Measuring device and measuring method

Abstract

A measurement device according to the present invention comprises: a sensor capable of generating a first biological signal in accordance with first biological information; a processing unit capable of generating a second biological signal by performing noise reduction processing on the basis of the first biological signal; and a storage unit capable of storing the processing result of the noise reduction processing. The processing unit is capable of using the processing result stored in the storage unit to perform noise reduction processing on the basis of the first biological signal.

Description

[Technical Field]

[0001] The present disclosure relates to a measurement device that measures a biological signal corresponding to biological information, and a measurement method that measures a biological signal corresponding to a biological state. [Background technology]

[0002] In the medical field, biological signals corresponding to biological information from living organisms such as the human body are often detected. For example, Patent Document 1 discloses a device that reduces noise components contained in biological signals obtained by a sensor and estimates brain activity based on the biological signals with the noise components reduced. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2001-000407 Summary of the Invention

[0004] Thus, it is desirable to reduce noise components in measurement devices, and effective reduction of noise components is expected.

[0005] It is desirable to provide a measurement device and a measurement method that can effectively reduce noise components.

[0006] A measurement device according to an embodiment of the present disclosure includes a sensor, a processing unit, a noise detection unit; The sensor is capable of generating a first biological signal corresponding to the first biological information, and the processing unit is capable of generating a second biological signal by performing noise reduction processing on the first biological signal. The noise detection unit is capable of calculating the degree of noise that is the amount of noise that is included in the first biological signal according to the second biological information. The storage unit is capable of storing the processing result of the noise reduction processing. If the noise level is greater than a predetermined value, It is possible to perform noise reduction processing based on the first biological signal using the processing results stored in the storage unit. Therefore, when the noise level is smaller than a predetermined value, it is possible to perform noise reduction processing based on the first biological signal. The noise reduction processing includes generating a first matrix by separating the first biological signal into a plurality of signals, performing matrix separation processing to separate the first matrix into a second matrix having components with little change in the time axis direction and a third matrix having components that are discrete in the time axis direction, and generating a second biological signal based on the second matrix.

[0007] A measurement method according to an embodiment of the present disclosure includes generating a first biological signal according to first biological information; Calculating a noise level of the first biological signal including noise according to the second biological information; The method includes performing noise reduction processing based on the first biological signal to generate a second biological signal, and storing the processing result of the noise reduction processing. If the noise level is greater than a predetermined value, Based on the first biological signal, a stored processing result is used. , when the noise level is smaller than a predetermined value, the noise reduction process is performed based on the first biological signal. The noise reduction process includes generating a first matrix by separating the first biological signal into a plurality of signals, performing a matrix separation process to separate the first matrix into a second matrix having components with little change in the time axis direction and a third matrix having components that are discrete in the time axis direction, and generating a second biological signal based on the second matrix.

[0008] In a measurement device and a measurement method according to an embodiment of the present disclosure, a first biological signal corresponding to first biological information is generated, and a second biological signal is generated by performing noise reduction processing based on the first biological signal. The processing result of the noise reduction processing is then stored. The noise reduction processing is performed based on the first biological signal using the stored processing result. [Brief explanation of the drawings]

[0009] [Figure 1] 1 is a block diagram illustrating an example configuration of a measurement device according to an embodiment of the present disclosure. [Figure 2] FIG. 2 is an explanatory diagram illustrating an example of the configuration of the sensor illustrated in FIG. [Figure 3A] 2 is a waveform diagram illustrating an example of a potential signal shown in FIG. 1. [Figure 3B] 2 is a waveform diagram illustrating an example of a potential signal shown in FIG. 1. [Figure 4] 2 is a block diagram illustrating an example of the configuration of a processing unit illustrated in FIG. 1. [Figure 5] 5 is an explanatory diagram illustrating an example of a matrix generated by the signal separation transform unit shown in FIG. 4. FIG. [Figure 6] 5 is an explanatory diagram illustrating an example of an operation of the matrix transformation unit illustrated in FIG. 4. [Figure 7] 2 is a flowchart illustrating an example of an operation of the measurement device shown in FIG. [Figure 8] 10 is an explanatory diagram illustrating an example of a matrix generated by a signal separation transform unit according to a reference example. FIG. [Figure 9] 10A and 10B are explanatory diagrams illustrating an example of an operation of a matrix transformation unit according to a reference example. [Figure 10] 5 is an explanatory diagram illustrating another example of the operation of the matrix transformation unit shown in FIG. [Figure 11] FIG. 10 is a block diagram illustrating an example of the configuration of a measurement device according to a modified example. [Figure 12] 12 is a block diagram illustrating an example of the configuration of a processing unit illustrated in FIG. 11. FIG. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings.

[0011] <Embodiment> [Configuration example] 1 shows an example of the configuration of a measurement device (measurement device 1) according to one embodiment. Measurement device 1 is an electroencephalogram potential measurement device that measures electroencephalograms in the human body. Measurement device 1 includes a sensor 11, a noise detection unit 12, a processing unit 13, and a storage unit 14. The noise detection unit 12, the processing unit 13, and the storage unit 14 are configured using, for example, a processor, a memory, etc.

[0012] The sensor 11 is configured to generate an electric potential signal E1 corresponding to an electroencephalogram in the human body when it is brought into contact with, for example, the head or ear of the human body. The electric potential signal E1 is a signal (electric potential signal E1(t)) that changes with time t. The measurement device 1 is a single-channel measurement device that performs processing based on such one electric potential signal E1.

