Method for extracting prescribed biological signal, program, biological signal extraction apparatus, and biological signal extraction system

Abstract

To provide a method capable of extracting a prescribed biological signal on the basis of one observation signal by a simple method.SOLUTION: A method for extracting a prescribed biological signal includes: generating two signals on the basis of one observation signal acquired by using a sensor; and extracting the prescribed biological signal from the generated two signals by using blind signal source separation. Generating the two signals includes: generating one of signals including sampling points equal to or more than a sample number corresponding to prescribed sampling frequency of the sampling points of the one observation signal; and generating other signals including a sample point different from the one signal equal to or larger than the sample number corresponding to the prescribed sampling frequency of the sampling point of the one observation signal.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a method, a program, a biological signal extraction device, and a biological signal extraction system for extracting a predetermined biological signal. [Background technology]

[0002] Pulse waves have been attracting attention as a means of understanding health conditions, but when measuring biological signals such as pulse waves, they typically contain stationary and non-stationary noise, making it difficult to obtain only the pulse waveform. A conventional method known as blind source separation (BSS) separates source signals using multiple observed signals acquired at multiple positions, extracting only the desired waveform. For example, Patent Document 1 discloses a signal source separation system for effectively separating and detecting vital signals. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent Publication No. 2021-071694 [Non-patent literature]

[0004] [Non-Patent Document 1] Journal of Signal Processing, Vol.9, No.5, pp.397-407 September 2015 "A Study on Blind Source Separation Using Kurtosis" Nakaya Kusakari, Hiroki Matsumoto et al. [Non-patent document 2] S. Amari “Natural gradient works efficiently in learning” Neural Computation Vol. 10, pp. 271-276, 1998 Summary of the Invention [Problem to be solved by the invention]

[0005] BSS is a known technique for separating multiple unknown source signals from multiple measurement signals that are a mixture of the source signals. When BSS is used to measure biological signals, a known method separates two observation signals acquired using two sensors placed at different locations into a pulse wave containing non-stationary noise and stationary noise containing non-stationary noise. However, this method has a problem in that it is difficult to determine an appropriate distance between the sensors. The present inventors have proposed a method using a quadrature mirror filter (QMF) to enable pulse wave extraction from a single observation signal acquired using a single sensor, but this method has a problem in that adjusting the filter characteristics when using QMF is complex and difficult.

[0006] In one aspect, the present invention has been made to solve such problems, and aims to provide a method, etc., that makes it possible to extract a predetermined biological signal based on a single observed signal in a simpler manner. [Means for solving the problem]

[0007] [1] A method according to one embodiment of the present invention comprises: 1. A method for extracting a predetermined biological signal, comprising: generating two signals based on one observation signal acquired using a sensor; extracting the predetermined biological signal from the two generated signals using blind source separation; Including, generating the two signals includes generating one signal including sampling points of the one observation signal that are equal to or greater than the number of samples corresponding to a predetermined sampling frequency, and generating another signal including sampling points of the one observation signal that are different from the one signal and are equal to or greater than the number of samples corresponding to the predetermined sampling frequency; It is a method.

[0008] [2] In one embodiment of the present invention, The method according to [1], wherein generating the two signals includes generating one signal including every other sampling point of the one observation signal, and generating another signal including every other sampling point of the one observation signal that is different from the one signal, and the one signal and the other signal are generated as signals at the same time.

[0009] [3] In one embodiment of the present invention, The method according to [1] or [2], wherein the one observation signal is a signal obtained by sampling a signal obtained using a sensor at a frequency that is at least twice the predetermined sampling frequency, or a signal obtained by using a sensor having a number of samples corresponding to a frequency that is at least twice the predetermined sampling frequency.

[0010] [4] In one embodiment of the present invention, The method according to any one of [1] to [3], wherein the predetermined biological signal is a signal that has periodicity and does not have symmetry within one period.

[0011] [5] In one embodiment of the present invention, The method according to any one of [1] to [4], wherein the predetermined biological signal is a pulse wave or an electrocardiogram signal.

[0012] [6] A program according to one embodiment of the present invention is a program that causes a computer to execute the method described in any one of [1] to [5].

[0013] [7] A biological signal extraction device according to one embodiment of the present invention comprises: A biological signal extraction device that extracts a predetermined biological signal, a signal generation unit that generates two signals based on one observation signal acquired using a sensor; a signal extraction unit that extracts the predetermined biological signal from the two signals generated by the signal generation unit using blind source separation; Equipped with the signal generator generates one signal including sampling points of the one observation signal that are equal to or greater than the number of samples corresponding to a predetermined sampling frequency, and generates another signal including sampling points of the one observation signal that are different from the one signal and that are equal to or greater than the number of samples corresponding to the predetermined sampling frequency. This is a biological signal extraction device.

