JP5691815B2 - Signal processing apparatus, signal processing method, and biological information measuring apparatus - Google Patents

Description

  The present invention relates to a signal processing device, a signal processing method, and a biological information measurement device that remove noise components from a time-series signal.

  Conventionally, techniques relating to signal processing for removing noise components from time-series data on which noise components are superimposed have been applied to various signal processing apparatuses. In particular, when the time-series data includes information related to biological information, the above signal processing device is called a biological information measuring device. The biological information measuring device is a device that non-invasively detects biological information from biological tissue, specifically, a measuring device called a pulse wave meter and a pulse oximeter that measures the pulse waveform and pulse rate of a living body called a photoelectric pulse wave meter. It is a measuring device or the like that measures the arterial blood oxygen saturation concentration. The principle of these measuring devices is that a living body such as the concentration of a light-absorbing substance in blood is obtained based on a signal component obtained by receiving light transmitted or reflected through a living tissue and corresponding to fluctuations due to pulsation of the living tissue. It seeks information.

  In general, various noise components are superimposed on data necessary for detection of biological information, which is obtained by receiving light transmitted or reflected through biological tissue. The noise component is mainly due to body movement such as movement of the body when the biological information measuring device is used. If the noise component is superimposed on the signal component, it causes an error in the calculation of biological information, and therefore it is desired to remove the noise component.

  There has been proposed a technique for calculating biological information based on a direct-current alternating current ratio in intensity of light transmitted or reflected through a living tissue when a living body is irradiated with a plurality of lights having different wavelengths. In particular, when a noise component is superimposed on a signal component, the DC / AC ratio for each wavelength is represented by the signal component and the noise component. As a technique for removing the noise component represented in this way, the signal component is extracted from the data including the signal component having the periodicity generated based on the measured data by using the periodicity, and the noise is extracted from the data. Techniques for removing components have been proposed (see Patent Document 1).

  According to this technique, it is possible to reduce the amount of calculation processing for removing noise components, compared to the conventional method using the cross-correlation between signal components and noise components.

International Publication No. 2010/073908 Specification

  However, although the amount of calculation processing is reduced by the technique of Patent Document 1, it cannot be denied that the power consumption increases because the number of multiplications is relatively large. In particular, since a portable biological information measuring device is usually driven by a battery, power consumption is a serious problem.

  In addition, the biological information measuring device is not only for operating rooms, intensive care units, etc., but also for respiratory failure patients, home oxygen therapy patients in daily life respiratory data collection and management, sleep apnea syndrome screening, It is being used even for sportswear such as mountain climbing that can be worn at all times. Even in view of such a situation, the biological information measuring apparatus is also required to be reduced in size, weight, power saving, price reduction, and the like.

  In addition to the biological information measuring device taken up as an example, in general, in signal processing devices that perform signal processing for removing noise components from time-series data, signal processing for removing noise components is a serious problem in terms of power consumption. I can say that.

  The present invention has been made in view of the above circumstances, and an object thereof is to provide a signal processing device, a signal processing method, and a biological information measurement device that reduce the amount of signal processing and suppress power consumption. .

  As a result of various studies, the present inventor has found that the above object is achieved by the present invention described below.

A signal processing apparatus according to an aspect of the present invention includes a first time-series signal including a first signal component having periodicity and a first noise component, and a second signal having a predetermined relationship with the first signal component. Based on a second time-series signal including a signal component and a second noise component having a predetermined relationship with the first noise component, and the predetermined relationship between the first noise component and the second noise component, A signal including the first signal component from the first time-series signal and the second time-series signal, wherein the predetermined relationship between the first noise component and the second noise component is a predetermined difference. The signal generation unit that generates a plurality of signals that are different from each other and the periodicity of each signal generated by the signal generation unit in the first predetermined time range, the first in the first predetermined time range. The predetermined relationship between one noise component and the second noise component is estimated. The signal generation unit generates one reference signal, and a recurrence formula using the signal and data calculated from the first time series signal and the predetermined difference To generate another signal that is a signal other than the reference signal among the plurality of signals , and further calculate a threshold value used for binarization processing of the reference signal. And the data calculated using the first time-series signal or the second time-series signal and the predetermined difference, and used for binarization processing of each of the other signals by a recurrence formula Each of the threshold values is calculated, and a binarization process is performed on the generated signal using the threshold value to generate a binarized signal, and the estimation unit uses the periodicity of the binarized signal. Estimating the period of the signal generated by the signal generator And estimating a predetermined relationship.

A signal processing method according to another aspect of the present invention is a signal processing method including a first signal component having a periodicity and a first noise component in a first predetermined time range. A second time-series signal including a second signal component having a predetermined relationship with the first signal component and a second noise component having a predetermined relationship with the first noise component; A signal including the first signal component from the first time-series signal and the second time-series signal based on the predetermined relationship between the noise component and the second noise component; A signal generation step for generating a plurality of signals in which the predetermined relationship between one noise component and the second noise component is different by a predetermined difference, and each signal generated in the generation step in a first predetermined time range Using the periodicity of the first predetermined time range An estimation step for estimating the predetermined relationship between the first noise component and the second noise component, wherein the signal generation step generates one reference signal, the signal and the first noise component. when the recurrence formula using the the data calculated from the time-series signal and the predetermined difference, generates other signal is a signal other than one signal serving as the reference of the plurality of signals, further, the A threshold value used for binarization processing of a reference signal is calculated, and the threshold value and data calculated using the first time-series signal or the second time-series signal and the predetermined difference; Is used to calculate a threshold value to be used for binarization processing of each of the other signals by a recurrence formula, and binarization is performed by performing binarization processing on the generated signal using the threshold value Generating a signal and said estimating step The estimates the period of the signal generated by said signal generating step using the periodicity of the binary signal, and estimates the predetermined relationship.

  In the signal processing apparatus having such a configuration, since the recurrence formula is used in the signal processing necessary for removing the noise component by using the periodicity, addition / subtraction is used more frequently than multiplication / division. In addition, the amount of signal processing can be reduced as compared with the prior art. For this reason, it is possible to reduce the required CPU performance by reducing the signal processing amount, and it is also possible to suppress the power consumption associated with the calculation of noise component removal.

  According to the signal processing apparatus having such a configuration, the threshold value for performing the binarization process can be set for each generated signal, and the amount of signal processing for obtaining the threshold value can be reduced as compared with the conventional technique. This makes it possible to perform binarization processing quickly.

  Further, in the above-described signal processing device, the signal generated by the signal generation unit is a time-series signal, and the signal generation unit uses the data obtained by thinning out the generated signal every predetermined number. A value processing is performed.

