JPS6247340A - Actual brain wave extraction method - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed description of the invention]

(a) Industrial application field This invention relates to a real brain wave extraction method. (b) Conventional technology Conventionally, electroencephalograms have been used to indicate the functional state of the brain and serve as an indicator of brain disorders. The electroencephalogram (EEG) is a record of electrical potential differences, and the electroencephalogram (EEG) is when the brain waves are guided to an intensifier and displayed using a pen recording device. As a derivation method, a reference electrode is placed on a part of the scalp that is considered to be relatively unaffected by brain waves, such as the earlobe, and an active electrode is placed on the part of the scalp to be detected, and brain potentials are derived. (c) Problems that the invention seeks to solveHowever,
In order to derive ideal real brain waves without interference, the active electrode should be placed in a location where the target brain waves can be easily detected, and the reference electrode, which should serve as the reference for membrane potential, should be placed in a location where the target brain waves can be easily detected. The ripples of noise spread

[It is essential to install the device in a location with a constant potential. However, in areas where unintended brain waves do not affect the brain, such as the hands and feet, the electrocardiogram level is higher than the membrane potential, making it extremely difficult to interpret liver and kidney brain waves. Also, it is difficult to decipher the target brain waves. Therefore, although unintended electroencephalograms and electrocardiograms are mixed in, the conventional unipolar derivation method for electroencephalograms is to place base wall electrodes on areas such as the earlobes, where these effects are considered to be relatively small. . However, the brain waves derived by placing a reference electrode on a lunar tube are mixed with noise waves as described above, so the target brain waves are masked by the noise waves and cannot accurately indicate brain functions. Therefore, it was impossible to perform more advanced analysis and diagnosis of brain function. Attempts have already been made to extract only the target real brain waves by removing single waves such as electrocardiograms and non-target brain waves using filters, etc., but these efforts have failed, and it is currently difficult to remove these interference waves. It was seen as impossible. (d) Means for solving the problem In this invention, electrodes are installed at positions on the body surface where it is easy to derive the desired brain waves, and from there, brain wave waveforms mixed with interference caused by electrocardiograms are derived, and on the other hand, separately 19 A trigger signal is generated from the electrocardiogram waveform, the electroencephalogram waveform is divided into individual segments starting from a point a predetermined period of time from the trigger signal, and each segment is sequentially added and averaged to obtain an estimated waveform of noise waves. This object is to provide a real brain wave extraction method characterized by extracting a real brain wave waveform by subtracting this estimated waveform from the original brain wave waveform mixed with interference while synchronizing the starting points. (e) Action and Effect In this invention, the electrocardiogram-derived waveform mixed with the brain waves is
At the same time as the electroencephalogram is derived, it is synchronized with the electrocardiogram derived separately.
Focusing on the fact that it is highly independent of the electroencephalogram waveform for testing purposes, the derived electroencephalogram is segmented using FI Ll synchronized with the electrocardiogram, and the successive segments are sequentially added and averaged in synchronization with the electrocardiogram. The waveform components that are synchronized with the electrocardiogram become apparent as their phases are superimposed, and the phases of asynchronous waveform components are dispersed and smoothed, so that only the synchronized waveforms are extracted.The target real brain wave is extracted from this synchronized waveform. fI of
It can be seen as a difference between the two. Therefore, real brain waves can be extracted by subtracting this synchronized waveform from the original derived brain wave waveform as an estimated noise waveform. As described above, since the waveform derived from the electrocardiogram can be erased, the reference electrode can be installed at a location where the influence of the electrocardiogram is large, as long as it is not contaminated with unintended electroencephalograms. The waveform thus obtained indicates the target brain wave with extremely high accuracy, and therefore has sufficient reliability as a resource for more advanced analysis and diagnosis of brain function. Embodiment If we describe the embodiment of this invention in detail based on the drawings, Figs. 1 to 4 are schematic diagrams for explanation, and Fig. 1 shows an electroencephalogram and a trigger signal. (1) shows an electrocardiogram that causes