[0014] [8] In one embodiment of the present invention, The signal generating unit generates one signal including every other sampling point among the sampling points of the one observation signal, and generates another signal including every other sampling point different from the one signal among the sampling points of the one observation signal, and the one signal and the other signal are generated as signals at the same time. This is the biological signal extraction device described in [7].

[0015] [9] A biological signal extraction system according to one embodiment of the present invention comprises: A biological signal extraction system for extracting a predetermined biological signal, a sensor for acquiring an observation signal; a signal generation unit that generates two signals based on one observation signal acquired by the sensor; a signal extraction unit that extracts the predetermined biological signal from the two signals generated by the signal generation unit using blind source separation; Equipped with the signal generator generates one signal including sampling points of the one observation signal that have a sampling frequency equal to or higher than a predetermined sampling frequency, and generates another signal including sampling points of the one observation signal that have a sampling frequency equal to or higher than the predetermined sampling frequency and that are different from the one signal. This is a biosignal extraction system.

[0016]

[10] In one embodiment of the present invention, The signal generating unit generates one signal including every other sampling point among the sampling points of the one observation signal, and generates another signal including every other sampling point different from the one signal among the sampling points of the one observation signal, and the one signal and the other signal are generated as signals at the same time. This is the biological signal extraction system described in [9]. [Effects of the Invention]

[0017] According to one aspect of the present invention, a predetermined biological signal can be extracted based on one observed signal using a simpler method. [Brief explanation of the drawings]

[0018] [Figure 1] 1 is a schematic configuration diagram of a biological signal extraction system according to an embodiment of the present invention. [Figure 2] FIG. 1 is a diagram for explaining an overview of the principle of blind source separation (BSS). [Figure 3] FIG. 1 is a diagram for explaining an outline of extracting a pulse wave using BSS. [Figure 4] 1 is a block diagram showing a hardware configuration of a biological signal extraction device according to an embodiment of the present invention. [Figure 5] 1 is a functional block diagram of a biological signal extraction device 3 according to an embodiment of the present invention. [Figure 6] 1 is a diagram showing an example of a processing model M of a processing signal generating unit and a biological signal extracting unit according to an embodiment of the present invention. FIG. [Figure 7] FIG. 7 is a diagram showing a model of oversampling in the processing model M of FIG. 6. [Figure 8] FIG. 1 is a diagram illustrating an overview of how a processing signal generator according to one embodiment of the present invention generates two signals (x(1)(n), x(2)(n)) based on one observed signal (x2(n)). [Figure 9] FIG. 10 is a diagram illustrating an outline of how a biological signal extraction unit of an embodiment of the present invention removes non-stationary noise using a threshold value. [Figure 10]3 is a flowchart showing the operation of the biological signal extraction system according to one embodiment of the present invention. [Figure 11] FIG. 10 is a diagram showing the results of a simulation experiment. DETAILED DESCRIPTION OF THE INVENTION

[0019] A biosignal extraction system 1 according to an embodiment of the present invention will be described below with reference to the drawings. The biosignal extraction system is a system for extracting a predetermined biosignal using blind source separation (BSS). The predetermined biosignal is, for example, a pulse wave or an electrocardiogram signal. In the description of the embodiment of the present invention, for convenience of explanation, the biosignal extraction system 1 may be described as a system for extracting a pulse wave, but is not limited to this. In this specification, extracting a pulse wave using BSS may mean separating and obtaining a pulse wave using BSS. Furthermore, in this specification, a pulse wave or an electrocardiogram may refer to a pulse wave or an electrocardiogram signal. For convenience of explanation, blind source separation will be abbreviated as BSS.

[0020] Figure 2 is a diagram illustrating the outline of the principle of BSS. The observed signal acquired by the sensor is composed of stationary noise S (1) (n) and pulse wave S including non-stationary noise (2) The observed signal x is obtained by a sensor arranged to obtain the pulse wave. (1) (n) and the observed signal x acquired by the sensor SB, which is arranged to acquire the pulse wave at a position different from the sensor SA. (2) (n) is the mixture matrix A shown in equation (2) and two source signals S2(n) = (S (1) (n),S (2) (n)). TIFF2025168870000002.tif14150(1) TIFF2025168870000003.tif11150(2)

[0021] As shown in equation (3), the separation matrix W and two observed signals x2(n) = (x (1) (n), x (2) (n)) and Y (2) By calculating (n), it is possible to obtain a pulse wave containing non-stationary noise from an observation signal actually acquired by a sensor. TIFF2025168870000004.tif14150(3) The separation matrix W is expressed by equation (4). TIFF2025168870000005.tif11150(4) The separating matrix W is ideally a generalized inverse matrix of the mixing matrix A, and can be obtained by a known method such as the method shown in Non-Patent Document 1.