  According to the signal processing device having such a configuration, since the signal processing is performed using a part of the measurement data, the amount of signal processing can be further reduced.

  In the above-described signal processing device, the binarization process is performed on data thinned between two pieces of the thinned data sandwiching the threshold value.

  According to the signal processing apparatus having such a configuration, binarization is performed using the data of the measured intervals in the vicinity of the threshold value, so that a quick and more accurate binarization process can be performed.

In addition, the biological information measuring device according to one aspect of the present invention is obtained by irradiating a living body with a plurality of lights having different wavelengths and receiving each of the lights transmitted or reflected through the living body. A biological information measuring apparatus for measuring biological information of the living body based on measurement data or second measurement data, and measuring first measurement data including a first signal component having periodicity and a first noise component. A first measurement unit; a second measurement unit that measures second measurement data including a second signal component having a predetermined relationship with the first signal component and a second noise component having a predetermined relationship with the first noise component; , Based on the first measurement data, the second measurement data, and the predetermined relationship between the first noise component and the second noise component, the first measurement data and the second measurement data, A signal generation unit that generates a plurality of signals that include the first signal component and that differ in the predetermined relationship between the first noise component and the second noise component by a predetermined difference; An estimation unit that estimates the predetermined relationship between the first noise component and the second noise component using the periodicity of the signal generated by the signal generation unit, and the signal generation unit is a reference 1 A signal other than the one signal serving as the reference among the plurality of signals is generated by a recurrence formula using the signals and the data calculated from the first measurement data and the predetermined difference. And further calculating a threshold value used for the binarization process of the reference signal, and the threshold value and the first time series signal or the second time series signal Data calculated using the predetermined difference The threshold value to be used for the binarization processing of each of the other signals is calculated by the recurrence formula, and the binarization processing is performed on the generated signal using the threshold value, thereby obtaining the binarized signal. And the estimation unit estimates the predetermined relationship by estimating a period of the signal generated by the signal generation unit using periodicity of the binarized signal .

  According to the biological information measuring apparatus having such a configuration, it is possible to provide a biological information measuring apparatus that measures the blood oxygen saturation of a living body that reduces the amount of calculation processing and suppresses power consumption. In addition, since the amount of calculation processing is reduced, it is less necessary to provide a high-performance processing function, and it is possible to realize a reduction in price, size, and weight.

Further, in the above-described biological information measurement device, based on a predetermined relationship in which the estimating unit has estimated, with a blood oxygen saturation calculating unit for calculating a blood oxygen saturation at a predetermined cycle, said signal producing formation unit Generates a plurality of signals with a period longer than the predetermined period to generate a binarized signal, and the estimation unit generates the first noise based on the periodicity of the binarized signal. The predetermined relationship between a component and the second noise component is estimated.

  A predetermined relationship k_v, which is a ratio between the first noise component and the second noise component, is calculated with a period longer than the period for obtaining blood oxygen saturation from the measurement data. This is because k_v does not change abruptly. Accordingly, the amount of signal processing is reduced compared to obtaining k_v every time data is measured, such as generating and binarizing a plurality of signals, and a more rapid signal processing amount is possible.

  Moreover, in the above-described biological information measuring device, a binarization signal is generated by performing binarization processing on a signal having a predetermined relationship estimated by the estimation unit, and the periodicity of the binarization signal is used. It is provided with the pulse rate calculation part which calculates a pulse rate.

  In the signal processing apparatus having such a configuration, noise components are removed by using the periodicity, and addition / subtraction is performed rather than multiplication / division by using a recurrence formula in signal processing necessary for estimating the heart rate. As a result, the amount of signal processing can be reduced as compared with the prior art.

  The signal processing device, the signal processing method, and the biological information measurement device according to the present invention can reduce the amount of signal processing necessary for noise component removal, thereby reducing power consumption associated with signal processing for noise component removal. It becomes possible to suppress.

It is a block diagram which shows the structure of the biological information measuring device in embodiment. It is a flowchart which shows the biological information measurement process in the biological information measuring device of embodiment. It is a flowchart which shows the initial process in the biometric information measuring apparatus of embodiment. It is a flowchart which shows the kv_min calculation process in the biological information measuring device of an embodiment. It is a figure which shows the waveform of R-kv * IR by measurement data. It is a flowchart which shows the binarization process in the biological information measuring device of embodiment. It is a figure for demonstrating the binarization process of FIG. It is a flowchart which shows the variation index calculation process in the biological information measuring device of embodiment. It is a flowchart which shows the pulse rate PR calculation process in the biological information measuring device of embodiment. It is a flowchart which shows the binarization process in the biometric information measuring apparatus of a modification. It is a flowchart which shows the pulse rate PR calculation process in the biological information measuring device of a modification.

The prior art and embodiments that are the premise of the present invention will be described below. For convenience, a biological information measuring device is taken as an example of signal processing devices, but the present invention is applicable not only to a biological information measuring device but also to a signal processing device that removes noise components.
<Conventional Technology as a Premise of the Present Invention>
First, the prior art as a premise of the present invention will be described. In this description, as an example, based on measurement data obtained by irradiating a living body with a plurality of lights having different wavelengths IR and R, respectively, and receiving or transmitting each light transmitted or reflected by the living body. A case where blood oxygen saturation is measured as the biological information of the living body will be described.

  According to Lambert-Beer's law, the ratio of the alternating current component to the direct current component in the intensity of light having a certain wavelength transmitted or reflected through living tissue is approximated to be equal to the change in absorbance of the living tissue at that wavelength.

  By using the Lambert-Beer law approximation described above, the infrared orthogonal ratio IR_signal, which is the ratio of the direct current component and the alternating current component of the transmitted or reflected light intensity for the infrared wavelength IR, It can be regarded as equal to the change in the absorbance of the tissue. Similarly, the red orthogonal ratio R_signal, which is the ratio of the direct current component to the alternating current component of the intensity of transmitted light or reflected light for the red wavelength R, may also be regarded as equal to the change in the absorbance of the living tissue for the wavelength R. it can.

  The infrared orthogonal ratio IR_signal is expressed as shown in Equation (1).

Here, s is a signal component corresponding to a change in absorbance, and n is a noise component superimposed on the signal component.

  The red orthogonal ratio R_signal is expressed by Expression (2).

Where k_a is the ratio of the signal component s of the absorbance change at the wavelength IR to the signal component of the absorbance change at the wavelength R, and k_v is the noise component n and the wavelength superimposed on the signal component of the wavelength IR. This is the ratio of the noise component superimposed on the R signal component.

  It is known that k_a and arterial blood oxygen saturation in the formula (2) correspond one-to-one, and the arterial oxygen saturation can be obtained by obtaining k_a.

