US3848586A

US3848586A - Signal detection system

Info

Publication number: US3848586A
Application number: US00316019A
Authority: US
Inventors: T Suzuki; T Ogawa
Original assignee: Hitachi Ltd
Current assignee: Hitachi Ltd
Priority date: 1971-12-17
Filing date: 1972-12-18
Publication date: 1974-11-19
Anticipated expiration: 1991-11-19
Also published as: JPS526039B2; JPS4868092A

Abstract

A signal detection system comprising detecting means for detecting the differential between a first and a second sampled value of an electrical signal subjected to sampling, comparing means for comparing the output of the detecting means with a predetermined first threshold value, discriminating means for discriminating the polarity of the output of the detecting means, adding means controlled by the output of the discriminating means for adding a predetermined second threshold value to the first sampled value, and means controlled by the output of the comparing means for deriving the result of addition from the adding means. The signal detection system serves for reducing or removing noises involved in or superposed on an electrical signal and more particularly a bioelectrical signal such as one recorded on an electroencephalogram or electrocardiogram.

Description

limited States Patent 1191 Suzuki et a1.

SIGNAL DETECTION SYSTEM Inventors: Takaji Suzuki, Kokubunji; Toshio Ogawa, Hachioji, both of Japan [30] Foreign Application Priority Data Dec. 17, 1971 Japan 46-101896 1 Nov. 19,1974

3,716,849 2/1973 Metcalf 235/183 3,745,463 7/1973 Klein 235/183 Primary Examiner-William E. Kamm Attorney, Agent, or Firm-Craig & Antonelli 5 7] ABSTRACT A signal detection system comprising detecting means for detecting the differential between a first and a second sampled value of an electrical signal subjected to sampling, comparing means for comparing the output of the detecting means with a predetermined first threshold value, discriminating means for discriminating the polarity of the output of the detecting means, adding means controlled by the output of the discriminating means for adding a predetermined second threshold value to the first sampled value, and means controlled by the output of the comparing means for deriving the result of addition from the adding means.

[56] References C'ted The signal detection system serves for reducing or re- UNITED STATES PATENTS moving noises involved in or superposed on an electri- 3,310,751 3/1967 Atzenbeck 328/151 cal signal and more particularly a bioelectrical signal 3,311,836 3/ 1967 DToro such as one recorded on an electroencephalogram or 3,506,813 4/1970 Trimble electrocardiogram, 3,590,811 7/1971 Harris 3,606,882 9/1971 Abe et a1. 128/206 A 7 Claims, 28 Drawlng Flgures DATA HOLDING JP SELECTOR ClRCUlT I ADDER S1 GATE THI5" n,

2 32* HOLDING 1* GATE CIRCUIT l6 5 6 8 1 20 I 4 x5 ABSOLUTE, r 7 M Tl I DATA x3 COMPARATOR SELECTOR x3 E H VIBRATOR I 3, HOLDING ,9 H6 51 ,11 53* CIRCUIT COMPARATOR MULT" 1 4 VIBRATOR L 14 MULTI- I X5 VIBRATOR PATENIE IIIIII I 9 I974 3,848,586 SHEET 20F 9 SI 82 S3 5 A IL I\ SL \LrfOR SHIFT SHIFT SHIFT IN I IST R s T iI' Iz ET A'EE E$AG INHIBIT SSSI s: S 2 2 A M Fl (5 6c 54 SI S2 83 II II l\ up SHIFT SHIFT SHIFT SHIFT {REGISTER REGISTER REGISTER REGISTER \J STAGE STAGE STAGE STAGE I s I T INHIBIT l S4S| S2 83 l TO DATA SELECTOR l8 E TO ADDER I? 52* HOLDING CIRCUIT 5 6 2 2 I I 7 ABSOLUTE 83- VALUE COMPARATOR (DETECTOR) 25 HOLDING TH6 SI CIRCUIT PATENI zzuv 1 91974 SHEET 50F 9 Fl G 7c F I G 7c H H C'I Fl 0 7b F G 8 X3 X4 X5 F l G 9 MUSCLEACTION POTENTIAL gg f CORRECTION PATTERN PATTERN A O O O O O LI EAR PATTERN B (I): I g MIDDLE I I I I POINT PATTERN c 0 O O LINEAR PATTERN 0 O O O 0 NONE SIGNAL DETECTION SYSTEM BACKGROUND OF THE INVENTION 1. FIELD OF THE INVENTION This invention relates to signal detection systems in which means are provided for reducing or removing noises involved in or superposed on an electrical signal so as to detect the electrical signal with a high signal-tonoise ratio (S/N ratio), and more particularly to a signal detection system preferably used for detecting a bioelectrical'signal such as one recorded on an electroencephalogram or electrocardiogram with a high S/N ratio by reducing or removing noises involved in or superposed on such bioelectric signal. I

2. DESCRIPTION OF THE PRIOR ART The present invention will be described withreference to the detection of a bioeleetri'cal signal, for example, a brain wave signal. The brain wave signal is generally derived and detected from electrodes mounted on the scalp and this signal is detected in many cases in a state in which noises or more specifically an electromyogram is superposed on the electroencephalogram.

The electroencephalogram is used for the diagnosis and discovery of brain diseases or persons having brain troubles by visual observation of the picture by brain specialists. It is demanded to remove the noise of the kind above described as much as possible since the presence of such noise leads to an erroneous diagnosis. An analog low-pass filter has been mainly employed in a prior art system for the purposeof removing the electromyogram superposed on the electroencephalogram. However, the prior art system employing the analog low-pass filter has had the following defects:

1. When a high-frequency component, for example,

a spike wave is involved in an electroencephalogram signal, this high-frequency component is greatly attenuated.

2. An artificially produced wave (artifact) tends to be involved in the signal component obtained by removing noise from the electroencephalogram.

In view of the above defects, it is common practice to compare an electroencephalogram obtained by removing the electromyogram by the use of the analog low-pass filter with an electroencephalogram obtained without using this filter for the diagnosis of brain diseases.

SUMMARY OF THE INVENTION With a view to obviate prior art defects as above described, it is an object of the present invention to provide a novel and improved signal detection system in which means are provided for reducing a muscle action potential signal relative to an electroencephalographic signal involving such a muscle action potential signal so as to detect the electroencephalographic signal with a high signal-to-noise ratio.

Another object of the present invention is to provide a signal detection system which can reproduce an electroencephalogram signal with good reproducibility even after the reduction of the noise.

A further object of the present invention is to provide a signal detection system which enables to make an accurate diagnosis of brain'diseases on the basis of an electroencephalogram.