[0022] Figure 3 is a diagram for explaining the outline of pulse wave extraction using BSS. The upper diagram of Figure 3 shows how the desired pulse wave S is extracted from two observation signals A and B acquired using two sensors placed at different positions. i The lower diagram of Fig. 3 illustrates a conventional method for extracting a desired pulse wave S from two signals C and C' generated from one observation signal acquired using one sensor. ii 1 is a diagram illustrating a method according to an embodiment of the present invention for extracting a signal obtained from a single sensor located at a single location. In this specification, a single observed signal refers to a signal obtained from a single sensor located at a single location, and is used to contrast with two observed signals obtained using two sensors located at different locations.

[0023] 1 is a schematic diagram of a biological signal extraction system 1 according to one embodiment of the present invention. The biological signal extraction system 1 includes a sensor 2 attached to the wrist or ankle, and a biological signal extraction device 3. The sensor 2 is a single sensor placed at a single position.

[0024] The sensor 2 is a known sensor capable of measuring an observation signal including a pulse wave. For example, the sensor 2 may be an optical sensor, an ultrasonic sensor, or a piezoelectric sensor. In the embodiment of the present invention, the sensor 2 is described as an analog sensor. The sensor 2 is communicatively connected to the biological signal extraction device 3, and the biological signal extraction device 3 receives the observation signal acquired by the sensor 2 from the sensor 2. For example, the sensor 2 may be configured to transmit the acquired observation signal to the biological signal extraction device 3 via wired or wireless communication. For example, if the sensor 2 is a piezoelectric sensor such as a piezoelectric pulse wave sensor, the sensor 2 may acquire a voltage signal caused by pressure fluctuations of the pulse wave and output the voltage signal to the biological signal extraction device 3. The sensor 2 acquires the observation signal as an analog signal, and the biological signal extraction device 3 samples the analog signal received from the sensor 2 to acquire a digital observation signal.

[0025] 4 is a block diagram showing the hardware configuration of a biosignal extraction device 3 according to one embodiment of the present invention. The biosignal extraction device 3 includes a processor 11, an input device 12, a display device 13, a storage device 14, and a communication device 15. These components are connected by a bus 16. Note that an interface is interposed between the bus 16 and each component device as necessary. For example, the biosignal extraction device 3 is a computer.

[0026] The processor 11 controls the overall operation of the biological signal extraction device 3. For example, the processor 11 is a CPU. The processor 11 performs various processes by reading and executing programs and data stored in the storage device 14. The processor 11 may be composed of multiple processors.

[0027] The input device 12 is a user interface that accepts input from a user to the biological signal extraction device 3, and is, for example, a touch panel, a touch pad, a keyboard, a mouse, a button, or a sensor. The display device 13 is a display that displays an application screen or the like to the user of the biological signal extraction device 3 under the control of the processor 11.

[0028] The storage device 14 includes a main storage device and an auxiliary storage device. The main storage device is, for example, a semiconductor memory such as RAM. RAM is a volatile storage medium that allows high-speed reading and writing of information, and is used as a storage area and a working area when the processor 11 processes information. The main storage device may also include ROM, which is a read-only nonvolatile storage medium. The auxiliary storage device stores various programs and data used by the processor 11 when executing each program. The auxiliary storage device may be any nonvolatile storage or nonvolatile memory that can store information, and may be removable.

[0029] The communication device 15 transmits and receives data to and from other computers, such as user terminals or servers, via a network, and is, for example, a wireless LAN module. The communication device 15 may be a device or module for other wireless communication, such as a Bluetooth (registered trademark) module, or a device or module for wired communication, such as an Ethernet (registered trademark) module or a USB interface. In one example, the sensor 2 transmits an acquired observation signal to the biological signal extraction device 3, and the biological signal extraction device 3 can acquire the observation signal via the communication device 15. Note that if the user's input and output to and from the biological signal extraction device 3 are only via the communication device 15, the biological signal extraction device 3 may not include the input device 12 and the display device 13.

[0030] 5 is a functional block diagram of a biosignal extraction device 3 according to one embodiment of the present invention. The biosignal extraction system 1 includes a processing signal generation unit 21 and a biosignal extraction unit 22. The processing signal generation unit 21 and the biosignal extraction unit 22 can be realized, for example, by a program being executed by the processor 11, and therefore part or all of one part (function) may be possessed by another part. However, these functions may also be realized by hardware by configuring electronic circuits or the like to realize part or all of each function. The biosignal extraction device 3 may also include other functions.