[0013] 2 shows an example configuration of the sensor 11. The sensor 11 has, for example, electrodes 21 and 22 and an amplifier 23. The electrodes 21 and 22 are configured to be brought into contact with the head 100 of a human body that is the measurement target. The amplifier 23 is configured to amplify the signals obtained by the electrodes 21 and 22. The sensor 11 is configured to generate an electric potential signal E1 corresponding to neural activity of the brain 101 in the head 100.

[0014] Muscle fibers 102 are present near the surface of the head 100. The muscle fibers 102 contract and relax based on action potentials generated by motor neurons. The potential signal E1 generated by the sensor 11 may contain signal components corresponding to the action potentials of the muscle fibers 102, in addition to signal components corresponding to neural activity in the brain 101. Since the measurement device 1 measures electroencephalograms, the signal components corresponding to the action potentials of the muscle fibers 102 are noise (hereinafter also referred to as myoelectric noise).

[0015] 3A and 3B show an example of myoelectric noise in the potential signal E1.

[0016] For example, when a person performs an action that moves the skeleton, such as chewing, a large action potential is generated in multiple muscle fibers 102. As a result, as shown in part W1 in Fig. 3A, a signal component with a large amplitude intensity is generated in the potential signal E1. That is, in the potential signal E1, a signal component with a large amplitude intensity corresponding to the action potential of the muscle fibers 102 is superimposed for a short period of time on the signal component of the electroencephalogram that appears continuously.

[0017] Furthermore, for example, when a person changes their facial expression, such as by distorting their face, small action potentials are generated in a smaller number of muscle fibers 102 closer to the surface of the head 100. As a result, as shown in part W2 in Fig. 3B, signal components with small amplitude intensity are generated in the potential signal E1. That is, in the potential signal E1, signal components with small amplitude intensity corresponding to the action potentials of the muscle fibers 102 are superimposed for a short period of time on the signal components of the electroencephalograms that appear continuously.

[0018] In this way, myoelectric noise of various amplitudes and intensities may be superimposed on the potential signal E1. Furthermore, although the myoelectric noise occurred for a short period of time in the examples of Figures 3A and 3B, it is also possible for the myoelectric noise to occur continuously over a longer period of time. The measuring device 1 is capable of generating a potential signal E2 by reducing such various myoelectric noises from a single potential signal E1.

[0019] The noise detection unit 12 is configured to calculate the degree to which myoelectric noise is included in the potential signal E1. In this example, the noise detection unit 12 calculates a time ratio RN of the period during which the potential signal E1 includes myoelectric noise to the entire period of the potential signal E1. Specifically, the noise detection unit 12 monitors the potential signal E1 for a period having a time length T, and calculates a time ratio RN of the period during which myoelectric noise occurs in the potential signal E1. The noise detection unit 12 then supplies information about this time ratio RN to the processing unit 13 and the storage unit 14.

[0020] The processing unit 13 (FIG. 1) is configured to perform noise reduction processing based on the potential signal E1 supplied from the sensor 11, thereby generating a potential signal E2. The processing unit 13 performs noise reduction processing based on the potential signal E1 when the time proportion RN of noise obtained by the noise detection unit 12 is equal to or smaller than a predetermined value TH. This predetermined value TH can be set to, for example, 20%. Furthermore, when the time proportion RN of noise is greater than the predetermined value TH, the processing unit 13 performs noise reduction processing based on the potential signal E1 using the potential signal E2 supplied from the storage unit 14. In this way, when the potential signal E1 contains a large amount of myoelectric noise, the processing unit 13 performs noise reduction processing using the potential signal E2 supplied from the storage unit 14.

[0021] Fig. 4 shows an example of the configuration of the processing unit 13. For ease of explanation, Fig. 4 also shows the memory unit 14. The processing unit 13 has a signal generation unit 31 and a noise reduction processing unit 39. The noise reduction processing unit 39 has a signal separation transformation unit 32, a matrix transformation unit 33, a threshold processing unit 34, a matrix inverse transformation unit 35, and a signal separation inverse transformation unit 36.

[0022] The signal generating unit 31 is configured to generate a potential signal E31 to be input to the noise reduction processing unit 39 based on the potential signal E1. Specifically, when the time proportion RN of noise obtained by the noise detecting unit 12 is equal to or smaller than a predetermined value TH, the signal generating unit 31 outputs the potential signal E1 as the potential signal E31 without modification. When the time proportion RN of noise obtained by the noise detecting unit 12 is greater than the predetermined value TH, the signal generating unit 31 combines the potential signal E1 and the potential signal E2 supplied from the memory unit 14 on the time axis to generate the potential signal E31.

[0023] The signal separation conversion unit 32 is configured to generate a matrix D by separating the potential signal E31 into signals of a plurality of frequency components. Specifically, the signal separation conversion unit 32 generates absolute values ​​d(t, f) of a spectrogram by performing a short-time Fourier transform based on the potential signal E31 over a period of time T (for example, one minute). Here, t is time and f is frequency. Then, the signal separation conversion unit 32 generates a matrix D having the absolute values ​​d(t, f) as matrix elements.

[0024] FIG. 5 shows a schematic representation of matrix D. The horizontal axis represents time t, and the vertical axis represents frequency f. Matrix D contains arrays of spectrogram absolute values ​​d(t,f), which are matrix elements. In FIG. 5, the magnitude of the matrix element values ​​is indicated by shading and dots.