This electroencephalogram is derived from the potential difference between the two electrodes by placing an active electrode (2) on the scalp, near the target area of L, and a reference electrode (3) on the chin or hand. The waveform contains a mixture of the target brain wave waveform and the waveform derived from the electrocardiogram, but does not contain any brain waves other than the target. Note that when setting the electrodes, it is not necessary to take into account the magnitude of the waveform level derived from the electrocardiogram, and in short, it is possible to select a position where the target brain waves can be most clearly derived. In addition, this waveform shows that the waveform periodicity derived from the electrocardiogram is not necessarily constant, and that the frequency bands of the target brainwave and noise waves to be removed are single and multiple, which has traditionally made it difficult to extract real brainwaves. I thought it was something. The trigger signal (1) amplifies the electrocardiogram waveform (4) to an appropriate level and generates an R wave (
5) is generated as a label. Note that the electrocardiogram waveform (4) includes a P wave (6) that precedes the R wave (5), and this preceding time is usually 200 m5 ecl.
It is said to be within X. Figure 2 shows the brain waveform <7> in Figure 1 as a trigger signal (1
) shows the synchronously divided waveform. Note that synchronous division here means temporal alignment of the electroencephalogram waveform (7) and the trigger signal (1), that is, the electrocardiogram waveform (4), and specifically, the electroencephalogram waveform (7>) is The in-phase J: of the next wave is divided into one segment (9) with the point (8) going back 200 m sec from (1), and one heliga signal (1
) The P wave (6) component of the electrocardiogram waveform (4) is located at the front of the segment (9) by setting 200 m5ec before the segment (8) as the starting point (8) of division. Figure 3 shows a waveform that has been sequentially averaged in synchronization with the segment (9) that was synchronously divided in Figure 2 and the starting point. Only the waveforms that are synchronized with the ECG are superimposed and become apparent, and the asynchronous waveforms are smoothed and only the synchronous waveforms are extracted. ). Note that this sequential averaging is performed by multiplying the past sequential averaging value by a weighting index, multiplying the newly arrived segment value by a value obtained by subtracting the same index from 1, and then adding both. , the estimated waveform (10) is updated sequentially to dynamically follow the interference waveform of the subject. Note that the weighting index can be set arbitrarily between 0 and 1. In this example, the above processing is performed for the active and reference electrodes (2
>, increase the potential difference between (3) to a maximum of about ±5V, △/
D conversion is performed and digital processing is performed, and the specifications of the Δ/D conversion are a sampling interval of 1 m5ec and a resolution of 12 bits. It should be noted that brain waves are relatively slow electrical oscillations within about 100 l-1z, and the derived potential difference is converted into a signal by dividing it into binary 12 bits, that is, 4096 steps. Sufficient accuracy is guaranteed. Then, the digital signals of brain potentials that arrive sequentially are stored in chronological order, and a series of digital signals obtained from a point (8) that is 200 m5ec back from the trigger signal (1) obtained from the electrocardiogram is defined as a segment (9).
Segments (9) that arrive one after another are synchronized with the starting point (8), and the latest estimated waveform (
10). The estimated waveform (10) is not updated after a series of digital signals forming the segment (9) are input, but is updated by segment ( 9) is sequentially updated starting from the part corresponding to the signal, thereby increasing the processing speed. Note that the main components of the waveform caused by the electrocardiogram are the R wave (5), which indicates the maximum peak, and the P wave (6), which precedes the R wave.
After the wave (5) is generated, it rapidly attenuates and the variation in the period increases, so in order to extract a repetitive waveform that is synchronized with the electrocardiogram waveform (4), it is necessary to extract 200 m5ec before the R wave that includes the P wave (6). By performing synchronous division and successive averaging using the starting point, sufficient accuracy can be obtained for this purpose. Furthermore, the travel time of the starting point (8) can be adjusted to accommodate individual differences among subjects. Figure 4 shows the target real brain wave waveform (11), and the estimated noise waveform (10) in Figure 3 is subtracted from the brain wave waveform (7) mixed with interference by synchronizing the starting point (8). The device configuration was carried out. That is, (2> and (3) indicate the active and reference electrodes, respectively, and they were attached to the skin of the subject's scalp and hand, respectively, with conductive glue containing a large amount of electrolyte. (12>