  Further, the equation (3) is obtained by multiplying the equation (1) by k_v, and the following equation (4) is obtained by subtracting the equation (2) from the equation (3).

  Similarly, equation (5) is obtained by multiplying the above equation (1) by k_a, and the following equation (6) is obtained by subtracting (equation 2) from equation (5).

Here, the relationship that the signal component s and the noise component n are independent, that is, the following relational expression (7) is used, and under the condition that k_a and k_v are constant within a short time ( Equation (8) is obtained by correlating equations 4) and (6).

Here, Σ is the sum for a short time such that k_a and k_v are constant. i is the data number of the time-series data IR_signal and R_signal corresponding to the change in the intensity of light, and the measurement time interval of the data is Δt, the measurement start time is t0, and t = Δt * i + t0 Tied.
Since Equation (8) contains two unknowns k_v and k_a, k_a and k_v cannot be obtained from Equation (8) alone.

  Here, in order to obtain k_v, paying attention to the fact that the right side of equation (4) is substantially periodic, k_v is obtained such that the left side of equation (4) has periodicity.

Substituting k_v obtained in this way into equation (8), k_a when equation (8) is satisfied is obtained. Based on this k_a, the arterial oxygen saturation with reduced noise components can be determined. In the above-described calculation method, for example, it is not necessary to use a Fourier transform or the like having a relatively large calculation processing amount, so that the calculation processing amount can be reduced accordingly.
<Embodiment>
In the present embodiment, the amount of calculation processing for obtaining k_v such that the left side of equation (4) has periodicity is significantly reduced compared to the amount of calculation processing for obtaining k_v by directly calculating the above equation. It is.

  Here, a method of reducing the arithmetic processing when obtaining k_v will be described.

  K_v is calculated such that q (see the following equation) corresponding to the left side of equation (4) has the most periodicity.

q (i) = IR_signal (i) * k_v-R_signal (i)
Here, i represents the i-th time series data from 1 to N, and the correspondence between i and time is as described above in <Prior Art as a Premise of the Present Invention>. For N, the number of data necessary for calculating the arterial oxygen saturation is selected, and a value such as 200 is used.

In order to obtain the arterial oxygen saturation, first, q is obtained, and q is binarized to determine k_v that minimizes an index representing the variation in the pulse period. Then, the determined k_v is set as kv_min (hereinafter, k_v at which the index indicating the variation in pulse period is minimized is referred to as “kv_min”).
ΣR_signal (i) ² -kv_min * ΣR_signal (i) * IR_signal (i)
-k_a * {ΣR_signal (i) * IR_signal (i) -kv_min * ΣIR_signal (i) ² } = 0
Solve
k_a = {ΣR_signal (i) ² -kv_min * ΣR_signal (i) * IR_signal (i)}
/ {ΣR_signal (i) * IR_signal (i) -kv_min * ΣIR_signal (i) ² }
To determine k_a. Then, the blood oxygen saturation is calculated based on k_a. Σ is a sum taken from 1 to N with respect to i (the same applies to the following equations). Further, ΣR_signal (i) ² , ΣIR_signal (i) ² , and ΣR_signal (i) * IR_signal (i) may be described as R_signal ² , ΣIR_signal ² , and ΣR_signal * IR_signal, respectively.

k_a = {ΣR_signal (i) ² -kv_min * ΣR_signal (i) * IR_signal (i)}
/ {ΣR_signal (i) * IR_signal (i) -kv_min * ΣIR_signal (i) ² }
, R_signal and IR_signal may be replaced with time differences, respectively.

  The method for obtaining kv_min is as follows.

  First, k_v is changed by Δk_v and q is obtained for each k_v.

  For example, k_v is incremented by Δk_v M times from a certain value, and is subtracted M times to calculate (2 * M + 1) ways of q. In this case, in the calculation for each k_v, N multiplications are performed. Therefore, to obtain q for all k_v, N * (2 * M + 1) multiplications are performed.

  Next, in order to determine the periodicity at each q, each q is binarized. It is preferable that the threshold used when generating the binarized signal is set according to each k_v. This is because the amplitude of q varies depending on each k_v.

The threshold is, for example, the following formula: threshold = f * √ (1 / N) Σq (i) ²
It can be obtained more. For example, f is 0.2.

(1 / N) Σq (i) ² is expressed as Q
Q = (1 / N) * Σq ²
= k_v * k_v * (1 / N) * ΣIR_signal ²
-2 * k_v * (1 / N) * ΣR_signal * IR_signal + (1 / N) * ΣR_signal ²
It is.

Therefore, (1 / N) * ΣIR_signal ² , (1 / N) * ΣR_signal * IR_signal, and (1 / N) * ΣR_signal ² need only be obtained once. However, since the calculation of Q for each k_v requires 4 multiplications, every k_v requires 4 * (2 * M + 1) multiplications, which is very computationally intensive. To be more.

  Furthermore, if more accuracy is required and k_v is changed finely to examine the variation in the pulse width of q (i), the amount of calculation will increase significantly.

  Therefore, in this embodiment, in order to reduce the amount of calculation, q and Q are obtained by the following procedure.

  q and Q corresponding to the m-th (m is an integer in the range of -M to M) k_v are expressed as q (m, i) and Q (m), and q (0, i) when m = 0 , Q (0), and T (0) necessary for obtaining Q (m) when m is not 0 are obtained by the following equations (A1) to (A3).

q (0, i) = R_signal (i) -k_v (0) * IR_signal (i) (A1)
Q (0) = k_v (0) ² * SUM_sq_IR-2 * k_v (0) * SUM_prdct_R_IR + SUM_sq_R (A2)
T (0) = 2 * k_v (0) * dkv_SUM_sq_IR-2 * dkv_SUM_prdct_R_IR (A3)
The following values are calculated in advance.

SUM_sq_R = (1 / N) * ΣR_signal (i) ² (A4)
SUM_sq_IR = (1 / N) * ΣIR_signal (i) ² (A5)
SUM_prdct_R_IR = (1 / N) * ΣR_signal (i) * IR_signal (i) (A6)
Next, q and Q when 1 ≦ m ≦ M (increase from m = 0) are obtained by the following equations (B1) to (B3).

q (m, i) = q (m-1, i) + dkv_IR (i) (B1)
Q (m) = Q (m-1) + T (m-1) (B2)
T (m) = T (m-1) + U (B3)
Further, q and Q when −M ≦ m ≦ −1 (decrease from m = 0) are obtained by the following equations (B4) to (B6).

q (m, i) = q (m + 1, i) -dkv_IR (i) (B4)
Q (m) = Q (m + 1) -T (m + 1) + U (B5)
T (m) = T (m + 1) -U (B6)
The following values are calculated in advance.

dkv_IR (i) = Δk_v * IR_signal (i) (A7)
U = 2 * sq_dkvSUM_sq_IR (A8)
dkv_SUM_prdct_R_IR = Δk_v * SUM_prdct_R_IR (A9)
dkv_SUM_sq_IR = Δk_v * SUM_sq_IR (A10)
sq_dkvSUM_sq_IR = Δk_v * dkv_SUM_sq_IR (A11)
Dkv_IR (i) may be calculated every time IR_signal (i) is calculated.