In an effort to attain the above objects, the inventors carried out the following experiment. In view of the for measuring the character of the muscle action potential signal. The inventors have found very interesting results as follows:

1. The gradient of the muscle action potential signal is steeper than that of an abnormal electroencephalographic or spike wave portion having a steepest gradient in the brain wave signal. It is known that the gradient of the spike wave is biologically given by 3.41 i 1.14 p. V/ms (mean value i standard deviation). Thus, when, for example, the signal is sampled at a sampling period of 2.5 ms (400 Hz), the differential between the sampled values or data of the spike wave sampled with the sampling period of 2.5 ms, in other words, the gradient of the spike wave, is given by (3.41 i 1.14) X 2.5 x 8.5 i 2.9 ,u V/2.5 ms. Due to the fact that the differential in the case of the muscle action potential signal is greater than the above value, the waveform portion in which the differential between the sampled values exceeds a predetermined value, that is, the waveform portion having a gradient steeper than a predetermined value may be removed so as to remove the electromyogram from the electroencephalogram.

2. The duration of the muscle action potential signal is less than 20 ms. Especially, the muscle action potential signal having a duration of less than 10 ms is objectionable. Therefore, observation of several sampled data can clearly identify the muscle action potential signal when the electroencephalographic signal is sampled at the sampling period of 2.5 ms.

On the basis of the above facts, the present invention contemplates the provision of a process which comprises subjecting an electroencephalographic signal including a muscle action potential signal to sampling for obtaining first sampled data x,- (

i

1, 2, 3, seeking the differential y, (n x between the first sampled data x,, and correcting x by the sum of x,- and a predetermined infinitesimal increment C for reducing the magnitude of x,- when the differential y exceeds a predetermined threshold value C or leaving i+1 in the existing value without correcting x when the differential y does not exceed the threshold value C so as to obtain second sampled data. In this manner, any waveform portion in which the differential exceeds the predetermined threshold value C that is, any waveform portion having a steeper gradient than a predetermined value can be removed for reducing the muscle action potential signal component.

In other words, the present invention contemplates the provision of a system in which a sampled data x which is replaced by x iC when I y,-l C (where x, C is used if y, C and x, C is used if y C,) and which remains in the existing values when Y, I C is derived from data obtained by sampling an electroencephalographic signal including a muscle action potential signal so as to remove any waveform portion in which the differential exceeds a predetermined value, that is, any waveform portion having a gradient steeper than a predetermined value, thereby reducing the objectionable muscle action potential signal component to a minimum.

In the present invention, the threshold value C is selected to be 15 p. V/2.5 ms in view of the gradient of the spike wave in the electroencephalogram signal, and the infinitesimal increment C is selected to be p. V/2.5 ms taken into consideration the reproducibility of the brain wave signal.

BRIEF DESCRIPTION OF THE DRAWING FIG. la shows a waveform of an electroencephalogram signal (spike wave) including a muscle action potential signal.

FIG. 1b shows a waveform of an output from an analog low-pass filter when the signal waveform shown in FIG. la is passed through a filter.

FIG. 2a shows a waveform of an electroencephalogram (slow wave) including a muscle action potential signal.

FIG. 2b shows a waveform of an output from an analog low-pass filter when the signal waveform shown in FIG. 2a is passed through the filter.

FIG. 3a shows schematically a waveform obtained by sampling an electroencephalogram including a muscle action potential signal having a single peak.

FIGS. 3b and 3c show schematically waveforms obtained by sampling a first differential and a second differential respectively of the waveform shown in FIG. 3a.

FIG. 3d shows schematically a waveform obtained after correction of the waveform shown in FIG. 3a according to the present invention.

FIG. 4a shows schematically a waveform obtained by sampling an electroencephalographic signal including a muscle action potential signal having consecutive peaks.

FIGS. 4b and 4c show schematically waveforms obtained after correction of the waveform shown in FIG. 4a according to the present invention.

FIG. 5a shows schematically a waveform obtained by sampling an electroencephalogram including a muscle action potential signal having a sustained peak.

FIG. 5b shows schematically a waveform obtained after correction of the waveform shown in FIG. 5a according to the present invention.

FIGS. 6a to 6e are block diagrams showing the structure of preferred embodiments of the present invention and of control devices used therewith.

FIG. 7a shows schematically a waveform obtained by sampling an abnormal electroencephalographic wave or spike wave contained in an electroencephalogram.

FIGS. 7b and 7c show schematically waveforms obtained by sampling a first differential and a second differential respectively of the waveform shown in FIG. 70.

FIG. 8 shows schematically a waveform obtained by sampling a noise involved in an electroencephalogram.

FIG. 9 is a chart showing classified patterns of muscle action potentials.

FIG. 10 is a block diagram showing the structure of another embodiment of the present invention adapted for middle point correction.

FIG. 11 is a block diagram showing the structure of a further embodiment of the present invention adapted for linear correction.

FIGS. 12a to 12 are block diagrams showing the structure of other embodiments of the present invention.

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. la shows a waveform of an electroencephalogram including a high-frequency component such as a spike wave. When such a waveform is passed through a conventional analog low-pass filter, a muscle action potential signal involved in and superposed on the electroencephalogram is removed to a certain extent, but the high-frequency component is also attenuated greatly as seen in FIG. lb.

FIG. 2a shows a waveform of an electroencephalogram having a muscle action potential signal involved and superposed thereon. When such a waveform is passed through a conventional analog low-pass filter, an artificially produced waveform portion or artifact as shown in FIG. 2b tends to be involved in the output waveform in which the muscle action potential signal is substantially removed by the filter.

Electroencephalographic signals including a muscle action potential signal as shown in FIGS. 1a and 2a are sampled at a frequency of 400 Hz and the electroencephalographic signal component in the sampled signals is omitted. FIGS. 3a-5b show schematically such waveforms and differentials between the sampled values. FIG. 3a represents the case in which the muscle action potential signal includes a single peak, and this case will hereinafter be called a pattern A. FIG. 4a represents the case in which a plurality of patterns A having different levels appear consecutively, and this case will hereinafter be called a pattern B. FIG. 5a represents the case in which the muscle action potential signal includes a sustained peak having a duration which is two or three times the duration (2.5 ms) of the pattern A, and this case will hereinafter be called a pattern C. FIGS. 3b and 30 show schematically waveforms obtained by sampling a first differential and a second differential described later respectively. FIGS. 3d, 4b, 4c and 5b show schematically waveforms obtained after reducing the muscle action potential signal component in a manner as described later.

