation method) and the electroencephalogram data D2 (RF) (LF) were collected by the difference between the potential of the electrode e1 and the average potential of the electrodes e2 to e5 (Laplacian derivation method).

Using the electroencephalogram data D1 (RF) and D2 (RF) for the right foot and the electroencephalogram data D1 (LF) and D2 (LF) for the left foot, the electroencephalogram data was analyzed for each leg by the following analysis method.
Analysis method 1: The intensity of the power spectrum in the time frequency domain was examined.
Analysis method 2: Based on the principle of the present invention, the coherence function of the power spectrum between the brain waves EEG1 and EEG2 was examined.
The FFT window width was 1 second, the hamming window, 94% overlap, and coherence were added and averaged in the trial direction.

  FIG. 18 shows the result of the analysis method 1 and shows the intensity distribution of the power spectrum in the time frequency domain. FIG. 19 shows the result of the analysis method 2 and shows the distribution of the coherence function in the time frequency domain. Channel 1 is data of electroencephalogram data D1 (RF) and D2 (RF), and Channel 2 is data of electroencephalogram data D1 (LF) and D2 (LF). The time zone indicated by the arrows of LeftIM and RightIM shows the time zone in the recall state (4 seconds), and before and after that shows the time zone in the resting state (9 seconds in total).

  As shown in FIG. 18, depending on the analysis method 1, a characteristic change was not observed in any frequency band in the time zones LeftIM and RightIM reminiscent of the left and right foot movements.

  As shown in FIG. 19, according to the analysis method 2, in Channel 1, a high correlation was observed in the β band in the time zone of RightIM (arrow part). That is, when recalling the movement of the right foot, a high correlation is observed in the β band of the electroencephalogram data D1 (RF) and D2 (RF), and it is confirmed that a motion-related potential signal related to the right foot can be detected by using this correlation. It was done. Similarly, in Channel 2, a high correlation was observed in the β band in the LeftIM time zone (arrow points). That is, when recalling the movement of the left foot, a high correlation is observed in the β band of the electroencephalogram data D1 (LF) and D2 (LF), and it is confirmed that a motion-related potential signal related to the left foot can be detected by using this correlation. It was done.

  Furthermore, in Channel 1 the correlation between the right and high correlations of RightIM is low, in Channel 2 the correlation remains low. Similarly, in the channel of Left IM where the high correlation is confirmed in Channel 2, the correlation remains low in Channel 1. It was. That is, using the same electrode group E (RF, LF) arranged in the vicinity of the foot motor field Cz, it was confirmed that the motion-related potential signals of the right foot and the left foot can be distinguished and detected.

  From the above, it has been confirmed that even if the conventional method using the power spectrum intensity cannot be detected, it can be detected according to the present invention, and the motion-related potential signal can be detected with high accuracy. Further, according to the present invention, it was confirmed that the motion-related potential signals of the right foot and the left foot can be detected separately.

  As described above, according to the present invention, it is possible to detect a motion-related potential signal with high accuracy and in real time only from electroencephalogram data collected from an electrode on the scalp without using an electromyogram, rehabilitation, diagnosis / treatment, It can be widely applied to various fields such as brain computer interface.

  The present invention is not limited to the embodiment described above. For example, in the above embodiment, the correlation value of the two electroencephalogram data D1 and D2 is used, but the correlation value of three or more electroencephalogram data can also be used.

  In addition, the collection position of the electroencephalogram data is set to the vicinity of Cz, C3, and C4 in the international 10-20 method, but may be any position where a motion-related potential signal for detection purposes can be obtained.

  In the above-described embodiment, the combination with the exercise assist system that performs rehabilitation of the wrist joint and the combination with the operation command generation system have been described as examples of the system unit, but in addition, a wheelchair, an artificial hand, an artificial leg, a game It can be combined with various systems such as devices. Moreover, as an exercise assistance system, it is applicable also to the system which assists the exercise | movement of various body parts, such as an upper arm, an ankle joint, an upper limb, a waist, and a neck. Further, the correlation value and the detection result can be displayed on the display device without using the detection result to control the external system, which can be used for diagnosis and treatment.

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

It is explanatory drawing explaining transmission of a motion command, and an electroencephalogram / electromyogram. It is the figure which made a graph the experimental result of the electroencephalogram-electromyogram coherence by the inventors. It is explanatory drawing which illustrates arrangement | positioning of the electrode on a scalp in the basic principle of this invention. FIG. 4 is a partial cross-sectional view taken along line AA ′ of the head shown in FIG. 3. It is explanatory drawing explaining the mechanism of the basic principle of this invention. It is a block diagram which illustrates notionally the motion related potential signal detection system shown as Example 1 of 1st Embodiment. In the said embodiment, it is explanatory drawing explaining the pattern of a different electrode layout exemplarily. In the said embodiment, it is explanatory drawing explaining the example of the collection pattern of the electroencephalogram data by a different derivation method and an electrode layout It is a block diagram which illustrates notionally the correlation value calculation part shown as the said embodiment. It is a block diagram which illustrates notionally the motion related potential signal detection system shown as Example 2 of 1st Embodiment. It is explanatory drawing explaining the collection method of the brain-related electroencephalogram data in the said embodiment. It is explanatory drawing which illustrates notionally the headset shown as 2nd Embodiment. It is an example of the headset shown as the said embodiment, (a) is a front view, (b) is a top view. It is explanatory drawing which illustrates notionally the system unit shown as Example 1 of 3rd Embodiment. It is explanatory drawing which illustrates notionally the system unit shown as Example 2 of 3rd Embodiment. It is explanatory drawing which illustrates notionally the system unit shown as Example 3 of 3rd Embodiment. It is the figure which made the result of Experiments 1-4 in Example 1 into a graph. FIG. 10 is a diagram showing an experimental result by the analysis method 1 in Example 2. It is a figure which shows the experimental result by the analysis method 2 in Example 2. FIG.

Explanation of symbols

P human, subject e, e1 to e5 electrode E, E1 to E6, E (RF, LF, RA, LA) electrode group S activity source EEG1, EEG2 electroencephalogram D1, D2 electroencephalogram data 101, 102, 103 motion-related potential signal detection System 1 Correlation value calculation unit 1a Segment division unit, 1b Frequency analysis unit, 1c calculation unit 2 Detection unit 3 Processing unit 200 Headset 10 EEG measurement unit 301 to 303 System unit 400 Exercise support system 20 Signal conversion unit 500 Operation command generation system 30 display device 40 monitoring unit 50 mode switching unit

Claims (4)

A system for detecting a specific motion-related potential signal contained in an electroencephalogram using electroencephalogram data collected from an electrode disposed on the scalp,
A different derivation method from a plurality of adjacent positions, or a frequency analysis of the data of the brain waves collected by different electrode layouts, and a correlation value calculation unit that calculates a correlation value between each brain wave for each frequency,
A motion-related potential signal detection system that detects a motion-related potential signal based on the correlation value in a predetermined frequency band.   A headset for collecting the electroencephalogram data used in the motion-related electroencephalogram potential detection system according to claim 1, wherein a plurality of electrodes are arranged close to each other corresponding to the position where the electroencephalogram data is collected. Headset characterized by.   A system unit comprising: the motion-related potential signal detection system according to claim 1; and a control target system controlled according to a detection result of the motion-related potential signal detection system.   4. The system unit according to claim 3, wherein the control target system is an exercise assistance system having a function of assisting body movement or an operation command generation system having a function of generating an operation command.

Images