[0031] As described above, the observed signal acquired by the sensor 2 includes two source signals: stationary noise including non-stationary noise, and pulse wave including non-stationary noise. One signal acquired by the sensor 2 is an analog signal. The processing signal generating unit 21 converts the signal acquired by the sensor 2 into an analog signal at a predetermined sampling frequency F S 1 twice the sampling frequency F S The observation signal obtained by the processing signal generation unit 21 is obtained by oversampling the digital signal at a sampling frequency F S 2, and the sampling frequency F S The data includes sampling points (samples) with a number of samples corresponding to a given sampling frequency F S 1 is the sampling frequency that has been used to acquire the observed signal. In this specification, the predetermined sampling frequency F S Sampling at a sampling frequency greater than 1 is called oversampling. S 1 twice the sampling frequency F S The observed signal obtained by sampling at a predetermined sampling frequency F S It contains twice as many samples as the observed signal sampled at 1.

[0032] The processing signal generator 21 generates one observation signal (S O ) based on two signals (SO1 , S O2 The processing signal generation unit 21 generates the observation signal (S O ) sampling points at a predetermined sampling frequency F S A signal (S O1 ) is generated, and the processing signal generation unit 21 acquires the observed signal (S O ) sampling points at a predetermined sampling frequency F S 1, the number of samples corresponding to the signal in question is equal to or more than the number of other signals (S O2 ) is generated. More specifically, the processing signal generation unit 21 generates the observed signal (S O ) sampling points, one signal (S O1 ) is generated, and the processing signal generation unit 21 acquires the observed signal (S O ) sampling points of the other signal (S O2 At this time, the processing signal generating unit 21 generates one signal (S O1 ) and other signals (S O2 ) can be regarded as signals of the same time. In other words, the processing signal generating unit 21 generates two signals so that one signal (S O1 ) and other signals (S O2 ) is generated as a simultaneous signal.

[0033] In one example, the processing signal generator 21 generates a signal (S O1 ) and other signals (S O2 ) are signals at the same time, O1 ) and other signals (S O2) can adjust the time of the sampling point of at least one of the signals. For example, the processing signal generation unit 21 can adjust the sampling point of one signal so that the corresponding sampling points of the two signals are sampling points at the same time. For example, the processing signal generation unit 21 can adjust the time of the sampling point of the signal acquired by the sensor 2 at a predetermined sampling frequency F S 1 twice the sampling frequency F S 2, and then divide it into two series at every other sampling point, and change the time of the sampling point of one signal so that the sampling points of the two series are at the same time, thereby generating two signals with sampling points at the same time. In this case, for example, if the sampling frequency is F S The kth (k=0, 1, …, n+1) sampling point among the sampling points sampled in 2 is S k Then, S0 is sorted into the A series and S1 into the B series, and S1 is set as the sampling point at time t0 when S0 is acquired, S2 is sorted into the A series and S3 into the B series, and S3 is set as the sampling point at time t2 when S2 is acquired. By repeating this process, two series of signals with sampling points at the same time, the A series signal (S0, S2, ..., S n ) and B series signals (S1, S3, ..., S n+1 ) is generated.

[0034] The biological signal extraction unit 22 extracts the two signals (x (1) (n), x (2) (n)) to extract the pulse wave.

[0035] Next, an example of a processing model M will be described, which explains the functions of the processing signal generation unit 21 and the biological signal extraction unit 22. Fig. 6 is a diagram showing an example of the processing model M of the processing signal generation unit 21 and the biological signal extraction unit 22 according to an embodiment of the present invention. In the processing model M, the processing signal generation unit 21 performs (a) oversampling, and the biological signal extraction unit 22 performs other processing (b) to (h). The processing of the biological signal extraction unit 22 can be similar to the processing in a known BSS.

[0036] FIG. 7 is a diagram showing a model of oversampling in the processing model M of FIG. The observed signal x(n) is expressed by equation (5). TIFF2025168870000006.tif7150(5) where n is a time parameter. x(n) is a predetermined sampling frequency F used in the conventional method of extracting a desired pulse wave using two sensors placed at different locations. S 1 shows the observed signal obtained.