[0025] Matrix D includes the characteristics of electroencephalograms and the characteristics of myoelectric noise. Since electroencephalograms occur continuously in the time axis direction, the components of electroencephalograms in matrix D change little in the time axis direction. Furthermore, the components of electroencephalograms are larger at lower frequencies and smaller at higher frequencies. Thus, the components of electroencephalograms have commonality in the time axis direction and the frequency axis direction. Therefore, matrix D has the characteristics shown by the shading in FIG. 5. Light shading indicates small matrix element values, and dark shading indicates large matrix element values. On the other hand, electroencephalogram noise occurs discretely, for example, in the time axis direction and the frequency axis direction. In other words, the components of electroencephalogram noise have sparsity in the time axis direction and the frequency axis direction. Therefore, matrix D has the characteristics shown by the dots in FIG. 5. Light dots indicate small matrix element values, and dark dots indicate large matrix element values.

[0026] The signal separation conversion unit 32 generates such a matrix D. Then, the signal separation conversion unit 32 supplies the generated matrix D to the matrix conversion unit 33.

[0027] The matrix transformation unit 33 (FIG. 4) is configured to generate a low-rank matrix L and a sparse matrix S by performing a matrix separation process based on the matrix D. The matrix D is expressed as the sum of the low-rank matrix L and the sparse matrix S as shown below. D = L + S The low-rank matrix L has matrix elements that change little in the time axis direction. This low-rank matrix L is a matrix with few eigenvalues ​​calculated by principal component analysis. The sparse matrix S has matrix elements that are discrete in the time axis direction and the frequency axis direction. This sparse matrix S is a matrix with few matrix elements that are not zero. The matrix conversion unit 33 separates the matrix D into the low-rank matrix L and the sparse matrix S so that the following value is minimized: ||L|| * + λ||Ψ(S)||1 Here, “||L|| * " is the nuclear norm of the low-rank matrix L. This "||L|| *" is the sum of the singular values ​​of the low-rank matrix L. In this example, Ψ is a transformation that generates a matrix by calculating the difference between multiple matrix elements of the sparse matrix S in the frequency axis direction. This transformation Ψ emphasizes the sparse components in the sparse matrix S. "||Ψ(S)||1" is the L1 norm of the matrix generated by the transformation Ψ. This "||Ψ(S)||1" is the sum of the absolute values ​​of the elements of the matrix generated by the transformation Ψ. λ is an adjustment parameter for the matrix separation process and is a scalar quantity indicating a predetermined value between "0" and "1".

[0028] FIG. 6 schematically illustrates an example of matrix separation processing in the matrix transformation unit 33. The matrix transformation unit 33 performs matrix separation processing based on the matrix D to generate a low-rank matrix L and a sparse matrix S. The low-rank matrix L has the characteristics indicated by the hatched areas in FIG. 5 and mainly includes electroencephalogram components. The sparse matrix S has the characteristics indicated by the dotted areas in FIG. 5 and mainly includes electromyogram noise components. The adjustment parameter λ can be adjusted in this manner so that the electroencephalogram components and the electromyogram noise components are separated into the low-rank matrix L and the sparse matrix S, respectively.

[0029] In this way, the matrix transformation unit 33 performs matrix separation processing to generate a low-rank matrix L that mainly contains electroencephalogram components and a sparse matrix S that mainly contains electromyographic noise components.

[0030] The thresholding unit 34 (FIG. 4) is configured to generate a low-rank matrix L' and a sparse matrix S' based on the low-rank matrix L and the sparse matrix S generated by the matrix transformation unit 33. Specifically, the thresholding unit 34 outputs the low-rank matrix L as the low-rank matrix L', and generates the sparse matrix S' by multiplying the sparse matrix S by a zero matrix. The low-rank matrix L' and the sparse matrix S' can be expressed as follows: L´ = L S´ = 0

[0031] The matrix inversion unit 35 is configured to calculate the matrix D' by calculating the sum of the low-rank matrix L' and the sparse matrix S'. In this example, the sparse matrix S' is a zero matrix, so the matrix D' is equal to the low-rank matrix L'.

[0032] The signal separation and inverse transformation unit 36 ​​is configured to generate a potential signal E2 by performing an inverse transformation of the short-time Fourier transform performed by the signal separation and transformation unit 32, based on the matrix D'. The potential signal E2 is a signal (potential signal E2(t)) that changes according to time t during a period of time length T (for example, one minute).

[0033] With this configuration, the noise reduction processing unit 39 generates the potential signal E2 based on the low-rank matrix L obtained by the matrix separation processing and the low-rank matrix L out of the sparse matrix S. In this way, the noise reduction processing unit 39 generates the potential signal E2 by reducing the myoelectric noise components contained in the potential signal E1.

[0034] The storage unit 14 (FIG. 2) is configured to store the potential signal E2 generated by the processing unit 13. Specifically, the storage unit 14 stores the potential signal E2 generated by the processing unit 13 when the time ratio RN of noise obtained by the noise detection unit 12 is equal to or smaller than a predetermined value TH. The storage unit 14 then supplies the stored potential signal E2 to the processing unit 13 when the time ratio RN of noise is greater than the predetermined value TH.

[0035] Here, the sensor 11 corresponds to a specific example of a "sensor" in the present disclosure. The electric potential signal E1 corresponds to a specific example of a "first biological signal" in the present disclosure. The processing unit 13 corresponds to a specific example of a "processing unit" in the present disclosure. The matrix D corresponds to a specific example of a "first matrix" in the present disclosure. The matrix D corresponds to a specific example of a "first matrix" in the present disclosure. The low-rank matrix L corresponds to a specific example of a "second matrix" in the present disclosure. The sparse matrix S corresponds to a specific example of a "third matrix" in the present disclosure. The electric potential signal E2 corresponds to a specific example of a "second biological signal" in the present disclosure. The memory unit 14 corresponds to a specific example of a "memory unit" in the present disclosure. The noise detection unit 12 corresponds to a specific example of a "noise detection unit" in the present disclosure.