  Thus, if q (0, i) is calculated using equation (A1) and dkv_IR (i) is calculated using equation (A7), q (m, i) is calculated using equation (B1) and As shown by the equation (B4), it can be executed only by addition and subtraction.

  Similarly, Q (0) is calculated using equation (A2), T (0) is calculated using equation (A3), and U is calculated using equation (A8). As shown in the formulas (B2) and (B3), and the formulas (B5) and (B6), it can be executed only by addition and subtraction. Therefore, the calculation amount can be greatly reduced.

Here, Formula (B2), Formula (B3), Formula (B5), and Formula (B6) will be described.
Q (m) (formula B (2)) when m is increased can be obtained as follows.
Q (m) = {k_v '(m)} ² * SUM_sq_IR
-2 * k_v '(m) * SUM_prdct_R_IR + SUM_sq_IR
= {k_v '(m-1) + Δk_v} ² * SUM_sq_IR
-2 * {k_v '(m-1) + Δk_v} * SUM_prdct_R_IR + SUM_sq_IR
= {k_v '(m-1)} ² * SUM_sq_IR + 2 * k_v' (m-1) * dkv_SUM_sq_IR + sq_dkv_SUM_sq_IR
-2 * k_v '(m-1) * SUM_prdct_R_IR-2 * dkv_SUM_prdct_R_IR + SUM_sq_IR
here
{k_v '(m-1)} ² * SUM_sq_IR-2 * k_v' (m-1) * SUM_prdct_R_IR + SUM_sq_IR = Q (m-1)
Q (m) = Q (m-1) + 2 * k_v '(m-1) * dkv_SUM_sq_IR + sq_dkv_SUM_sq_IR-2 * dkv_SUM_prdct_R_IR
2 * k_v '(m) * dkv_SUM_sq_IR + sq_dkv_SUM_sq_IR-2 * dkv_SUM_prdct_R_IR = T (m)
2 * k_v '(m-1) * dkv_SUM_sq_IR + sq_dkv_SUM_sq_IR-2 * dkv_SUM_prdct_R_IR = T (m-1)
Q (m) = Q (m-1) + T (m-1)
Also,
T (m) = 2 * k_v '(m) * dkv_SUM_sq_IR + sq_dkv_SUM_sq_IR-2 * dkv_SUM_prdct_R_IR
= 2 * k_v '(m-1) * dkv_SUM_sq_IR + sq_dkv_SUM_sq_IR-2 * dkv_SUM_prdct_R_IR
+ 2 * sq_dkv_SUM_sq_IR
= T (m-1) + 2 * sq_dkv_SUM_sq_IR
It is.

When expressed as 2 * sq_dkv_SUM_sq_IR = U, T (m) can be obtained by the following recurrence formula.
T (m) = T (m-1) + U
Therefore, Q (m) can be obtained sequentially by the following two recurrence formulas.
Q (m) = Q (m-1) + T (m-1)
T (m) = T (m-1) + U
Next, Q (m) (formula B (5)) when m is decreased can be obtained as follows.
From T (m) = T (m-1) + U when increasing m
T (m-1) = T (m) -U
So substituting this into Q (m) = Q (m-1) + T (m-1)
Q (m) = Q (m-1) + T (m) -U
Q (m-1) = Q (m) -T (m) + U
If m is replaced by m + 1, Q (m) can be obtained by sequentially decreasing m from −1 using the following recurrence formula.
Q (m) = Q (m + 1) -T (m + 1) + U
T (m) = T (m + 1) -U
DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, an embodiment of the invention will be described with reference to the drawings. In addition, the structure which attached | subjected the same code | symbol in each figure shows that it is the same structure, The description is abbreviate | omitted suitably.

<Configuration>
First, the configuration of the embodiment of the present invention will be described. FIG. 1 is a block diagram illustrating a configuration of a biological information measuring device according to an embodiment. In addition, the solid line in the figure represents the flow between each block of the electric signal component corresponding to the time-series data of the pulse group described later.

  This biological information measuring device 40 uses time-series data of the intensity of light transmitted or reflected from the living body of the measurement target (subject) in order to measure biological information from which the influence of noise components due to body movement or the like is removed. Thus, it is a measuring device that measures biological information such as blood oxygen saturation, for example, based on time-series data from which the influence on the biological information measurement value due to body movement or the like to be measured is removed and the noise component is removed.

  More specifically, in accordance with one aspect of the present invention, the biological information measuring device main body of the biological information measuring device 40 irradiates the living body with a plurality of lights having different wavelengths from each other and transmits or reflects the living body. The biological information of the living body is measured based on each measurement data obtained by receiving each light. The biological information measuring apparatus main body removes a noise component from the data by using the periodicity of the signal component from the data including the signal component having a periodicity generated based on the measurement data, and converts the data into the data from which the noise component has been removed. A measurement unit 10 that measures biological information of a living body is provided.

  That is, the biological information measuring device 40 includes a data acquisition unit 1 that acquires each measurement data by irradiating a living body with a plurality of lights having different wavelengths and transmitting or reflecting the light transmitted through or reflected from the living body. The living body information measuring apparatus main body for measuring living body information of the living body based on each measurement data obtained by the acquisition unit 1.

  Hereinafter, the configuration of the biological information measuring device 40 will be described in more detail. The biological information measurement device 40 includes, for example, a data acquisition unit 1, a measurement unit 10, and an output unit 30, as shown in FIG.