FIG. 6b is a block diagram showing the structure of an embodiment of the present invention. The system shown in FIG. 6b is constructed so that, in response to the application of a first sampled signal including a muscle action potential signal component of the kind corresponding to anyone of the patterns A, B and C, a second sampled signal substantially free from the muscle action potential signal component can be obtained at the output. FIG. 6a is a block diagram showing the structure of a control device which generates a plurality of control signals S S and S for controlling the system shown in FIG. 6b. Referring to FIG. 6a, a threestage shift register SL is controlled by a train of shift pulses S having a frequency which is three times the sampling frequency of 400 Hz. Output signals S and S of the first and second stages respectively of the shift register SL are applied to an OR gate OR. An inhibit gate IN is connected to the OR gate OR for inhibiting the output of the OR gate OR. The control signals 8,, S and S appear successively from the first, second and third stages respectively of the shift register SL in response to the shift action by the shift pulse S. The control signals S and S are applied to the OR gate OR and an input is applied from the OR gate OR to the inhibit gate lN. In this case, no input is applied to the first stage of the shift register SL. However, when the control signal S appears from the shift register SL, an input is applied from the inhibit gate IN to the first stage of the shift register SL to provide the control signal S In this manner, the control signals appear in the order of S S S within a period of time of 2.5 ms in response to the application of the shift pulses S.

Referring to H0. 6b, the first sampled signal is applied to an input terminal 1] of the system. A data selector 20 is controlled by a control signal *3. A pair of holding

circuits

2 and 3 are controlled by the control signals S and 8;, respectively and are connected at their output terminals to a differential amplifier 4. An absolute value detecting circuit 5 detects the absolute value of the output of the differential amplifier 4. A comparator 6 compares the output of the absolute value detecting circuit 5 with a predetermined threshold voltage representing a constant C, 15 1. W25 ms applied through a terminal TH The output of the comparator 6 and the control signal S, are applied to an AND gate 7, and a monostable multivibrator 8 is triggered by the output of the AND gate 7. The monostable multivibrator 8 delivers an output or control signal *3 for controlling a data selector l8 and the data selector 20 described above. More precisely, when the multivibrator 8 delivers the output *3, the data selectors l8 and 20 are connected to an adder l7 and a holding circuit 19 respectively, while when no output appears from the multivibrator 8, the

data selectors

18 and 20 are connected to the holding circuit 3. Another comparator 9 is connected at one of the input terminals to the output terminal of the differential amplifier 4 and is grounded at the other input terminal. The output of the comparator 9 and the control signal S, are applied to an AND gate 10, and anothermonostable multivibrator 11 is triggered by the output of the AND gate to deliver an output or control signal *4. An inhibit gate 12 is connected to the output terminal of the comparator 9. The output of the inhibit gate 12 and the control signal S are applied to an AND gate 13, and another monostable multivibrator 14 is triggered by the output of the AND gate 13 to deliver an output or control signal *5. A pair of

gates

15 and 16 are turned on in response to the application of the control signals *4 and *5 respectively and remain in the off state in the absence of such control signals. Predetermined threshold voltages representing a constant C 5 p. V/2.5 ms and a constant -C 5 p. V/2.5 ms are applied to the

gates

15 and 16 through terminals TH and TH respectively. The output of the holding circuit 2 and one of the predetermined threshold voltages C and C are applied to the adder 17. in the presence of the control signal *3, the value held in the holding circuit 19 is supplied through the data selector 20 to be held in the holding circuit 2 in response to the application of the control signal S to the holding circuit 2.

In operation, when, for example, a first sampled signal having a pattern A waveform as shown in FIG. 3a is applied to the input terminal ll of the system, a sampled value x, is held in the holding circuit 3 in response to the application of the control signal S to the holding circuit 3. Since no sampled value is held in the holding circuit 2 at this time, no output appears from the differential amplifier 4. Therefore, even when the control signal S, is applied to the AND gate 7, no output appears from the AND gate 7 due to the fact that no output is delivered from the comparator 6. The data selector 20 is connected to the holding circuit 3 due to the absence of the control signal *3, and the holding circuit 2 operates in response to the application of the control signal S to hold the sampled value x, held in the holding circuit 3. This completes one cycle of the control signals in the order of S S, S

Subsequently, another or next sampled value x is applied to the holding circuit 3 to be held therein in response to the application of the control signal S The value x held in the holding circuit 3 and the value x, held in the holding circuit 2 are applied to the differential amplifier 4 and an output (x -x is delivered from the differential amplifier 4. The absolute value detecting circuit 5 detects the absolute value x x l of the output of the differential amplifier 4. An output appears from the comparator 6 due to the fact that the absolute value I x x,\ is greater than the threshold value C in the case of the pattern A. An output appears from the AND gate 7 in response to the application of the control signal S The monostable multivibrator 8 is triggered thereby and delivers the control signal *3 so that the data selectors l8 and 20 are connected to the adder l7 and holding circuit 19 respectively. At the same time, the comparator 9 compared the output (x x,) of the differential amplifier 4 with the zero potential so as to determine whether (x x is positive or negative.

Since x x 0 in the case of the pattern A, an out put appears from the comparator 9 to be applied to the AND gate 10, and an output appears from the AND gate 10in response to the application of the control signal S thereto. The monostable multivibrator 11 is triggered by the output of the AND gate 10 and delivers the control signal *4. Since no output appears from the inhibit gate 12 in this case, no output appears from the AND gate 13 and the multivibrator 14 does not deliver the control signal *5. The gate 15 is turned on in response to the application of the control signal *4 and the predetermined threshold value C is applied to the adder 17 through the terminal TH and gate 15. The value x held in the holding circuit 2 is also applied to the adder 17. Thus, a corrected output x x C appears from the adder 17 to be applied through the data selector 18 to the holding circuit 19 to be held therein. The corrected output of the data selector 18 is held in the holding circuit 19 for the sampling period of time of 2.5 ms and is then delivered from an output terminal 1-2 of the system as an output signal due to the fact that the holding circuit 19 is controlled by the control signal 5,. Subsequently, the corrected value is transferred from the holding circuit 19 to the holding circuit 2 through the data selector 20 in response to the appearance of the control signal S Another sampled value x is applied to and held in the holding circuit 3 in response to the application of the control signal S and an output (x x appears from the differential amplifier 4 to be applied to the absolute value detecting circuit 5, thence to the comparator 6. Since the value (x;, x is smaller than the threshold value C in the case of the pattern A, no output appears from the comparator 6, and no output appears from the AND gate 7 even with the application of the control signal S, to

the AND gate 7. The monostable multivibrator 8 is not triggered and the control signal *3 does not appear from the multivibrator 8. Thus, the value x held in the holding circuit 3 is supplied through the data selector 18 to the holding circuit 19 to be held therein, and this value is subsequently delivered from the output terminal 1-2 of the system. In this manner, a second sampled signal as shown in FIG. 3d is obtained at the output terminal 1-2 of the system in the case of the pattern A. It will be seen from FIG. 3d that the level of the muscle action potential signal component in this second sampled signal is remarkably reduced.