, (12)- are electrodes for deriving the electrocardiogram waveform (4), and are each attached to the skin of the subject other than the head. The potential difference between the active and reference electrodes (2>, (3)) is amplified to a maximum of about ±5 V by an amplifier (13) with a constant amplification rate, and converted into a binary signal by A/D conversion (14). It is converted into a 12-bit digital signal and stored in a temporary storage device (15).This Δ/D conversion is performed at a sampling interval i rr
1 S e C, and are sequentially stored in addresses of the temporary storage device (15) corresponding to the time series. On the other hand, the electrode (12) for deriving the electrocardiogram waveform (4) is connected via the amplifier (16) to the built-in Schmidts 1~element (17) and differential element (18), and the IC1~Riga signal generator (19).
It captures the rising edge of the electrocardiogram waveform (4) and R wave (5) when they cross the reference voltage from below, and generates the trigger signal (1).
) while synchronizing with temporary storage fa (15).
EEG waveform stored in (7) 200 m from the current time of memory
The memory of the device (15) is accessed from an address corresponding to a point in time 5 ec back, and output to the calculation unit (20). In addition, in order to generate one signal (1) by digital processing, the electrocardiogram waveform (4) is amplified and then A/D converted, and the values of the digital signals input sequentially are compared before and after, and the subsequent signal is It is also possible to issue the trigger signal (1) when the value of is equal to or smaller than the value of the previous signal. The arithmetic unit (20) has a built-in register (21), which stores the estimated waveform (10) of the noise waves that has been successively averaged, and stores the newly input temporarily stored waveform (10). An average is performed with the electroencephalogram waveform (7) from the device (15), and the latest average value, that is, the estimated noise waveform (10) is always stored. Then, the estimated waveform (10) stored in this register (21) is subtracted from the brain wave waveform (7) stored in the temporary storage device (15) and output to the D/A converter (22). Actual electroencephalogram waveform (11) was measured using a pen recording recorder (23) connected to (22).
The online processing is performed with a delay of 2QQmsec. Note that the arithmetic unit (20) has a built-in clock (20)', and all timings such as the sampling strobe are determined by dividing the pulse from the clock (20>'').
”, the entire device operates in synchronization and synchronization, and the trigger signal (1) is 200m5ec.
To go back in time, the temporary storage device (1) at the time of the generation of the signal (1)
This is obtained by starting the address from -200 with the relative address of address 5), and this retrograde time can be changed and set by operating the keyboard (24) connected to the calculation section. In addition, a computer (2) for diagnosis etc. is connected to the calculation unit (20).
5), and processing such as diagnosis can be performed using the program and database built into the computer (25). Note that the calculation unit (20) can be performed by loading a program for the above procedure into a general-purpose computer, and if it is a high-speed computer, it can scan the brain waves from several channels of active electrodes and process them simultaneously. However, to cover an even larger number of channels, increase 1]
This is realized by integrating the circuits from the converter (13) to the D/A converter (22) and arranging a large number of them in parallel, which is preferable from the viewpoint of equipment operation and handling.

[Brief explanation of drawings]

Figure 1 shows electroencephalogram and electrocardiogram Figure 2 shows the synchronously divided waveform. Figure 3 shows the sequentially averaged waveform. (Estimated waveform of noise) Figure 4 shows the real brain wave waveform. Figure 5 is a block diagram of the equipment configuration. (1) Nidoriger signal (4): Electrocardiogram waveform (7): Brain wave waveform (8): Starting point Kuh): Segment (10): Estimated waveform of noise (11): Real brain wave waveform Figure 1 ■ 9 Figure 2 Figure 3 Figure 4 Figure 5

Claims (1)

[Claims] 1) Place electrodes on the body surface where it is easy to derive the desired brain waves, and from there measure the brain wave waveform mixed with interference caused by the electrocardiogram and pulse (
7), generate a trigger signal (1) from the electrocardiogram waveform (4) obtained separately, and generate the trigger signal (1).
The electroencephalogram waveform (7) is divided into individual segments (9) using a point (8) a predetermined time back from
10) is obtained, and this estimated waveform (10) is subtracted from the original EEG waveform (7) mixed with interference while synchronizing the starting point (8) to extract the actual EEG waveform (11). Real brain wave extraction method.