[0036] [Actions and Actions] Next, the operation and function of the measurement device 1 according to this embodiment will be described.

[0037] (Overview of overall operation) First, an overview of the overall operation of the measurement device 1 will be described with reference to FIG. 1. The sensor 11, when brought into contact with the head, ear, or the like of a human body, generates a potential signal E1 corresponding to an electroencephalogram (EEG) in the human body. The noise detection unit 12 calculates the degree to which myoelectric noise is included in the potential signal E1. Specifically, the noise detection unit 12 calculates the time ratio RN of the period during which the potential signal E1 includes myoelectric noise relative to the entire period of the potential signal E1. The processing unit 13 performs noise reduction processing based on the potential signal E1 supplied from the sensor 11, thereby generating a potential signal E2. Specifically, when the time ratio RN of noise obtained by the noise detection unit 12 is equal to or less than a predetermined value TH, the processing unit 13 performs noise reduction processing based on the potential signal E1. Furthermore, when the time ratio RN of noise is greater than the predetermined value TH, the processing unit 13 performs noise reduction processing based on the potential signal E1 using the potential signal E2 supplied from the storage unit 14. The storage unit 14 stores the potential signal E2 generated by the processing unit 13. Specifically, when the time ratio RN of noise obtained by the noise detection unit 12 is equal to or smaller than a predetermined value TH, the storage unit 14 stores the potential signal E2 generated by the processing unit 13. When the time ratio RN of noise is greater than the predetermined value TH, the storage unit 14 supplies the stored potential signal E2 to the signal generation unit 31 of the processing unit 13.

[0038] (Detailed operation) 7 shows an example of the operation of the measuring device 1. Every time the measuring device 1 acquires a potential signal E1 in a period having a time length T, the measuring device 1 performs the following process.

[0039] First, the noise detection unit 12 calculates the time ratio RN of the period during which the potential signal E1 contains myoelectric noise to the entire period of the potential signal E1 (step S101).

[0040] Next, the measuring device 1 checks whether the time ratio RN of noise is equal to or less than a predetermined value TH (R≦TH) (step S102). The predetermined value TH can be set to, for example, 20%.

[0041] In step S102, if the time proportion RN of noise is equal to or less than the predetermined value TH ("Y" in step S102), the signal generating unit 31 of the processing unit 13 supplies the potential signal E1 as the potential signal E31 to the noise reduction processing unit 39 as is, and the noise reduction processing unit 39 generates the potential signal E2 by performing noise reduction processing based on this potential signal E31 (step S103).

[0042] Specifically, in the noise reduction processing unit 39, the signal separation conversion unit 32 first separates the potential signal E31 into signals of multiple frequency components to generate a matrix D as shown in FIG. 5. The matrix conversion unit 33 performs matrix separation processing on the matrix D as shown in FIG. 6 to generate a low-rank matrix L and a sparse matrix S. The low-rank matrix L mainly contains electroencephalogram components, and the sparse matrix S mainly contains electromyographic noise components. The threshold processing unit 34 outputs the low-rank matrix L as a low-rank matrix L' and generates a sparse matrix S' by multiplying the sparse matrix S by a zero matrix. The matrix inverse conversion unit 35 calculates a matrix D' by calculating the sum of the low-rank matrix L' and the sparse matrix S'. The signal separation inverse conversion unit 36 ​​generates a potential signal E2 by performing an inverse transform of the short-time Fourier transform performed by the signal separation conversion unit 32 based on the matrix D'. In this way, the noise reduction processing unit 39 generates a potential signal E2 based on the potential signal E1 by reducing the electromyographic noise contained in the potential signal E1.

[0043] Next, the storage unit 14 stores the potential signal E2 generated in step S103 (step S104). The potential signal E2 stored in the storage unit 14 is used in the next and subsequent processes.

[0044] In step S102, if the time proportion RN of noise is greater than the predetermined value TH ("N" in step S102), the signal generating unit 31 of the processing unit 13 generates the potential signal E31 by combining the potential signal E1 and the potential signal E2 supplied from the memory unit 14 on the time axis (step S105).

[0045] Next, the noise reduction processing unit 39 of the processing unit 13 performs noise reduction processing based on the potential signal E31 to generate a potential signal E2 (step S106). The noise reduction processing in step S106 is the same as the noise reduction processing in step S103, except that the input potential signal E31 includes the potential signal E1 and the electric signal E2.

[0046] Then, the signal separation and inverse conversion unit 36 ​​of the noise reduction processing unit 39 updates the potential signal E2 by removing the signal portion corresponding to the potential signal E1 from the potential signal E2 (step S107).

[0047] This is the end of this flow.

[0048] In this way, in the measurement device 1, when the time ratio RN of noise is greater than the predetermined value TH ("N" in step S102), the potential signal E1 and the potential signal E2 supplied from the storage unit 14 are combined on the time axis to generate the potential signal E31 (step S105), and the potential signal E2 is generated by performing noise reduction processing based on this potential signal E31 (step S106). In this way, the measurement device 1 can effectively reduce myoelectric noise when it occurs continuously for a long period of time.