On the other hand, when a first sampled signal having a pattern B waveform as shown in FIG. 4a is applied to the input terminal l1 of the system, an output apl5 pears from the AND gate 7 to trigger the monostable multivibrator 8 in response to the application of the control signal S to the AND gate 7 due to the fact that the absolute value [x x 'l of the differential (x x is greater than the predetermined threshold value C However, no output appears from the AND gate 10 and the monostable multivibrator 11 is not triggered since x x O in this case. Consequently, the inhibit gate 12 delivers an output and the control signal *5 is delivered from the monostable multivibrator 14 to turn on the gate 16. Due to the turn-on of the gate 16, the predetermined threshold'value -C is supplied to the adder 17 through the terminal TH and gate 16 to be added to the value x x C held in the holding circuit 2 to give the result x x C x C C x, This corrected value x;, is supplied from the holding circuit 19 to the holding circuit 2 through the data selector in response to the appearance of the control signal S Subsequently, a new sampled value x. is applied to and held in the holding circuit 3 in response to the appearance of the control signal S The operation similar to that above described is repeated thereafter. Consequently, a corrected waveform or a second sampled signal waveform as shown in FIG. 4b can be obtained.

When a first sampled signal having a pattern C waveform as shown in FIG. 5a is applied to the input terminal 11 of the system, the system operates in a manner entirely similar to that above described so that a second sampled signal waveform as shown in FIG. 5b can be obtained.

The above description has referred to the case in which the differential between a sampled value corrected by the threshold value C and the next sampled value is compared with the predetermined threshold value C However, the present invention is in no way limited to the comparison between such values, and the differentials between the sampled values, that is, the differential between the sampled values x and x and the differential between the sampled values x and x may be compared with the threshold value C and correction may then be applied to the result of comparison. In this case, the system shown in FIG. 6b may be suitably modified as shown in FIG. 6e. Referring to FIG. 60, the data selector 20 in FIG. 6b is eliminated to directly connect the holding circuit 2 with the holding circuit 3 and another holding circuit 20' is connected between the holding circuit 19 and the adder 17. In the system shown in FIG. 6e, the control signal S is applied to the holding circuit 20 to supply a shift pulse, the output of the holding circuit 19 is applied to the holding circuit 20' as an input thereto, and the output of the holding circuit 20 is applied to the adder 17 as one of the inputs thereto.

In the system shown in FIG. 6e, the holding circuit 20' responds to the control signal S to hold the value 5 held in the holding circuit 19 and the value held in the holding circuit 20 is applied to the adder 17. Thus, when the absolute value Ix x is greater than the threshold value C the adder l7 delivers a corrected output x C Therefore, the patterns A and B can be 10 corrected in a manner similar to that described with reference to FIG. 6b. In the case of the pattern C, any substantial correction is applied due to the fact that the values except the value (x, x are zero as will be apparent from FIG. 5a.

In the above description, the threshold value representing the constant C, has been set at l5 p.V/2.5 ms. However, depending on persons whose electroencephalogram is to be measured, there are some cases in which an electroencephalogram includes a waveform portion which exceeds this threshold value and can still be considered as a spike wave. If the threshold voltage is set at the specified value and correction is applied in such a case, a great reduction may occur in the amplitude of the peak point (waveform portion having a greatest amplitude value) having an especially steep gradient in the spike wave and poor reproducibility of the electroencephalogram may result. In order to deal with the case in which a spike wave having a steep gradient exists in an electroencephalogram, the present invention employs an arrangement in which means are provided for correcting the result of measurement on the basis of a second differential for the purpose of reducing the muscle action potential signal component 5 without attenuating the spike wave. In other words, a

second differential is derived from first differentials between sampled values and correction is applied to the result of comparison between this second differential and a predetermined threshold value representing a constant C, 30 [LV/(2.5 ms) instead of the threshold value C above described. Examples of second differentials of the muscle action potential signal and brain wave signal are shown in FIGS. 30 and 7c respectively.

FIG. 6d is a block diagram showing the structure of parts of another embodiment ofv the present invention adapted for deriving such a second differential, and FIG. 6c is a block diagram showing the structure of a control device which generates a plurality of control signals S S S and S for controlling the system shown in FIG. 6d. Referring to FIG. 6c, a four-stage shift register SL is controlled by a train of shift pulses S having a frequency which is four times the sampling frequency of 400 Hz. Output signals S S and S of the first, sec- 0nd and third stages respectively of the shift register SL are applied to an OR gate OR. An inhibit gate IN is connected to the OR gate OR for inhibiting the output of the OR gate OR. This control device operates in a manner similar to that described with reference to FIG. 6a, and the control signals appear in the order of 8 8 S S within a period of time of 2.5 ms.

The system shown in FIG. 6d is actually a modification of the system shown in FIG. 6b and differs from the latter in that three holding circuits 2-1, 2-2 and 2-3 and three differential amplifiers 4-1, 4-2 and 4-3 are provided in place of the holding

circuits

2 and 3 and differential amplifier 4 and the data selector 20 is eliminated.

When a first sampled signal having a pattern A waveform as shown in FIG. 3a is applied to the input terminal 1-1 of the system, a value x is held in the holding circuit 2-3 in response to the application of the control signal S The control signal S is subsequently applied to one of the input terminals of the AND gate 7. At this time, the contents of the holding circuits 2-1 and 2-2 are zero and no outputs appear from the differential amplifiers 4-1, 4-2 and 4-3. Thus, no input is applied to the other input terminal of the AND gate 7 and the monostable multivibrator 8 is not triggered. Subsequently, the control signal S is applied to the holding circuit 2-1, but the content of the holding circuit 2-1 is zero due to the fact that the content of the holding circuit 2-2 is zero at this time.