[0037] 6(a), the processing signal generator 21 generates two signals by dividing the observed signal x2(n) obtained by sampling at twice the sampling frequency of x(n) into two signals at each point. The observed signal x2(n) is expressed by equation (6). TIFF2025168870000007.tif12150(6) In addition, the signal generated by dividing x2(n) into two signals at each point is called x (1) (n), x (2) (n), x2(n) is expressed by equation (7). TIFF2025168870000008.tif7150(7)

[0038] FIG. 8 shows a process in which the processing signal generator 21 of an embodiment of the present invention generates two signals (x2(n)) based on one observed signal (x2(n)). (1) (n), x (2)The processing signal generator 21 generates a signal (x2(n)) that includes every other sampling point among the sampling points of the observed signal (x2(n)). (1) (n)) and generates another signal (x2(n)) that includes alternate sampling points different from the one signal among the sampling points of the observed signal (x2(n)). (2) (n)).

[0039] In the processing model M, as explained in Figure 2, the observed signal x2(n) is a convolution mixture model of two source signals: a pulse wave containing non-stationary noise and stationary noise containing non-stationary noise. The observed signal x2(n) is expressed by the mixing matrix A 2,2 (m) and the source signal vector s2(nm) are expressed by equation (8). TIFF2025168870000009.tif17150(8) Here, the mixing matrix A 2,2 Since (m) and the source signal vector s2(nm) are expressed by equations (9) and (10), equation (8) can be expressed as equation (11). TIFF2025168870000010.tif12150(9) TIFF2025168870000011.tif7150(10) TIFF2025168870000012.tif14150(11)

[0040] In the block extraction shown in FIG. 6 (b), the biological signal extraction unit 22 blocks each piece of data for the observed signal. In processing model M, data blocking is necessary because processing is performed in the frequency domain. When each observed signal consists of L pieces of data, 2L pieces of data for the observed signal are used in overlapping block processing to generate L pieces of pulse wave data for each block. If the block number is l (lowercase L), the matrix obtained by extracting x2(n) into 2L blocks from n = lL-L / 2 to n = lL+L-1+L / 2 is expressed by equation (12). TIFF2025168870000013.tif13150(12) where the row vector TIFF2025168870000014.tif8150 is expressed by equation (13), where i indicates the block number. TIFF2025168870000015.tif9150(13)

[0041] In the discrete Fourier transform (c) shown in FIG. 6, the biological signal extraction unit 22 calculates the following as shown in equation (14): Perform a discrete Fourier transform on TIFF2025168870000016.tif8150, Calculate TIFF2025168870000017.tif8150. TIFF2025168870000018.tif9150(14) TIFF2025168870000019.tif8150 can also be expressed by equation (15). TIFF2025168870000020.tif9150(15) In this case, it represents a column vector TIFF2025168870000021.tif8150 is expressed by equation (16), where k is the frequency. TIFF2025168870000022.tif9150(16)

[0042] In the signal separation (d) shown in FIG. 6, the biological signal extraction unit 22 extracts the separated signal vectors using a separation matrix as shown in Equation (17). TIFF2025168870000023.tif8150 is calculated, which separates the pulse wave containing non-stationary noise from stationary noise containing non-stationary noise. TIFF2025168870000024.tif9150(17) Here, the separation matrix TIFF2025168870000025.tif8150 is expressed by equation (18), where k is the frequency. TIFF2025168870000026.tif16150(18) Here, a known method such as the method by Amari et al. shown in Non-Patent Document 2 can be used for updating the separating matrix, and it can be expressed by equation (19). TIFF2025168870000027.tif14150(19) Here, η represents the step gain. The nonlinear function is expressed by equation (20) because the probability density function of the pulse wave, which is the desired signal, is a super-Gaussian distribution. TIFF2025168870000028.tif9150(20) where R is the real part and I is the imaginary part.

[0043] In the non-stationary noise removal shown in Fig. 6(e), the biological signal extraction unit 22 removes non-stationary noise from the separated signal. Separation signal matrix composed of TIFF2025168870000029.tif8150 TIFF2025168870000030.tif8150 is expressed by equation (21). TIFF2025168870000031.tif9150(21) where the row vector TIFF2025168870000032.tif8150 is expressed by equation (22), where i is the signal sequence number. TIFF2025168870000033.tif9150(22) Next, The amplitude spectrum vector of TIFF2025168870000034.tif8150 is If the image is TIFF2025168870000035.tif9150, the relationship in equation (23) holds. TIFF2025168870000036.tif9150(23) If the block length is long, the amplitude of the non-stationary noise will be small. Non-stationary noise is removed by performing the process shown in equation (24) on each component of TIFF2025168870000037.tif9150, where th is a threshold value. TIFF2025168870000038.tif15150(24)

[0044] 9 is a diagram for explaining an outline of how the biological signal extraction unit 22 of one embodiment of the present invention removes non-stationary noise using a threshold value. Amplitude spectrum after removing non-stationary noise. The frequency spectrum corresponding to TIFF2025168870000039.tif8150 If the image is TIFF2025168870000040.tif9150, the relationship in equation (25) holds. TIFF2025168870000041.tif9150(25)