  The data acquisition unit 1 is a device for measuring and acquiring time series data related to pulsation of a living body necessary for measuring biological information in the measuring unit 10 at a predetermined time interval (measurement cycle). Here, as a method for detecting pulsation, various methods can be adopted. For example, a method using the light absorption characteristic of hemoglobin in a living tissue can be preferably employed. As is well known, oxygen is carried by hemoglobin to each cell in the living body, but hemoglobin combines with oxygen in the lung to become oxygenated hemoglobin, and returns to hemoglobin when oxygen is consumed by cells in the living body. Oxygen saturation refers to the proportion of oxygenated hemoglobin in the blood. The absorbance of these hemoglobins and the absorbance of oxyhemoglobin are wavelength-dependent.For example, hemoglobin absorbs more light than oxyhemoglobin for red light with a wavelength R in the red region, but the wavelength in the infrared region. It absorbs less light than oxyhemoglobin for IR infrared light. This method obtains biological information such as blood oxygen saturation and pulse rate by utilizing the difference in absorption characteristics of hemoglobin and oxyhemoglobin with respect to red light and infrared light. For example, as shown in FIG. 1, the data acquisition unit 1 includes a light emitting element (R) that irradiates a predetermined living tissue with red light (hereinafter, R), and a light emitting element (R) that is irradiated with the light emitting element (R). A sensor (R) unit 2 including a light receiving element (R) that receives light transmitted or reflected through a biological tissue, and a light emitting element (IR) that irradiates the predetermined biological tissue with infrared light (hereinafter, IR); And a reflective or transmissive sensor including a sensor (IR) unit 3 having a light receiving element (IR) that receives each light that has been irradiated with the light emitting element (IR) and transmitted or reflected through the biological tissue to be measured. is there. The data acquisition unit 1 having such a configuration is set in a predetermined living tissue, monitors each received light amount by the light receiving element (R, IR), and converts each received light into an electrical signal according to the light intensity. Each time series data related to the pulse wave is acquired by performing photoelectric conversion. The data acquisition unit 1 is connected to the measurement unit 10 and outputs these time series data to an AC / DC (R) unit 11 described later.

  The output unit 30 is a device that is connected to the measurement unit 10 and outputs measurement values of biological information related to the living body measured by the measurement unit 10. The output unit 30 is, for example, a display device such as a liquid crystal display (LCD), a 7-segment LED, an organic photoluminescence display device, a CRT (Cathode Ray Tube) display device, a plasma display device, a microphone, or a speaker. A sound output device such as a printer or a printing device such as a printer. For example, various pieces of measurement information such as data analysis results of pulse wave data are output in an arbitrary form such as light lighting (including turning on / off and blinking), characters, images, sounds, and printing.

  For example, as shown in FIG. 1, the output unit 30 includes a SpO2 output unit 31, a pulse rate output unit 32, and a reliability output unit 33. The SpO2 output unit 31 is a device for outputting the blood oxygen saturation calculated by the SvO2 estimation SpO2 determination unit 19 described later. The pulse rate output unit 32 is a device for outputting the pulse rate calculated by the pulse rate calculation unit 20 described later. And the reliability output part 33 is an apparatus for outputting the reliability calculated by the reliability calculation part 21 mentioned later.

For example, as shown in FIG. 1, the measurement unit 10 is functionally an AC / DC (R) conversion unit 11, an AC / DC (IR) conversion unit 12, a BPF (R) unit 13, and a BPF (IR). Unit 14, ΣR_signal * IR_signal calculation unit 15, ΣR ² calculation unit 16, ΣIR ² calculation unit 17, initial value calculation unit 18, SvO2 estimation SpO2 determination unit 19, pulse rate calculation unit 20, reliability calculation unit 21, and difference calculation unit 22 and controls the data acquisition unit 1 and the output unit 30 according to the function according to a control program stored in advance.

  The measurement unit 10 calculates initial values required for noise component removal and biological information calculation after performing predetermined preprocessing on the time-series data of the pulse wave acquired by the data acquisition unit 1, Using the initial value, noise component removal processing is performed to remove the noise component included in the time series data of the pulse wave, and the biological information is measured based on the time series data of the pulse wave after the noise component is removed. It is a device that outputs measurement values of biological information to the output unit 30. The measurement unit 10 includes, for example, a ROM (Read Only Memory) that stores arithmetic processing programs for performing arithmetic processing, which will be described later, a RAM (Random Access Memory) that functions as a so-called working memory, and temporarily stores data. A central processing unit (CPU) that reads a processing program from the ROM and executes it, and its peripheral circuits.

  The AC / DC (R) conversion unit 11 converts the red light (R) time-series data input from the sensor (R) unit 2 of the data acquisition unit 1 into data that is a ratio of a direct current component and an alternating current component. It is a circuit to convert. In the present embodiment, the AC / DC (R) conversion unit 11 performs dark processing for removing components due to dark current as the preprocessing.

The BPF (R) unit 13 receives time-series data of red light (R) from the AC / DC (R) conversion unit 11 and removes noise components to obtain frequency components for red light (R). It is a band pass filter (BPF). The BPF (R) unit 13 outputs R_signal, which is time-series data after filtering, to the ΣR_signal * IR_signal calculation unit 15, the Σ (R_signal) ² calculation unit 16, and the SvO2 estimation SpO2 determination unit 19, respectively.

  The AC / DC (IR) conversion unit 12 is data that is a ratio of a DC component to an AC component with respect to time series data of infrared light (IR) input from the sensor (IR) unit 3 of the data acquisition unit 1. It is a circuit that converts to In the present embodiment, the AC / DC (IR) conversion unit 12 performs dark processing for removing components due to dark current as the preprocessing.

The BPF (IR) unit 14 receives time series data of infrared light (IR) from the AC / DC (IR) conversion unit 12 and removes noise components to obtain frequency components for infrared light (IR). This is a bandpass filter for the purpose. The BPF (IR) unit 14 outputs IR_signal, which is time-series data after filtering, to the ΣR_signal * IR_signal calculation unit 15, the ΣIR_signal ² calculation unit 17, and the SvO2 estimation SpO2 determination unit 19, respectively.

  The ΣR_signal * IR_signal calculation unit 15 includes time-series data of red light (R) input from the BPF (R) unit 13 and time-series data of infrared light (IR) input from the BPF (IR) unit 14. Is used to calculate the cross-correlation ΣR_signal * IR_signal and output the calculated cross-correlation ΣR_signal * IR_signal to the initial value calculation unit 18 and the reliability calculation unit 21.

The Σ (R_signal) ² calculating unit 16 calculates autocorrelation Σ (R_signal) ² using the time series data of the red light (R) input from the BPF (R) unit 13, and calculates the calculated autocorrelation Σ ( R_signal) ² is output to the initial value calculation unit 18 and the reliability calculation unit 21.

The Σ (IR_signal) ² calculating unit 17 calculates autocorrelation Σ (IR_signal) ² using time series data of infrared light (IR) input from the BPF (IR) unit 14, and calculates the calculated autocorrelation Σ This is a circuit that outputs (IR_signal) ² to the initial value calculation unit 18 and the reliability calculation unit 21.