In response to the application of the control signal S to the holding circuit 2-2, the value x held in the holding circuit 2-3 is transferred to the holding circuit 2-2. Another or next value x is then supplied to and held in the holding circuit 2-3 in response to the application of the control signal S However, the content of the holding circuit 2-1 is still zero at this time and no outputs appear from the differential amplifiers 4-1 and 4-3. No output appears from the AND gate 7 and the monostable multivibrator 8 is not triggered even when the control signal S is applied to the AND gate 7. The value x held in the holding circuit 2-2 is transferred to the holding circuit 2-1 in response to the application of the control signal S the value x held in the holding circuit 2-3 is transferred to the holding circuit 2-2 in response to the application of the control signal S and a new value x is supplied to and held in the holding circuit 2-3 in response to the application of the control signal S Thus, the differential amplifiers 4-1, 4-2 and 4-3 deliver a first differential (x x a first differential (x x and a second differential [(x x (x x respectively. The absolute value detecting circuit 5 detects the absolute value of the second differential l (x x (x x I and this absolute value is supplied to the comparator 6 to be compared with a predetermined threshold value representing the constant C 30 uV/(2.5 ms) suppled through the terminal TI-I Due to the fact that this second differential is greater than the threshold value C as seen in FIG. 30 in the case of the pattern A, an output appears from the comparator 6 and an output appears from the AND gate 7 in response to the application of the control signal S The system operates thereafter in the same manner as that described with reference to FIG. 6b, except that. the predetermined threshold value representing the constant C used for correction is 10 uV/(2.5 ms)? In this manner, necessary correction is applied to the result of sampling to obtain a second sampled signal. A first sampled signal including a pattern B waveform as shown in FIG. 4a can be similarly processed.

It can be seen from the above description that little attenuation occurs in an portion (spike wave) of the electroencephalogram having a steep gradient as shown in FIG. 7a. This is because the peak point having a steep gradient is solely corrected relative to the second differential due to the fact that the gradient of the spike wave is not so steep at points except the'peak point as seen in FIG. 7b. I

The above description has referred to the case in which an electroencephalogram including a muscle action potential signal is sampled to detect the muscle action potential signal component having a gradient steeper than a predetermined value and the values obtained by sampling are corrected by a predetermined threshold value for reducing the muscle action potential signal component. However, in the case of the pattern C, distortion may occur in the electroencephalographic signal waveform after correction or an artifact may be artificially produced in the corrected electroencephalogram. FIG. 8 shows schematically a waveform obtained after sampling a brain wave signal including such an artifact and this waveform will hereinafter be called a pattern D. In order to improve the reproducibility of such a waveform and also to improve the reproducibility of the pattern B waveform, linear correction or middle point correction is preferably employed. In the linear correction, a first signal obtained by sampling an electroencephalogram is corrected by a linearly approximated value obtained by linear approximation of two sampled values and a period of time therebetween. In the middle point correction, such a signal is corrected by the middle point values.

In order to carry out such linear correction or middle point correction, the differential between first sample values x, and x of a signal obtained by sampling is compared with the threshold value representing the constant C l5 ,u.V/2.5 ms, and 1 is used to represent the case in which the absolute value of the differential is greater than C while 0 is used to represent the case in which such value is not greater than C Referring to FIG. 9, the patterns A to D of the muscle action potential signal are represented by 5-bit coded patterns of l and 0, and the linear correction or middle point correction is applied or is not applied to the patterns A to D depending on the coded patterns. The potterns A to D in the chart shown in FIG. 9 are coded in the following manner:

1. In the case of the pattern A, the differential between sampled values is either 1 or 0, an the pattern A is represented by, for example, (00000) or (001 10). In the pattern A waveform shown in FIG. 3a, the absolute values of the differentials between x and x and between x and x are greater than C and they are represented by 1, while the differentials between any other values of x, and x, are represented by 0. Thus, the pattern A waveform shown in FIG. 3a is represented by (01 2. The pattern B corresponds to the case in which the pattern A appears consecutively. Thus, the pattern B includes more ls than the pattern A as seen in FIG. 9 and the patter B waveform shown in FIG. 4a is represented by (0111 1).

3. The pattern C corresponds to the case in which the duration (2.5 ms) of the pattern A is increased to two or three times. The pattern C is represented by, for example, (01010) or (01001) as seen in FIG. 9 depending on the appearance of l and 0. The pattern C waveform shown in FIG. 5 is represented by (01001) due to the fact that the absolute values of the differentials between x and x and between x and x are greater than C 4. The pattern D is a stepwise varying pattern. This pattern D is represented by, for example, (00100) as seen in FIG. 9 due to the fact that the absolute value of the differential between values x, and x at the stepped portion is greater than C and this case is represented by 1, while anyother differentials are represented by 0. The pattern D waveform shown in FIG. 8 is represented by (01000) since the absolute value of the differential between x and x is greater than C,. In the case of this pattern D, no correction is applied to its coded pattern in view of the fact that unsatisfactory waveform reproducibility may result when corrected by a predetermined threshold value as pointed out hereinbefore.

FIG. 10 is a block diagram showing the structure of another embodiment of the present invention adapted for middle point correction, and like reference numerals are used therein to denote like parts appearing in FIG. 6b. Referring to FIG. 10, a first signal obtained by sampling is applied through an input terminal 1-l to a pair of delay means 20 and 21. The delay means 20 acts to delay the input signal by five hits, while the delay means 21 acts to delay the input signal by four bits. Another delay means 22 is provided to delay the output of a differential amplifier 4 by five bits. A voltage divider 23 divides the output of the delay means 22 into V2. A -stage shift register 24 is connected to a monostable multivibrator 8 to receive the output of the latter. An inhibit gate 25 inverts the output from the first stage of the shift register 24. The output of the inhibit gate 25 and outputs from the remaining stages of the shift register 24 are applied to an AND gate 26.

In operation, suppose, for example, that a first sampled signal including a muscle action potential signal component having a pattern B vaveform as shown in FIG. 4a is applied to the input terminal l1 of the system. In response to the application of such an input signal, the differentials (x x (x x (x x are successively delivered from the differential amplifier 4. This output is applied through an absolute value detecting circuit 5 to a comparator 6 to be compared with a predetermined threshold value C supplied through a terminal TH An output appears from the comparator 6 due to the fact that each differential is greater than the threshold value C in the case of the pattern B. An AND gate 7 is turned on in response to the application of a control signal S and the monostable multivibrator 8 is triggered to apply its output to the shift register 24. As a result, the shift register 24 registers (l l l 10) and the output from the first stage of the shift register 24 is inverted into 1 by the inhibit gate 25. An output appears from the AND gate 26 due to the application of all the inputs from the shift register 24.