[0045] In the inverse discrete Fourier transform (f) shown in FIG. 6, the biological signal extraction unit 22 calculates the following as shown in equation (26): Perform a discrete Fourier transform on TIFF2025168870000042.tif9150, Calculate TIFF2025168870000043.tif9150. TIFF2025168870000044.tif9150(26)

[0046] In the selection of the (g) signal shown in Figure 6, TIFF2025168870000045.tif8150 and TIFF2025168870000046.tif9150 is a signal indicating a pulse wave, and the other is a signal indicating stationary noise, so the biological signal extraction unit 22 selects the signal indicating the pulse wave. The biological signal extraction unit 22 focuses on the fact that the amplitude spectrum of the pulse wave has regularity in its peaks, and selects the signal having regularity in its frequency peaks as the pulse wave signal.

[0047] In the extraction of element (h) shown in Figure 6, the processing model M is a block process with overlaps, so It is necessary to extract the central L signals from TIFF2025168870000047.tif9150. Therefore, the signal indicating the pulse wave TIFF2025168870000048.tif8150 is expressed by equation (27). TIFF2025168870000049.tif8150(27)

[0048] FIG. 10 is a flowchart showing the operation of the biological signal extraction system 1 according to one embodiment of the present invention.

[0049] In step 101, the sensor 2 acquires an observation signal. In step 102, the processing signal generation unit 21 generates two signals based on one observation signal acquired by the sensor 2. In step 103, the biological signal extraction unit 22 extracts a pulse wave from the two signals generated by the processing signal generation unit 21 using BSS.

[0050] Next, the following simulation experiment will be used to demonstrate that the biosignal extraction device 3 according to the embodiment of the present invention can extract pulses with accuracy equal to or higher than that of conventional methods. In this simulation experiment, a signal acquired by a piezoelectric pulse wave sensor attached to the wrist was used as the observed signal. Other experimental parameters are as shown in Table 1 below. Table 1 TIFF2025168870000050.tif53155

[0051] In this simulation experiment, evaluation was performed using cosine similarity. Similarity sim is expressed as in equation (28). TIFF2025168870000051.tif15150(28) Similarity is an index that measures the similarity between two vectors, and is 1 if the two vectors match perfectly.

[0052] FIG. 11 shows the results of a simulation experiment. (i) in FIG. 11 is a pulse wave obtained using the biosignal extraction system 1 according to one embodiment of the present invention, (iv) is a signal obtained from a piezoelectric pulse wave sensor attached to the subject's wrist, i.e., an observed signal, and (v) is a signal obtained from the index finger for similarity confirmation. Noise can be seen in the observed signal. (ii) in FIG. 11 is a pulse wave obtained using a conventional method using QMF, and (iii) is a pulse wave obtained using a conventional method using two sensors. The similarities between (i), (ii), and (iii) in FIG. 11 and (v) are 0.676, 0.641, and 0.587, respectively. The waveform and similarity results shown in FIG. 11 confirm that the biosignal extraction system 1 according to one embodiment of the present invention can extract pulse waves with accuracy equal to or higher than that of conventional methods.