[0049] That is, if myoelectric noise occurs continuously for a long time, performing noise reduction processing based only on the potential signal E1 as shown in step S103 may not be able to reduce the myoelectric noise. That is, if myoelectric noise occurs continuously for a long time, the matrix D may include the features indicated by lines in addition to the features indicated by hatching and dots, as shown in FIG. 8. These lines are formed by arranging matrix elements with large values ​​in the time axis direction. The matrix conversion unit 33 generates a low-rank matrix L and a sparse matrix S by performing a matrix separation processing based on such a matrix D, as shown in FIG. 9. The low-rank matrix L has the features indicated by hatching and lines in FIG. 8. That is, the low-rank matrix L includes components of continuously occurring myoelectric noise in addition to components of electroencephalograms. As a result, in this example, the myoelectric noise cannot be effectively reduced.

[0050] Therefore, when myoelectric noise occurs continuously for a long period of time, the measurement device 1 generates a potential signal E31 by combining the potential signal E1 and the potential signal E2 supplied from the storage unit 14 on the time axis (step S105).The measurement device 1 then performs noise reduction processing based on this potential signal E31 to generate a potential signal E2 (step S106).

[0051] FIG. 10 is a schematic diagram illustrating an example of the matrix separation process performed in step S106. Since matrix D is generated based on the combined potential signal E31, matrix D is twice as large in the time axis direction as matrix D shown in FIG. 9. In this matrix D, the left half corresponds to potential signal E1, and the right half corresponds to potential signal E2 supplied from the storage unit 14. The matrix conversion unit 33 performs matrix separation process based on such matrix D to generate a low-rank matrix L and a sparse matrix S. The low-rank matrix L is also twice as large in the time axis direction as the low-rank matrix L shown in FIG. 9, and the sparse matrix S is also twice as large in the time axis direction as the sparse matrix S shown in FIG. 9. As shown in FIG. 10, the low-rank matrix L has the characteristics indicated by the hatching in FIG. 8 and mainly contains electroencephalogram components. The sparse matrix S has the characteristics indicated by the dots and lines in FIG. 8 and mainly contains electromyographic noise components. Continuously occurring electromyographic noise components are included in the sparse matrix S. That is, in the example of Fig. 9, the low-rank matrix L includes a component of myoelectric noise that occurs continuously, whereas in the example of Fig. 10, the sparse matrix S includes a component of myoelectric noise that occurs continuously. This allows the measurement device 1 to effectively reduce noise components.

[0052] As described above, the measurement device 1 includes a sensor 11 that generates a first biosignal (potential signal E1) corresponding to electroencephalogram information, a processing unit 13 that performs noise reduction processing based on the first biosignal (potential signal E1) to generate a second biosignal (potential signal E2), and a storage unit 14 that stores the processing result (potential signal E2) of the noise reduction processing. The processing unit 13 performs noise reduction processing based on the first biosignal (potential signal E1) using the processing result (potential signal E2) stored in the storage unit 14. As a result, the measurement device 1 can reduce myoelectric noise, for example, when myoelectric noise occurs continuously for a long period of time, as shown in FIG. 10 . As a result, the measurement device 1 can effectively reduce noise components.

[0053] The measurement device 1 also includes a noise detection unit 12 that calculates the noise level (time ratio of noise RN) of the first biosignal (potential signal E1) containing noise corresponding to muscle fiber activity information. The processing unit 13 performs noise reduction processing based on the first biosignal (potential signal E1) using the processing result (potential signal E2) stored in the storage unit 14 when the noise level is greater than a predetermined value. The processing unit 13 also performs noise reduction processing based on the first biosignal (potential signal E1) when the noise level is less than a predetermined value. This allows the measurement device 1 to store the second biosignal (potential signal E2) obtained when there is little noise in the storage unit 14, and to perform noise reduction processing based on the potential signal E1 and the potential signal E2 stored in the storage unit 14 when there is much noise, as shown in FIG. 10 . As a result, the measurement device 1 can effectively reduce noise components.

[0054] In the measurement device 1, the noise reduction process includes generating a first matrix (matrix D) by separating the first biological signal (electric potential signal E1) into multiple signals, performing a matrix separation process to separate the first matrix (matrix D) into a second matrix (low-rank matrix L) having components with little change in the time axis direction and a third matrix (sparse matrix S) having components that are discrete in the time axis direction, and generating a second biological signal (electric potential signal E2) based on the second matrix (low-rank matrix L). In the matrix separation process, the processing unit 13 generates a fourth matrix (Ψ(S)) by performing an enhancement process to enhance the third matrix (sparse matrix S), and separates the first matrix (matrix D) into the second matrix (low-rank matrix L) and the third matrix (sparse matrix S) so that the second matrix (low-rank matrix L) and the fourth matrix (Ψ(S)) satisfy a predetermined condition. 3B, for example, even when the amplitude intensity of the myoelectric noise is close to the amplitude intensity of the electroencephalogram, the measurement device 1 can reduce the myoelectric noise by emphasizing the signal component of the myoelectric noise. As a result, the measurement device 1 can effectively reduce the noise component.

[0055] This allows the measurement device 1 to reduce the number of channels, thereby reducing the number of sensors 11 connected to the user and improving comfort. Even with such a small number of channels, the measurement device 1 can generate a potential signal E2 with fewer noise components, thereby improving the accuracy of electroencephalogram measurement.

[0056] [effect] As described above, this embodiment includes a sensor that generates a first biological signal according to electroencephalogram information, a processing unit that performs noise reduction processing based on the first biological signal to generate a second biological signal, and a storage unit that stores the processing result of the noise reduction processing.The processing unit then performs noise reduction processing based on the first biological signal using the processing result stored in the storage unit.This makes it possible to effectively reduce noise components.