On the other hand, the differentials are also applied from the differential amplifier 4 to the delay means 22 to be delayed by five bits thereby, and the output of the delay means 22 is divided into /2 by the voltage divider 23, this divided voltage being then applied to one of the input terminals of an adder 17. The values x x x x and x of the first sampled signal are successively applied to the other input terminal of the adder 17 after being delayed by five bits by the delay means 21. The adder 17 executes the addition of the half value of the differential and the sampled value, for example, the addition of /a(x x and x In the meantime, a data selector 18 is controlled by the control signal applied from the AND gate 26 and is placed in the position in which it supplies the result of addition carried out by the adder 17 to a holding circuit 19. Thus, the result of addition of the sampled value to the half value of the differential appears at an output terminal 1-2 of the system as a second sample signal having a waveform as shown in FIG. 40.

When the coded pattern corresponding to the pattern B does not exist within the five bits of the first sampled signal, no control signal appears from the AND gate 26 and the data selector 18 is placed in the position in which it supplies the output of the delay means 20 directly to the holding circuit 19 so that the values of the first sampled signal appear successively at the output terminal 1-2 of the system. A plurality of combinations each consisting of a shift register and an inhibit gate as above described may be arranged in parallel so as to detect a group of coded patterns corresponding to various kinds of the pattern B. In this manner, the pattern B to be subjected to the middle point correction can be detected and a second sampled signal substantially free from the muscle action potential signal component of the pattern B can be obtained.

FIG. 11 is a block diagram showing the structure of still another embodiment of the present invention adapted for linear correction, and like reference numerals are used therein to denote like parts appearing in FIG. 6b. Referring to FIG. 11, a first signal obtained by sampling is applied through an input terminal 1-1 to a delay means 20 to be suitably delayed thereby. A function generator 21 generates a linear function for the purpose of linear correction. A timer 27 is connected to a monostable multivibrator 8 to be controlled by the output of the latter. A plurality of inhibit gates 25-1, 25-2 and 25-3 are provided for inverting the outputs from the first, third and fourth stages respectively of a five-stage shift register 24.

In operation, suppose, for example, that a first sampled signal including a muscle action potential signal component having a pattern C waveform as shown in FIG. 5a is applied to the input terminal 11 of the system to be subjected to linear correction. In response to the application of such an input signal to the input terminal 1-1, the differentials (x x (x x (x, x and (x x are successively delivered from a differential amplifier 4 in a manner similar to that described with reference to FIG. 11. Due to the fact that the absolute values of the differentials (x -x and (,1: x are greater than a predetermined threshold value C,, an output appears from the monostable multivibrator 8 each time such differential appears from the differential amplifier 4. As a result, the shift register 24 registers (01001) and the outputs from the first, third and fourth stages of the shift register 24 are inverted into I by the respective inhibit gates 25-1, 25-2 and 25-3. An output appears from the AND gate 26 to be applied to a data selector 18 for controlling same. The output of the monstable multivibrator 8 is also applied to the timer 27, which is set in response to the first pulse applied from the multivibrator 8 and is reset in response to the application of the next input pulse so that the period of time between these pulses is measured. The output of the timer 27 is applied to the function generator 21. Since the data selector 18 is now placed in the position in which it supplied the output of the delay means 20 to the function generator 21 by being controlled by the control signal applied from the AND gate 26, a linear function obtained by linear correction of the sampled values x and x and the period of time therebetween is generated from the function generator 21 for the above period of time following the sampled value x,. Thus, a linearly corrected second sampled signal is applied through a holding circuit 19 to appear at an output terminal 12 of the system.

When the coded pattern corresponding to the pattern C does not exist within the five bits of the first sampled signal, no control signal appears from the AND gate 26 and the data selector 18 is placed in the position in which it supplies the output of the delay means 20 directly to the holding circuit 19 so that the first sampled signal appears at the output terminal 1-2 without being subjected to the linear correction. A plurality of combinations each consisting of a shift register and AND gates as above described may be arranged in parallel so as to detect a group of coded patterns corresponding to various kinds of the patterns A and C. In this manner, the patterns A and C to be subjected to the linear correction can be satisfactorily detected and the desired reduction of the muscle action potential signal component can be attained.

No correction is applied to the muscle action potential signal component of the pattern D as described previously. Thus, even if there is a correcting system including the shift registers and AND gates, which exclusively detects the coded patterns except that corresponding to the pattern D, the first sampled signal is not detected, and thus, the desired objects can be attained.

It will be understood that a high S/N ratio can be obtained by virtue of the fact that a muscle action potential signal superposed on a brain wave signal can be substantially removed. Further, an electroencephalogram having better reproducibility than heretofore can be obtained and the diagnosis with better accuracy can be expected. This contributes more to the advance of medical science.

The foregoing description has referred to a system for reducing a muscle action potential signal component superposed on or involved in a brain wave signal and detecting the brain wave signal with good reproducibility. However, the present invention is in no way limited to such a system. Diagnostically very important is a system which comprises means for measuring noise involved in an electroencephalogram, means for automatically generating an alarm to tell the clinical exam iner the fact that the allowable quantity of the noise exceeds a predetermined threshold value so that the diagnosis can be automatically interrupted temporarily, and means for restarting the detection of the brain wave signal after the source of the noise has been removed.

In an effort to obtain such a system, the inventors have made experiments to determine the allowable quantity of noise involved in an electroencephalogram and found that the diagnosis should be ceased when the integrated value of the noise measured for a predetermined period of time of, for example, one second exceeds a predetermined threshold value representing a constant voltage gradient value of, for example, mV/sec.

According to another aspect of the present invention, the present invention contemplates the provision of a system which measures noise continuously and integrates the quantity of the noise involved in a brain wave signal so that the clinic examiner can make an accurate diagnosis of brain diseases on an electroencephalogram.

An embodiment of the present invention suitable for this purpose will be described with reference to FIG. 12a and 12b. FIG. 12b is a block diagram showing the structure of parts of such an embodiment adapted for obtaining an integrated value of anoise, and FIG. 12a is a block diagram showing the structure of a control device. Referring to FIG. 12a, a four-stage shift register SL is controlled by a train of shift pulses S having a frequency which is five times the sampling frequency of 400 Hz. Outputs 5,, S and S from the first, second and third stages respectively of the shift register SL are applied to an OR gate OR to which an inhibit gate IN is connected. The operation of the control device shown in FIG. 12a is similar to that of the control device shown in FIG. and the control signals appear in the order of S S, S S within a period of time of 2.5 ms. The function of the control signals S S and S is the same as that of the respective control signals S S and S in FIG. 6a. The control signal S is used for integration of a noise.

In FIG. 12b in which like reference numerals are used to denote like parts appearing in FIG. 6b, a plurality of blocks designated by the reference numerals 201 to 207 are provided for attaining the integration of the noise. A control signal S is generated by a pulse generator (not shown) at intervals of a predetermined period of time of, for example, 1 second.