  The difference calculation unit 22 uses time series data of red light (R) input from the BPF (R) unit 13 and time series data of infrared light (IR) input from the BPF (IR) unit 14. Q (0, i) and dkv_IR (i) (refer to equations (A1) and (A7)) are calculated, and the calculated q (0, i) and dkv_IR (i) are sent to the SvO2 estimation SpO2 determination unit 19. It is a circuit to output.

The initial value calculation unit 18 is a circuit that calculates an initial value and a value to be used in the subsequent calculation. More specifically, the cross-correlation ΣR_signal * IR_signal, Σ (R_signal) input from the ΣR_signal * IR_signal calculation unit 15. ^second calculating unit 16 autocorrelation Σ (R_signal) ² input from, and, with Σ (IR_signal) ² calculator 17 autocorrelation Σ (IR_signal) ² inputted from, for use in the calculation of the initial value and after A value (hereinafter may be collectively referred to as “initial value”) is calculated, and the calculated initial value or the like is output to the SvO2 estimation SpO2 determination unit 19.

  The SvO2 estimation SpO2 determination unit 19 removes the noise component superimposed on the signal component using the periodicity from the time series data including the signal component having periodicity, and based on the time series data from which the noise component is removed, This is a circuit for calculating the blood oxygen saturation level. The SvO2 estimation SpO2 determination unit 19 corresponds to the signal generation unit and the estimation unit disclosed in the means for solving the problem. More specifically, the SvO2 estimation SpO2 determination unit 19 inputs the time series data R_signal input from the BPF (R) unit 13, the time series data IR_signal input from the BPF (IR) unit 14, and the difference calculation unit 22. Is a circuit for calculating the arterial blood oxygen saturation based on the initial value calculated by the initial value calculation unit 18 from the time series data q (0, i) and dkv_IR (i) Then, the blood oxygen saturation level of the vein is calculated based on the noise component, and the calculated blood oxygen saturation level is output to the SpO2 output unit 31.

  The pulse rate calculation unit 20 calculates the pulse rate based on the time-series data from which the noise component is removed by the SvO2 estimation SpO2 determination unit 19, and outputs the calculated pulse rate to the pulse rate output unit 32.

  The reliability calculation unit 21 is a circuit that calculates the reliability indicating the degree of error for the measured biological information, and outputs the calculated reliability to the reliability output unit 33.

  Note that the biological information measuring device 40 may further include an external storage unit (not shown) as necessary. The external storage unit includes, for example, a memory card, a flexible disk, a CD-ROM (Compact Disc Read Only Memory), a CD-R (Compact Disc Recordable), a DVD-R (Digital Versatile Disc Recordable), and a Blu-ray Disc (Blue-ray Disc). For example, a memory card interface, a flexible disk drive, a CD-ROM drive, a CD-R drive, a DVD-R drive, and a Blu-ray disk drive. Etc.

  Here, when an arithmetic processing program or the like is not stored in the measurement unit 10, the biological information measurement device 40 is installed in the measurement unit 10 from the recording medium on which these programs are recorded via the external storage unit. You may be comprised so that. Alternatively, the biological information measuring device 40 may be configured such that data such as detected biological information is recorded on a recording medium via the external storage unit.

<Operation>
Next, the operation of this embodiment will be described with reference to FIGS.

  The biological information measurement process flowchart (steps S10 to S23) in FIG. 2 mainly performs three processes. Step S10 shows an initial process, steps S11 to S20 show a blood oxygen saturation detection process, and step S21 shows a pulse rate detection process. Step S22 shows a process for displaying the calculated blood oxygen saturation, pulse rate, etc., and step S23 shows a process for controlling the blood oxygen saturation etc. to be measured in the measurement cycle. FIG. 3 is a flowchart showing details of the initial processing, and FIGS. 4 to 8 are flowcharts showing details of some of the blood oxygen saturation detection processing and diagrams for explanation. . FIG. 9 is a flowchart showing details of the pulse rate detection process.

  The biological information measuring device 40 executes the arithmetic processing program by, for example, activation thereof. Each part (11-22) of the measurement part 10 is functionally comprised by execution of this arithmetic processing program. The biological information measuring device 40 measures biological information such as blood oxygen saturation with reduced noise components based on the time-series data acquired by the data acquisition unit 1 by the following operation.

<Initial processing (step 10)>
The biological information measuring device 40 first performs an initial process (step S10). In the initial process, four processes are mainly performed. One is a process of collecting a predetermined number of time-series data. The second is a process for calculating and storing a value (difference of q to be described later) used in the process of obtaining k_v, and the third is a process for calculating a fixed value (such as a constant starting with SUM) used in the calculation. The fourth is a process for calculating an initial value.

  Details of the initial processing will be described with reference to FIG. FIG. 3 is a flowchart showing the initial processing.

  When the measurement of the biological information measuring device 40 is started, the measurement unit 10 receives from the data acquisition unit 1 the component R_dark due to the dark current of the sensor (R) unit 2 and the dark current of the sensor (IR) unit 3. Each component IR_dark is input. Then, R_signal_and_dark (i) and IR_signal_and_dark (i), which are data time series related to the intensity change of the light with the wavelength IR and the wavelength R transmitted or reflected through the living tissue, are input from the data acquisition unit 1 to the measurement unit 10. The Here, i represents the i-th time series data from 1 to N, and the correspondence between i and time is as described in the principle of the invention. As N, the number of data necessary for calculating the oxygen saturation is selected, and a value such as 200 is used. For the sake of simplicity, hereinafter, i in time series data may be omitted. The data R_signal_and_dark related to the wavelength R includes the component R_dark due to the dark current of the sensor (R) unit 2, and the data IR_signal_and_dark related to the wavelength IR includes the component IR_dark due to the dark current of the sensor (IR) unit 3. .

  Next, the AC / DC (R) unit 11 performs dark processing by subtracting R_dark from R_signal_and_dark. Similarly, the AC / DC (IR) unit 12 performs dark processing for subtracting IR_dark from IR_signal_and_dark (step S30).

  Next, the dark-processed signal of each wavelength is converted into a ratio between the DC component and the AC component (AC component / DC component) (step S31). More specifically, the IR_dark component is removed from IR_signal_and_dark, and IR_signal, which is the ratio of the direct current component consisting only of the intensity of light derived from the living body to the alternating current component, is calculated. Similarly, the R_dark component is removed from R_signal_and_dark, and R_signal, which is the ratio of the direct current component consisting only of the intensity of light derived from the living body to the alternating current component, is calculated.

  Next, the BPF (R) unit 13 performs processing for filtering R_signal and removing unnecessary frequency components to obtain only desired frequency components. Similarly, the BPF (IR) unit 14 performs processing for filtering IR_signal and removing unnecessary frequency components to obtain a desired frequency component signal component (step S32).