[0057] In this embodiment, a noise detection unit is provided that calculates the noise level of the first biological signal, which is determined based on the activity information of muscle fibers. If the noise level is greater than a predetermined value, the processing unit performs noise reduction processing based on the first biological signal using the processing results stored in the storage unit. If the noise level is less than the predetermined value, the processing unit performs noise reduction processing based on the first biological signal. This effectively reduces noise components.

[0058] In addition, in the measurement device 1, the noise reduction process includes generating a first matrix by separating the first biological signal into multiple signals, performing a matrix separation process to separate the first matrix into a second matrix having components with little change in the time axis direction and a third matrix having components that are discrete in the time axis direction, and generating a second biological signal based on the second matrix. In the matrix separation process, the processing unit generates a fourth matrix by performing an emphasis process to emphasize the third matrix, and separates the first matrix into the second matrix and the third matrix so that the second matrix and the fourth matrix satisfy a predetermined condition. This allows for effective reduction of noise components.

[0059] [Variation 1] In the above embodiment, the measurement device 1 is configured as a single-channel measurement device that performs noise reduction processing based on one potential signal E1, but this is not limited to this. Alternatively, the measurement device 1 may perform noise reduction processing based on multiple potential signals E1. In this case, for example, the measurement device 1 may have multiple circuit groups, such as the sensor 11, noise detection unit 12, processing unit 13, and storage unit 14 shown in FIG. 1. Also, for example, multiple sensors 11 may be provided, and the processing unit 13 may generate, for example, a three-dimensional matrix D based on the multiple potential signals E1 generated by these multiple sensors 11, and perform processing using this matrix D.

[0060] [Variation 2] In the above embodiment, the processor of the measurement device 1 performs all of the processes for generating the potential signal E2 by performing noise reduction processing based on the potential signal E1, but this is not limited to this. Instead, for example, some of these processes may be performed by a computing device other than the measurement device. Specifically, for example, a personal computer, a smartphone, a cloud network, or the like may perform some of these processes.

[0061] [Variation 3] In the above embodiment, the signal separation conversion unit 32 generates the matrix D by performing a short-time Fourier transform based on the potential signal E31, but this is not limiting. Instead, for example, the matrix D may be generated by performing a wavelet transform, a discrete cosine transform, an empirical mode decomposition, or the like based on the potential signal E31. Furthermore, when performing processing based on a plurality of potential signals E1, for example, the signal separation conversion unit 32 can generate the matrix D by performing principal component analysis, canonical correlation analysis, or independent component analysis, for example.

[0062] [Variation 4] In the above embodiment, the matrix transformation unit 33 calculates the differences between multiple matrix elements in the matrix S in the frequency axis direction in the transformation Ψ, but this is not limiting. Instead, for example, the matrix transformation unit 33 may calculate the differences between multiple matrix elements in the time axis direction, or may calculate the differences between multiple matrix elements in the frequency axis direction and also in the time axis direction. Also, in this example, the matrix transformation unit 33 calculates the differences in the transformation Ψ, but this is not limiting, and the matrix transformation unit 33 may perform a wavelet transform or a discrete cosine transform. Even in this case, the transformation Ψ can emphasize sparse components in the matrix.

[0063] [Variation 5] In the above embodiment, myoelectric noise is reduced, but the present invention is not limited to this. Alternatively, for example, electrooculography noise due to eye movement, electrocardiography noise due to heartbeat, sweat noise, and contact noise caused by changes in the contact state between the electrodes 21 and 22 and the skin may be reduced. The signal separation and transformation unit 32 may select one of short-time Fourier transform, wavelet transform, discrete cosine transform, and empirical mode decomposition as the processing to be used, depending on the type of noise to be reduced. Similarly, the matrix transformation unit 33 may select one of a process for calculating a difference in the transformation Ψ, a wavelet transform, and a discrete cosine transform as the processing to be used, depending on the type of noise to be reduced.

[0064] [Variation 6] In the above embodiment, the thresholding unit 34 outputs the low-rank matrix L as the low-rank matrix L' without modification and generates the sparse matrix S' by multiplying the sparse matrix S by a zero matrix. That is, the weighting factor of the low-rank matrix L is set to "1" and the weighting factor of the sparse matrix S is set to "0." This is not limiting, and other weighting factors may be set. For example, the thresholding unit 34 may generate the low-rank matrix L' by multiplying the low-rank matrix L by a matrix whose multiple matrix elements are each "1" or "0," or may generate the low-rank matrix L' by multiplying the sparse matrix S by a matrix whose multiple matrix elements are each a value between "0" and "1." Similarly, the thresholding unit 34 may generate the sparse matrix S' by multiplying the sparse matrix S by a matrix whose multiple matrix elements are each a value between "0" and "1," or may generate the sparse matrix S' by multiplying the sparse matrix S by a matrix whose multiple matrix elements are each a value between "0" and "1." The thresholding unit 34 can set a matrix to be multiplied by the low-rank matrix L and a matrix to be multiplied by the sparse matrix S, for example, using a weighting coefficient.

[0065] Furthermore, the thresholding unit 34 may evaluate the autocorrelation in the time axis direction in the low-rank matrix L and may also evaluate the autocorrelation in the time axis direction in the sparse matrix S, and perform processing based on the evaluation results. Specifically, the thresholding unit 34 may set weighting factors based on the evaluation results, and use the weighting factors to set a matrix to be multiplied by the low-rank matrix L and a matrix to be multiplied by the sparse matrix S.