Referring to FIG. 12b, a monostable multivibrator 8 delivers a control signal *3 for controlling a gate 201, and the output of the gate 201 is applied to one of the input terminals of an adder 202 to the other input terminal of which another input is applied from a holding circuit 204. A data selector 203 is connected between the adder 202 and the holding circuit 204 and is changed over to be connected to ground in response to the application of the control signal S The output of the data selector 203 is applied to the holding circuit 204 to be held therein in response to the application of the control signal S A gate 205 is turned on in response to the application of the control signal S and the output of the gate 205 is applied to a comparator 206 to be compared with a predetermined threshold value C supplied through a terminal TH A monostable multivibrator 207 is triggered by the output of the comparator 206, and the output of the multivibrator 207 appears at an output terminal l-3. The output of the gate 205 appears also at another output terminal 1-4.

In operation, suppose, for example, that a first sampled signal including a pattern A waveform as shown in FIG. 3a is applied through an input terminal l-l of the system to a holding circuit 3 to be held therein. The differential between the value held in the holding circuit 3 and the preceding sampled value held in another holding circuit 2 appears at the output terminal of a differential amplifier 4 and this output is applied to a comparator 6 through an absolute value detecting circuit 5 and to another comparator 9 directly. Outputs appear from the

comparators

6 and 9 when this input is greater than a predetermined threshold value supplied through a terminal TH, and is greater than the zero potential respectively. The outputs of the

comparators

6 and 9 are applied to one of the input terminals of AND

gates

7 and 10 respectively. In response to the application of the control signal S to the other input terminal of the AND

gates

7 and 10, outputs appear from the AND

gates

7 and 10 to trigger the

monostable multivibrators

8 and 11 and control signals *3 and *4 are generated from these

multivibrators

8 and 11 respectively. The control signal *3 is appliedto the gate 201 to turn on same. This control signal *3 is also applied to a data selector 18 to connect same to an adder 17 for receiving a corrected output from the adder 17. A gate 15 is turned on in response to the application of the control signal *4 so that a predetermined threshold value supplied through a terminal TH is applied to the adder 17 for correction and the corrected value is delivered from the adder 17. This corrected value is supplied through the data selector 18 to a holding circuit 19 to be held therein, and in response to the application of the control signal S to the holding circuit 2, the corrected value is supplied through a data selector 20 to the holding circuit 2 to be held therein. Thus, the output of the differential amplifier 4 is given by the differential between the corrected value, for example, x held in the holding circuit 3.

At this time, the gate 201 is already in the on position due to the application of the control signal *3 thereto. Therefore, the output of the gate 201 represents the differential between the sampled value x and the corrected value x of x that is, the quantity of the removed noise which is the muscle action potential signal component. The output of the gate 201 and the output of the holding circuit 204 are added to each other in the adder 202, and the result of addition is kept held in the holding circuit 204 until the one-second signal S is applied to the data selector 203. In other words, integration is carried out for this period of time. The gate 205 is turned on in response to the application of the control signal 5 and the output of the gate 205 representing the quantity of the noise appears at the output terminal l-4 and is applied to, for example, a digital display means to be displayed thereon. At the same time, the holding circuit 204 is reset due to the fact that the data selector 203 is connected to ground in response to the application of the control signal S The output of the gate 205 is also applied to one of the input terminals of the comparator 206 to be compared with the predetermined threshold value C supplied to the other input terminal of the comparator 206 through the terminal TH This threshold value representing the constant C is, for example, 15 mV/sec. Therefore, when the output of the holding circuit 204 is greater than the threshold value C, an output appears from the comparator 206 to trigger the monostable multivibrator 207, and a control signal appears at the output terminal 1-3 to inform the clinical examiner of the fact that the muscle action potential signal component is included in a large quantity in the electroencephalogram. An alarm means or the like may be connected to the output terminal l-3 to give an alarm for the above situation so that the diagnosis can be interrupted temporarily and restarted automatically after eliminating the source of inclusion of the muscle action potential signal and confirming the disappearance of the output at the output terminal l-3. 1

While the system shown in FIG. 12b has been described with reference to the integration of a noise for a predetermined period of time in the case of the system shown in FIG. 6b, it will be apparent to those skilled in the art that, in the case of the system shown in FIG. 6d too, a system as shown in FIG. 120 may be employed for the integration of a noise in a manner similar to that described with reference to FIG. 12b.

It will be understood from the above description that the present invention provides a system in which means are provided for integrating the separated muscle action potential signal for a predetermined period of time, continuously displaying the integrated value and automatically issuing an alarm when the integrated value exceeds a predetermined threshold value, so that the diagnosis can be ceased temporarily and restarted automatically when the integrated value is reduced to less than the threshold value. Thus, an accurate diagnosis of brain diseases on the basis of the electroencephalogram can be ensured.

Further, while the foregoing description has referred solely to the removal of muscle action potentials involved in a brain wave signal, it will be apparent to those skilled in the art that the present invention is also effectively applicable to the detection of noises involved in a heart action potential signal and any other bioelectrical signals as well as in common electrical signals.

We claim:

1. A signal detection system comprising holding means for holding an electrical signal subjected to sampling, detecting means for detecting the differential between sampled values x, and x (where i is an integer) held in said holding means, comparing means for comparing the output of said detecting means with a predetermined first threshold value, discriminating means for discriminating the polarity of the output of said detecting means, adding means controlled by the output of said discriminating means for adding a predetermined second threshold value to the sampled value x,- in response to the appearance of the sampled value x,-,,, means for applying the result of addition by said adding means to said detecting means for the purpose of detecting the differential between the result of addition and another sampled value x and means controlled by the output of said comparing means to be connected to said adding means in lieu of said holding means for deriving the result of addition from said adding means, said result of addition and said sampled value being selectively derived depending on the appearance or disappearance of the output from said comparing means.

2. A signal detection system as claimed in claim 1, further comprising integrating means for integrating the differential between said result of addition and said sampled value for a predetermined period of time, another comparing means for comparing the output of said integrating means with a predetermined third threshold value, and means for displaying the output of said comparing means.

3. A signal detection system comprising holding means for holding an electrical signal subjected to sampling, a pair of first detecting means for detecting the differentials between sampled values x and x and between sampled values x and x respectively (where i is an integer) held in said holding means, second detecting means for detecting the differential between the outputs of said first detecting means, comparing means for comparing the output of said second detecting means with a predetermined first threshold value, discriminating means for discriminating the polarity of the output of said second detecting means, adding means controlled by the output of said discriminating means for adding a predetermined second threshold value to the sampled value x, in response to the appearance of the sampled value n means for applying the result of addition by said adding means to said first detecting means for the purpose of detecting the differential between the result of addition and the sampled value x and means controlled by the output of said comparing means to be connected to said adding means in lieu of said holding means for deriving the result of addition fromsaid adding means, said result of addition andsaid sampled value being selectively derived depending on the appearance or disappearance of the output from said comparing means.