  Next, the difference calculation unit 22 calculates q (0, i) using Expression (A1), and calculates dkv_IR (i) from IR_signal (i) using Expression (A7) ( Step S33). The difference calculation unit 22 outputs the calculated q (0, i) and dkv_IR (i) to the SvO2 estimation SpO2 determination unit 19.

Further, with respect to the data R_signal and IR_signal extracted in step S32, the ΣR_signal * IR_signal calculation unit 15, the Σ (R_signal) ² calculation unit 16, and the Σ (IR_signal) ² calculation unit 17 perform cross-correlation ΣR_signal * IR_signal and wavelength R. autocorrelation Shigumaaru_signal ² and autocorrelation ShigumaIR_signal ² for the wavelength IR is calculated respectively, and outputs the initial value calculation section 18. Note that Σ is a sum taken from 1 to N with respect to i, and it is assumed that k_v and k_a are constant over a short time interval.

  Next, when the initial value calculation unit 18 collects a predetermined number (N) of data (step S34: Yes), the SUM_sq_R, SUM_sq_IR, and SUM_prdct_R_IR are expressed by the equations (A4), (A5), and Calculation is performed using (A6) (step S35).

  Next, the initial value calculation unit 18 substitutes an appropriate value for k_v_0 that is an initial value of k_v (step S36). From the second measurement, kv_min obtained last time is stored and substituted.

  Next, using the SUM_sq_R, SUM_sq_IR, SUM_prdct_R_IR, and k_v_0 obtained in steps S35 and S36, the initial value calculation unit 18 uses the temporary values k_a_0 and k_ns of the k_a that are necessary in the blood oxygen saturation detection flow. Each value k_ns_0 is calculated (step S37). The initial value calculation unit 18 outputs the calculated values to the SvO2 estimation SpO2 determination unit 19.

  The SvO2 estimation SpO2 determination unit 19 stores SUM_sq_R, SUM_sq_IR, SUM_prdct_R_IR, k_v_0, k_a_0, and k_ns_0 in the memory or the like (not shown).

  Here, k_ns is an index for determining whether or not to extract a noise component, and is represented by a ratio between the strength of the noise component and the strength of the signal component, for example, as shown in Expression (9). Is done.

Here, n_square, which is a value related to the intensity of the noise component, and s_square, which is a value related to the intensity of the signal component, are given by different expressions depending on whether or not an equal sign holds by substituting k_v_0 into Expression (8).

  When k_v_0 is substituted into equation (8) and the value of k_a such that the equal sign holds (that is, the left side becomes equal to zero) is obtained, the value of k_a is set to k_a_0, and the following equation ( 10) and equations (11) and (9) are used to calculate k_ns_0.

  On the other hand, if k_v_0 is substituted into the equation (8) and a physiologically valid k_a for which the equal sign holds is not obtained, an appropriate k_a such as the previous k_a is set to k_a_0, and the following equations (12) and K_ns_0 is calculated using (13) and Equation (9). Further, assuming that the difference in oxygen saturation between arterial blood and venous blood is 10%, for example, a value obtained by subtracting a corresponding value from k_v_0 may be k_a_0.

<Blood oxygen saturation detection (steps S11 to S20)>
After the above-described initial process (step S10 in FIG. 2) is completed, the blood oxygen saturation detection process is executed by the SvO2 estimated SpO2 determination unit 19.

  The SvO2 estimation SpO2 determination unit 19 first determines the noise component state by comparing k_ns_0, which is an index of the noise component, with a predetermined value (step S11). Processing is divided into the following two types according to the determination result. The predetermined value is appropriately selected according to the level of the noise component to be removed, and is, for example, 0.05 in this embodiment.

  When the SvO2 estimation SpO2 determination unit 19 determines that the noise component index is equal to or less than a predetermined value (step S11: No), for example, k_a is known as a change in the absorbance of the wavelength IR, which is known as a conventional technique. K_a is calculated using equation (14), which is a method for obtaining the average value of the ratio between the signal component of the minute and the signal component of the change in absorbance at wavelength R (step S12), and then step S20 is executed. .

  It should be noted that the equations (8) to (14) may be obtained by replacing R_signal and IR_signal with time differences, respectively.

  On the other hand, when the SvO2 estimation SpO2 determination unit 19 determines that the noise component index is larger than the predetermined value (step S11: Yes), kv_min has been calculated, that is, the biological information measurement device 40 It is determined whether or not the first measurement process has been started (step S12).

  This determination is made with reference to the calculated flag. When the calculated flag is ON, indicating that “calculation has been performed” (step S12: Yes), the stored kv_min is used (step S15). If the calculated flag is OFF, indicating that “calculation has not been performed” (step S12: No), step S16 is executed, kv_min is calculated, and then step S18 is executed. . The kv_min calculation process will be described later in <kv_min calculation process>.

  When kv_min is obtained, it is stored in the memory, the calculated flag is turned on and "calculated" is set (step S17).

  Next, the SvO2 estimation SpO2 determination unit 19 calculates k_a by substituting kv_min obtained in step S16 into equation (8) and solving this equation (step S18).

The SvO2 estimation SpO2 determination unit 19 that has calculated k_a next calculates k_ns, which is an index of the noise component, using Equation (9) (step S19). For calculating k_ns, ΣR_signal * IR_signal, ΣR_signal ² and ΣIR_signal ² stored in the memory in the initial process (step S10) are used.

  Next, the SvO2 estimation SpO2 determination unit 19 calculates the reliability of k_a or the reliability of k_v, and calculates the blood oxygen saturation based on k_a (step S20). Here, the reliability of k_a is a value representing the degree of error included in k_a. Similarly, the reliability of k_v is a value indicating the degree of error included in k_v. These reliability calculation processes will be described later in <Reliability Calculation Process>.

  The SvO2 estimation SpO2 determination unit 19 stores kv_min in preparation for the next measurement.

  Next, the pulse rate calculation unit 20 calculates the pulse rate PR (step S21). The pulse rate PR calculation process will be described later in <Pulse rate PR calculation process>.

  The SpO2 output unit 31, the pulse rate output unit 32, and the reliability output unit 33 output the blood oxygen saturation, the pulse rate PR, and the reliability of k_a detected in steps S20 and S21, respectively (step S22).