[0066] [Variation 7] In the above embodiment, the noise detection unit 12 calculates the time ratio RN of the period during which the potential signal E1 contains myoelectric noise to the entire period of the potential signal E1, but this is not limited to this. Instead, for example, the noise detection unit 12 may use a machine learning technique to output a score indicating the degree to which the potential signal E1 contains myoelectric noise, based on the potential signal E1.

[0067] [Variation 8] In the above embodiment, the signal generating unit 31 generates the potential signal E31 by combining the potential signal E1 and the potential signal E2 supplied from the storage unit 14 on the time axis, but this is not limiting. Alternatively, the signal generating unit 31 may combine the potential signal E1 and a portion of the potential signal E2 on the time axis. For example, the signal generating unit 31 can adjust the length of the portion of the potential signal E2 that is connected to the potential signal E1 so that the proportion of the time period containing myoelectric noise in the entire combined period becomes a predetermined proportion.

[0068] [Variation 9] In the above embodiment, the storage unit 14 stores the potential signal E2 for a period of time T. However, the storage unit 14 may store multiple past potential signals E2. In this case, the signal generation unit 31 may generate the potential signal E31 by, for example, combining the potential signal E1 with one of the multiple potential signals E2 stored in the storage unit 14 on the time axis. In this case, the signal generation unit 31 may combine, for example, the most recent potential signal E2 of the multiple potential signals E2 stored in the storage unit 14 with the potential signal E1, or may combine, for example, a potential signal E2 different from the most recent potential signal E2 of the multiple potential signals E2 stored in the storage unit 14 with the potential signal E1. Furthermore, the signal generation unit 31 may generate an average signal of the multiple potential signals E2 stored in the storage unit 14 and combine this average signal with the potential signal E1 on the time axis to generate the potential signal E31. Furthermore, the signal generating section 31 may generate the potential signal E31 by combining, for example, the potential signal E1 and a plurality of potential signals E2 supplied from the storage section 14 on the time axis.

[0069] [Variation 10] In the above embodiment, the storage unit 14 stores the potential signal E2, but the present invention is not limited to this. A measurement apparatus 1A according to this modified example will be described in detail below.

[0070] 11 shows an example of the configuration of a measurement device 1A. The measurement device 1A has a processing unit 13A and a storage unit 14A.

[0071] The processing unit 13A is configured to generate a potential signal E2 by performing noise reduction processing based on the potential signal E1 supplied from the sensor 11. When the time proportion RN of noise obtained by the noise detection unit 12 is equal to or smaller than a predetermined value TH, the processing unit 13A performs noise reduction processing based on the potential signal E1. When the time proportion RN of noise is greater than the predetermined value TH, the processing unit 13A performs noise reduction processing based on the potential signal E1 using a low-rank matrix L supplied from the storage unit 14.

[0072] 12 shows an example of the configuration of the processing unit 13A. The processing unit 13A has a noise reduction processing unit 39A. The noise reduction processing unit 39A has a signal separation conversion unit 32A and a matrix conversion unit 33A.

[0073] The signal separation conversion unit 32A is configured to generate a matrix D by separating the potential signal E31 into signals of a plurality of frequency components. This operation is similar to the operation of the signal separation conversion unit 32 according to the above embodiment. When the time proportion RN of noise obtained by the noise detection unit 12 is equal to or smaller than a predetermined value TH, the signal separation conversion unit 32A outputs the matrix D as is. When the time proportion RN of noise obtained by the noise detection unit 12 is greater than the predetermined value TH, the signal separation conversion unit 32A updates the matrix D by combining the matrix D with a low-rank matrix L supplied from the storage unit 14A, and outputs the updated matrix D. In this way, the signal separation conversion unit 32A can generate a matrix D similar to that in the above embodiment (FIG. 10), for example.

[0074] The matrix transformation unit 33A is configured to generate a low-rank matrix L and a sparse matrix S by performing a matrix separation process on the matrix D. This operation is similar to the operation of the matrix transformation unit 33 according to the above embodiment. The matrix transformation unit 33A then supplies the generated low-rank matrix L and sparse matrix S to the threshold processing unit 34, and also supplies this low-rank matrix L to the storage unit 14A.

[0075] The storage unit 14A (FIG. 11) is configured to store the low-rank matrix L generated by the processing unit 13A. Specifically, the storage unit 14A stores the low-rank matrix L generated by the processing unit 13A when the time proportion RN of noise obtained by the noise detection unit 12 is equal to or smaller than a predetermined value TH. Then, when the time proportion RN of noise is greater than the predetermined value TH, the storage unit 14A supplies the stored low-rank matrix L to the processing unit 13A.

[0076] Even with this configuration, the measurement apparatus 1A can operate in the same manner as the measurement apparatus 1 according to the above embodiment.

[0077] [Other variations] Two or more of these variations may be combined.

[0078] The present technology has been described above by giving embodiments and some modified examples, but the present technology is not limited to these embodiments and can be modified in various ways.

[0079] For example, in each of the above embodiments, electroencephalograms are detected, but the present invention is not limited to this, and various types of biological information can be detected.

[0080] The effects described in this specification are merely examples and are not limiting, and other effects may also be present.

[0081] The present technology can be configured as follows: According to the present technology configured as follows, it is possible to improve detection accuracy.