[0053] Next, the effects of the biological signal extraction system 1 according to the embodiment of the present invention will be described. Conventionally, there is a method for extracting a pulse wave from two observation signals acquired using two sensors placed at different positions, but this method has the problem of difficulty in determining the appropriate distance between the sensors. To solve this problem, the present inventors have proposed a method using a quadrature mirror filter (QMF) to enable pulse wave extraction from one observation signal acquired using one sensor. However, this method has the problem that adjusting the filter characteristics when using QMF is complicated and difficult. In the embodiment of the present invention, the processing signal generator 21 converts one observation signal acquired by the sensor 2 into a signal having a predetermined sampling frequency F, which is a sampling frequency that has been used to acquire conventional observation signals. S 1 twice the sampling frequency F S 2, one observed signal (x2(n)) is obtained. The processing signal generator 21 generates one signal (x (1)(n)) and generates another signal (x2(n)) that includes alternate sampling points different from the one signal among the sampling points of the observed signal (x2(n)). (2) (n)) where one signal (x (1) (n)) and other signals (x (2) (n)) are generated as simultaneous signals. The predetermined sampling frequency F used to acquire the conventional observation signal S If 1 is a frequency at which the observed signal can be restored with high accuracy, then, considering the sampling theorem, a given sampling frequency F S A signal (x) sampled at 1 (1) (n)) and other signals (x (2) In the embodiment of the present invention, since the pulse wave is a signal that has periodicity and does not have symmetry within one cycle, has regular peaks, and the amplitude at the peak is steep, the sampling theorem is not satisfied around the peak, and therefore, a predetermined sampling frequency F S A signal (x) sampled at 1 (1) (n)) and other signals (x (2) In the embodiment of the present invention, a signal (x2(n)) generated at substantially the same time from one observed signal (x2(n)) is used as an input signal for the BSS. (1) (n)) and other signals (x (2) (n)) can be treated as simultaneous signals. (1) (n)) and other signals (x (2) By generating (n)) as simultaneous signals (treating them as simultaneous signals), a change occurs in the frequency characteristics, making these two signals more suitable as input signals for BSS. In view of the above circumstances, the biological signals targeted by the biological signal extraction system 1 of the embodiment of the present invention are periodic signals that do not have symmetry within one cycle, and have regular peaks with steep peak amplitudes, such as pulse waves and electrocardiograms. With this configuration, in this embodiment, it is possible to generate two signals required for using BSS to measure a pulse wave based on a single observed signal without requiring complex adjustments such as adjustment of filter characteristics, etc. This makes it possible to extract a predetermined biological signal based on a single observed signal in a simpler manner. This type of processing, which focuses on waveform characteristics, changes the sampling interval of one observed signal, generates two signals from one observed signal, and uses BSS, is something that has never been done before.

[0054] The above-described effects and advantages are similar to those of other embodiments and examples unless otherwise specified.

[0055] In an embodiment of the present invention, the present invention may be a biosignal extraction system 1, a biosignal extraction device 3, or a method for realizing the functions of the embodiment of the present invention described above and the processing shown in the flowcharts. Furthermore, in an embodiment of the present invention, the present invention may be a program for realizing the functions of the embodiment of the present invention described above and the information processing shown in the flowcharts, or a computer-readable storage medium storing the program. Furthermore, in an embodiment of the present invention, the present invention may be a server that can supply a computer with a program for realizing the functions of the embodiment of the present invention described above and the information processing shown in the flowcharts. Furthermore, in an embodiment of the present invention, the present invention may be a virtual machine that realizes the functions of the embodiment of the present invention described above and the information processing shown in the flowcharts.

[0056] In the embodiment of the present invention, the biological signal extraction system 1 may be a system for extracting a pulse wave at the wrist, or may be a system for extracting a pulse wave at the ankle.

[0057] In an embodiment of the present invention, the biological signal extraction system 1 may be a system for extracting other predetermined biological signals, such as an electrocardiogram, rather than a system for extracting a pulse wave. However, the predetermined biological signal is a signal that has periodicity, does not have symmetry within one cycle, has regular peaks, and the peak amplitudes are steep. It is understood that the location of the sensor 2 may be changed as appropriate depending on the type of the predetermined biological signal.

[0058] In one or more embodiments of the present invention, if the observation signal obtained by the sensor 2 has already been obtained, the flowchart of FIG.

[0059] In one or more embodiments of the present invention, the biological signal extraction system 1 may include a known amplifier circuit for amplifying a voltage signal acquired by the sensor 2 and a known filter circuit for extracting a signal of a predetermined frequency from the voltage signal. The biological signal extraction system 1 may also include a known transmitter for wirelessly transmitting the amplified and filtered signal to the outside.

[0060] In one or more embodiments of the present invention, the sensor 2 may be a digital sensor or a sensor that stores signals in digital format, rather than an analog sensor. In this embodiment, the sensor 2 may be configured to transmit the acquired observation signals to the biological signal extraction device 3 via wired or wireless communication, or the processing signal generation unit 21 may be configured to acquire the observation signals from the sensor 2 via an external storage device. For example, when the processing signal generation unit 21 is configured to acquire the observation signals from the sensor 2 via an external storage device, the sensor 2 does not need to be communicatively connected to the biological signal extraction device 3. In this embodiment, the observation signals acquired by the processing signal generation unit 21 are sampled at a sampling frequency F S2. In this embodiment, the biological signal extraction device 3 can also have some of the functions of the sensor 2 by including a software module for controlling the sensor 2. Note that, if the biological signal extraction device 3 does not communicate with other devices, as in this embodiment, the biological signal extraction device 3 does not need to include the communication device 15.

[0061] In one or more embodiments of the present invention, the processing signal generator 21 converts the signal acquired by the sensor 2 into a signal at a predetermined sampling frequency F S Sampling frequency F greater than 2 times 1 S By sampling at 2, the observed signal of the digital signal can be obtained.