4. A signal detection system as claimed in claim 3,

' further comprising integrating means for integrating the differential between said result of addition and said sampled value for a predetermined period of time, another comparing means for comparing the output of said integrating means with a predetermined third threshold value, and means for displaying the output of said comparing means.

5. A signal detection system comprising holding means for holding an electrical signal subjected to sampling, detecting means for detecting the differential between sampled values x, and x (where i is an integer) held in said holding means, means for dividing the output of said detecting means into /2, comparing means for comparing the output of said detecting means with a predetermined first threshold value, adding means for adding the output of said dividing means to the'sampled value x,-, determining means for determining the pattern of said electrical signal depending on the appearance or disappearance of the output from said comparing means, and means controlled by the output of said determining means to be connected to said adding means in lieu of said holding means for deriving the result of addition from said adding means, said result of addition and said sampled value being selectively derived depending on the appearanceor disappearance of the output from said determining means.

6. A signal detection system comprising holding means for holding an electrical signal subjected to sampling, detecting means for detecting the differential between sampled values at, and x (where i is an integer) held in said holding means, comparing means for comparing the output of said detecting means with a predetermined threshold value, measuring means controlled by the output of said comparing means for measuring the period of time between successive outputs of said comparing means, determining means for determining function, said linear function and said sampled value being selectively derived depending on the appearance or disappearance of the output of said determining means.

7. A signal detection system comprising holding means for holding an electrical signal subjected to sampling, detecting means for detecting the differential between sampled values x, and x (where i is an integer) held in said holding means, comparing means for com-' paring the output of said detecting means with a predetermined first threshold value, discriminating means'for discriminating the polarity of the output of said detecting means, adding means controlled by the output of said discriminating means for adding a predetermined second threshold value to the sampled value x, in response to the appearance of the sampled value x means controlled by the output of said discriminating means for supplying the predetermined second threshold value to said adding means for addition to thereby said result of addition, and means controlled by the output of said comparing means to be connected to said adding means in lieu of said holding means for deriving said result of addition from said adding means, said result of addition and said sampled value being selectively derived depending on the appearance or disappearance of the output from said comparing means.

* l =l r

Claims

1. A signal detection system comprising holding means for holding an electrical signal subjected to sampling, detecting means for detecting the differential between sampled values xi and xi 1 (where i is an integer) held in said holding means, comparing means for comparing the output of said detecting means with a predetermined first threshold value, discriminating means for discriminating the polarity of the output of said detecting means, adding means controlled by the output of said discriminating means for adding a predetermined second threshold value to the sampled value xi in response to the appearance of the sampled value xi 1, means for applying the result of addition by said adding means to said detecting means for the purpose of detecting the differential between the result of addition and another sampled value xi 2, and means controlled by the output of said comparing means to be connected to said adding means in lieu of said holding means for deriving the result of addition from said adding means, said result of addition and said sampled value being selectively derived depending on the appearance or disappearance of the output from said comparing means.

3. A signal detection system comprising holding means for holding an electrical signal subjected to sampling, a pair of first detecting means for detecting the differentials between sampled values x1 and xi 1 and between sampled values xi 1 and xi 2 respectively (where i is an integer) held in said holding means, second detecting means for detecting the differential between the outputs of said first detecting means, comparing means for comparing the output of said second detecting means with a predetermined first threshold value, discriminating means for discriminating the polarity of the output of said second detecting means, adding means controlled by the output of said discriminating means for adding a predetermined second threshold value to the sampled value xi in response to the appearance of the sampled value xi 2, means for applying the result of addition by said adding means to said first detecting means for the purpose of detecting the differential between the result of addition and the sampled value xi 2, and means controlled by the output of said comparing means to be connected to said adding means in lieu of said holding means for deriving the result of addition from said adding means, said result of addition and said sampled value being selectively derived depending on the appearance or disappearance of the output from said comparing means.

4. A signal detection system as claimed in claim 3, further comprising integrating means for integrating the differential between said result of addition and said sampled value for a predetermined period of time, another comparing means for comparing the output of said integrating means with a predetermined third threshold value, and means for displaying the output of said comparing means.

5. A signal detection system comprising holding means for holding an electrical signal subjected to sampling, detecting means for detecting the differential between sampled values xi and xi 1 (where i is an integer) held in said holding means, means for dividing the output of said detecting means into 1/2 , comparing means for comparing the output of said detecting means with a predetermined first threshold value, adding means for adding the output of said dividing means to the sampled value xi, determining means for determining the pattern of said electrical signal depending on the appearance or disappearance of the output from said comparing means, and means controlled by the output of said determining means to be connected to said adding means in lieu of said holding means for deriving the result of addition from said adding means, said result of addition and said sampled value being selectively derived depending on the appearance or disappearance of the output from said determining means.

6. A signal detection system comprising holding means for holding an electrical signal subjected to sampling, detecting means for detecting the differential between sampled values xi and xi 1 (where i is an integer) held in said holding means, comparing means for comparing the output of said detecting means with a predetermined threshold value, measuring means controlled by the output of said comparing means for measuring the period of time between successive outputs of said comparing means, determining means for determining the pattern of said electrical signal depending on the appearance or disappearance of the output from said comparing means, a function generator for generating a linear function on the basis of the sampled value xi and the output of said measuring means, and means controlled by the output of said determining means for deriving the sampled value xi 1 in lieu of the linear function, said linear function and said sampled value being selectively derived depending on the appearance or disappearance of the output of said determining means.

7. A signal detection system comprising holding means for holding an electrical signal subjected to sampling, detecting means for detecting the differential between sampled values xi and xi 1 (where i is an integer) held in said holding means, comparing means for comparing the output of said detecting means with a predetermined first threshold value, discriminating means for discriminating the polarity of the output of said detecting means, adding means controlled by the output of said discriminating means for adding a predetermined second threshold value to the sampled value xi in response to the appearance of the sampled value xi 1, means controlled by the output of said discriminating means for supplying the predetermined second threshold value to said adding means for addition to thereby said result of addition, and means controlled by the output of said comparing means to be connected to said adding means in lieu of said holding means for deriving said result of addition from said adding means, said result of addition and said sampled value being selectively derived depending on the appearance or disappearance of the output from said comparing means.