[0082] (1) a sensor capable of generating a first biological signal according to the first biological information; a processing unit capable of generating a second biological signal by performing noise reduction processing on the first biological signal; a storage unit capable of storing the processing results of the noise reduction processing; Equipped with The processing unit is capable of performing the noise reduction processing based on the first biological signal and using the processing result stored in the storage unit. Measuring equipment. (2) a noise detection unit capable of calculating a noise level in which the first biological signal includes noise corresponding to the second biological information; The processing unit When the degree of noise is greater than a predetermined value, the noise reduction process can be performed based on the first biological signal using the processing result stored in the storage unit; When the degree of noise is smaller than the predetermined value, the noise reduction process can be performed based on the first biological signal. The measuring device according to (1) above. (3) In the noise reduction process, the processing unit generating a first matrix by separating the first biological signal into a plurality of signals; performing a matrix separation process of separating the first matrix into a second matrix having components that change little in a time axis direction and a third matrix having components that are discrete in a time axis direction; generating the second biological signal based on the second matrix; It is possible to The measuring device according to (2) above. (4) In the matrix separation process, the processing unit is capable of generating a fourth matrix by performing an enhancement process that enhances the third matrix, and is capable of separating the first matrix into the second matrix and the third matrix so that the second matrix and the fourth matrix satisfy a predetermined condition. The measuring device according to (3) above. (5) the processing result stored in the storage unit is the second biological signal generated by the previous noise reduction processing, When the noise level is greater than the predetermined value, the processing unit: The first biological signal can be updated by combining the processing result stored in the storage unit with the first biological signal; The noise reduction process can be performed based on the updated first biological signal. The measuring device according to (3) or (4). (6) the processing result stored in the storage unit is the second matrix generated by the previous noise reduction processing, When the noise level is greater than the predetermined value, the processing unit: The first matrix can be updated by combining the processing result stored in the storage unit with the first matrix, The updated first matrix can be separated into the second matrix and the third matrix. The measuring device according to (3) or (4). (7) The storage unit is capable of storing the processing result of the noise reduction processing when the noise level is smaller than the predetermined value. The measuring device according to any one of (2) to (6). (8) The second biological information is activity information of muscle fibers. The measuring device according to any one of (2) to (7). (9) The first biological information is electroencephalogram information. The measuring device according to any one of (1) to (8). (10) generating a first biological signal according to the first biological information; generating a second biological signal by performing noise reduction processing based on the first biological signal; storing a processing result of the noise reduction processing; Including, The noise reduction processing is performed based on the first biological signal using the stored processing result. Measurement method.

[0083] This application claims priority based on Japanese Patent Application No. 2021-151823, filed on September 17, 2021, with the Japan Patent Office, the entire contents of which are incorporated herein by reference.

[0084] Those skilled in the art will recognize that various modifications, combinations, subcombinations, and variations may occur depending on design requirements and other factors, and are intended to be within the scope of the appended claims and their equivalents.

Claims

1. a sensor capable of generating a first biological signal according to the first biological information; a processing unit capable of generating a second biological signal by performing noise reduction processing on the first biological signal; a noise detection unit capable of calculating a noise level in which the first biological signal includes noise corresponding to the second biological information; a storage unit capable of storing the processing results of the noise reduction processing; Equipped with The processing unit When the degree of noise is greater than a predetermined value, the noise reduction process can be performed based on the first biological signal using the processing result stored in the storage unit, When the degree of noise is smaller than the predetermined value, the noise reduction process can be performed based on the first biological signal; The noise reduction process includes: generating a first matrix by separating the first biological signal into a plurality of signals; performing a matrix separation process of separating the first matrix into a second matrix having components that change little in a time axis direction and a third matrix having components that are discrete in a time axis direction; generating the second biological signal based on the second matrix; Contains Measuring equipment.

2. In the matrix separation process, the processing unit is capable of generating a fourth matrix by performing an emphasis process that emphasizes the third matrix, and is capable of separating the first matrix into the second matrix and the third matrix such that the second matrix and the fourth matrix satisfy a predetermined condition. The measurement device according to claim 1 .

3. the processing result stored in the storage unit is the second biological signal generated by the previous noise reduction processing, When the noise level is greater than the predetermined value, the processing unit: The first biological signal can be updated by combining the processing result stored in the storage unit with the first biological signal, The noise reduction process can be performed based on the updated first biological signal. The measurement device according to claim 1 .

4. the processing result stored in the storage unit is the second matrix generated by the previous noise reduction processing, When the noise level is greater than the predetermined value, the processing unit: The first matrix can be updated by combining the processing result stored in the storage unit with the first matrix, The updated first matrix can be separated into the second matrix and the third matrix. The measurement device according to claim 1 .

5. The storage unit is capable of storing the processing result of the noise reduction processing when the noise level is smaller than the predetermined value. The measurement device according to claim 1 .

6. The second biological information is activity information of muscle fibers. The measurement device according to claim 2 .

7. The first biological information is electroencephalogram information. The measurement device according to claim 1 .

8. generating a first biological signal according to the first biological information; generating a second biological signal by performing noise reduction processing on the first biological signal; calculating a noise level of the first biological signal including noise according to the second biological information; storing a processing result of the noise reduction processing; Including, The noise reduction process includes: When the degree of noise is greater than a predetermined value, the stored processing result is used based on the first biological signal; When the degree of noise is smaller than the predetermined value, the noise level is determined based on the first biological signal; The noise reduction process includes: generating a first matrix by separating the first biological signal into a plurality of signals; performing a matrix separation process of separating the first matrix into a second matrix having components that change little in a time axis direction and a third matrix having components that are discrete in a time axis direction; generating the second biological signal based on the second matrix; Contains Measurement method.

Images