[0062] In one or more embodiments of the present invention, the processing signal generation unit 21 selects a predetermined sampling frequency F from among the sampling points of the observation signal (x2(n)) acquired by the processing signal generation unit 21. S A signal (x) containing sampling points equal to or greater than the number of samples corresponding to 1 (1) (n)) is generated, and a predetermined sampling frequency F is generated from the sampling points of the observation signal (x2(n)) acquired by the processing signal generation unit 21. S Other signals (x) that contain sampling points different from the signal in question, the number of samples corresponding to 1 or more (2) (n)) can be generated. (1) (n)) and other signals (x (2) (n)) do not contain the same sampling points and are at a given sampling frequency F S As long as the signal includes sampling points equal to or greater than the number of samples corresponding to 1 and can be treated as signals at the same time, the signal does not have to include alternate sampling points. For example, the processing signal generating unit 21 may generate a signal obtained by the sensor 2 at a sampling frequency F S It is also possible to generate two signals by sampling at 2 and dividing every other sample into two series according to a predetermined rule, rather than dividing every other sample into two series.

[0063] In the processes or operations described above, the processes or operations can be freely changed as long as no inconsistencies in the processes or operations occur, such as the use of data that should not yet be available in a certain step. Furthermore, the embodiments described above are merely examples for explaining the present invention, and the present invention is not limited to these embodiments. The present invention can be embodied in various forms without departing from the spirit of the invention. [Explanation of symbols]

[0064] 1: Biological signal extraction system, 2: Sensor, 3: Biological signal extraction device, 11: Processor, 12: Input device, 13: Display device, 14: Storage device, 15: Communication device, 16: Bus, 21: Processing signal generation unit, 22: Biological signal extraction unit

Claims

1. 1. A method for extracting a predetermined biological signal, comprising: generating two signals based on one observation signal acquired using a sensor; extracting the predetermined biological signal from the two generated signals using blind source separation; Including, generating the two signals includes generating one signal including, among sampling points of the one observation signal, a number of sampling points equal to or greater than a predetermined sampling frequency, and generating another signal including, among sampling points of the one observation signal, a number of sampling points different from the one signal equal to or greater than a predetermined sampling frequency; method.

2. 2. The method of claim 1, wherein generating the two signals includes generating one signal including every other sampling point of the one observation signal, and generating another signal including every other sampling point of the one observation signal that is different from the one signal, wherein the one signal and the other signal are generated as signals at the same time.

3. 2. The method according to claim 1, wherein the one observation signal is a signal acquired using a sensor sampled at a frequency equal to or greater than twice the predetermined sampling frequency, or a signal acquired using a sensor having a number of samples corresponding to a frequency equal to or greater than twice the predetermined sampling frequency.

4. The method according to claim 1 , wherein the predetermined biological signal is a signal that has periodicity and does not have symmetry within one period.

5. The method according to claim 1 , wherein the predetermined biological signal is a pulse wave or an electrocardiogram signal.

6. A program that causes a computer to execute the method according to any one of claims 1 to 5.

7. A biological signal extraction device that extracts a predetermined biological signal, a signal generating unit that generates two signals based on one observation signal acquired using a sensor; a signal extraction unit that extracts the predetermined biological signal from the two signals generated by the signal generation unit using blind source separation; Equipped with the signal generator generates one signal including, among the sampling points of the one observation signal, a number of sampling points equal to or greater than a predetermined sampling frequency, and generates another signal including, among the sampling points of the one observation signal, a number of sampling points different from the one signal equal to or greater than a predetermined sampling frequency. Biosignal extraction device.

8. 8. The biological signal extraction device according to claim 7, wherein the signal generation unit generates one signal including every other sampling point of the one observation signal, and generates another signal including every other sampling point of the one observation signal that is different from the one signal, and the one signal and the other signal are generated as signals at the same time.

9. A biological signal extraction system for extracting a predetermined biological signal, a sensor for acquiring an observation signal; a signal generating unit that generates two signals based on one observation signal acquired by the sensor; a signal extraction unit that extracts the predetermined biological signal from the two signals generated by the signal generation unit using blind source separation; Equipped with the signal generator generates one signal including sampling points of the one observation signal that have a sampling frequency equal to or higher than a predetermined sampling frequency, and generates another signal including sampling points of the one observation signal that have a sampling frequency equal to or higher than the predetermined sampling frequency and that are different from the one signal; Biosignal extraction system.

10. 10. The biological signal extraction system according to claim 9, wherein the signal generation unit generates one signal including every other sampling point of the one observation signal, and generates another signal including every other sampling point of the one observation signal that is different from the one signal, and the one signal and the other signal are generated as signals at the same time.

Images