  The reliability output unit 33 may output the reliability of k_v instead of the reliability of k_a. In addition, the measurement unit 10 compares k_ns calculated in step S19 with a predetermined value (predetermined value in step S11), and if k_ns exceeds a predetermined value (predetermined value in step S11), k_ns is A warning for warning that the predetermined value has been exceeded is output to the SpO2 output unit 31. Furthermore, the measurement unit 10 compares the reliability of k_a with a preset value for determining the reliability of k_a. If k_a exceeds the value, k_a indicates that k_a has exceeded the value. The warning may be further output to the reliability output unit 33. Further, the measurement unit 10 uses the reliability of k_v in place of the reliability of k_a, compares the reliability of k_v with a preset value for determining the reliability of k_v, and k_v determines the value. If it exceeds, a warning indicating that k_v has exceeded that value may be further output to the reliability output unit 33.

  The measurement unit 10 waits until it detects from the internal timer that the measurement period has elapsed (step S23: No), and if it detects that the measurement period has elapsed (step S23: Yes), has the k_v calculation period elapsed? It is determined whether or not (step S24).

  If the k_v calculation cycle has elapsed (step S24: Yes), the calculated flag is turned off (step S25). If it has not elapsed (step S24: No), the calculated flag is left on. Keep it. And a process is started from step S10.

  Here, the k_v calculation cycle is a cycle in which kv_min is calculated, and is a time during which k_v is assumed not to change. Since k_v does not change abruptly, the calculation cycle of kv_min is made longer than the calculation cycle of k_a and SpO2. For example, it is assumed that k_a and SpO2 are calculated every second, for example (corresponding to the measurement period of step 23). On the other hand, the cycle for calculating kv_min is set to a cycle (for example, 2 seconds) longer than the calculation cycle of k_a and SpO2 (corresponding to the k_v calculation cycle in step S24). By doing so, it is possible to reduce the amount of calculation by reducing the number of times kv_min is calculated without changing the correctness of the measurement result.

<Reliability calculation processing>
In general, it is known that the error contained therein varies depending on the value of the calculated arterial oxygen saturation, for example, the arterial oxygen saturation is approximately 95% and 70%. The degree of error in degrees is different. Therefore, the predetermined value of k_ns is changed as follows. The predetermined value of k_ns is set to a small value when the arterial oxygen saturation is smaller than the predetermined arterial oxygen saturation (when k_a and k_v are large and the venous blood oxygen saturation is small) On the other hand, if the arterial oxygen saturation is greater than the predetermined arterial oxygen saturation (k_a and k_v are small and the venous blood oxygen saturation is large), the predetermined value of k_ns is set large. Good. By changing the predetermined value of k_ns according to the value of arterial oxygen saturation calculated in this way, for example, when arterial oxygen saturation is large, the threshold value is increased, and the error in arterial oxygen saturation is Warning and output prohibition can be performed with almost the same value as when the degree of saturation is small. Here, for the predetermined arterial blood oxygen saturation and the predetermined value of k_ns corresponding to the predetermined arterial oxygen saturation, values determined based on past measurement data may be stored in the measurement unit 10 in advance. Alternatively, it may be appropriately changed according to the living body or the state of the living body. In addition to k_ns, for example, a value z given by any one of Equations (16) to (21) may be used as an index of the reliability of the obtained k_a or k_v. When the absolute value of z is large, the reliability of k_a is low. Further, when the absolute value of z is large, the reliability of k_v is low.

  Further, as a substitute for the reliability index of k_a, the variation of the value w given by the following equation (22) (standard deviation, maximum value−minimum value, etc.) may be used. A small variation in the value w indicates that the reliability of the obtained k_a is high.

  In addition, the reliability of k_a may be evaluated from the variation of y with respect to the regression line of (x, y), where R_signal is y and IR_signal is x, the difference between the maximum value and the minimum value, or the like. Expressions (16) to (22) may be obtained by replacing R_signal and IR_signal with time differences, respectively.

<Kv_min calculation process>
The kv_min calculation process in step S16 will be described with reference to FIG.

  As described above in <Prior Art as a Premise of the Present Invention>, k_v selected so that q corresponding to the left side of Expression (4) (see Expression (15) below) has the strongest periodicity is Is the optimal k_v.

  Here, q represents a value for the left side of Equation (15), which is the i-th data series, for a certain k_v.

  More specifically, how equation (15) changes according to the value of k_v will be described below with reference to FIG.

  FIG. 5 is a diagram illustrating a waveform of R-kv * IR based on measurement data. FIG. 5A shows a waveform according to the data sequence given by equation (15) for k_v when k_v is given by a value smaller than the optimum value, and FIG. 5B shows that k_v is approximated to the optimum value. FIG. 5 (c) shows an equation for k_v when k_v is given by a value larger than the optimum value (FIG. 5C). The waveform by the data series given by 15) is shown. Note that the waveform is interpolated between the data of the time series data. 5A to 5C, these horizontal axes are predetermined short times (corresponding to data series from 1 to N in the data series number i), and the vertical axis is q in Expression 15. .

  When k_v smaller than the optimum value is given to equation (15), when the value k_v approximated to the optimum value is given to equation (15) with the waveform shown in FIG. The waveform is as shown in b). As can be seen by comparing these two figures, when the value k_v approximated to the optimum value is substituted (in FIG. 5B), the waveform has strong periodicity even when the noise component is superimposed on the signal component. However, when k_v smaller than the optimum value is given in Equation (15), if the noise component is superimposed on the signal component, the waveform is disturbed and the periodicity is weak, and is larger than the optimum value. When k_v is given in equation (15), the waveform is as shown in Fig. 5 (c), as can be seen by comparing Fig. 5 (c) with Fig. 5 (b). Is given to equation (15) (FIG. 5C), if a noise component is superimposed on a signal component, the waveform is disturbed and the periodicity is weak.

  In order to use the change in the equation (15) with respect to k_v, the periodicity of the equation (15) is detected.

  FIG. 4 is a flowchart of the kv_min calculation process.

  In the kv_min calculation process of the embodiment, the initial value of k_v is k_v_s, and k_v is within a predetermined range (for example, k_v_s-0.5 <k_v <k_v_s + 0.5), Δk_v by M, plus and minus by M Change to obtain (2 * M + 1) q (see equation (15)). Then, the width of each waveform is calculated by binarizing each q.

  Hereinafter, the kv_min calculation process will be specifically described.

  First, the SvO2 estimation SpO2 determination unit 19 sets an initial value in this process (step S40). A counter m indicating the number of times k_v is changed is set to 0 (zero), and a variable Var_PW_min indicating the minimum value of the pulse width variation index is set to a value that is not possible as a variation index, for example, 100. Further, for example, 1 is set to k_v_s for the first time, and the previous kv_min is set for the second time.

  Next, a value used when calculating a threshold suitable for each q is calculated and stored in a memory or the like (step S41). Specifically, Q (0) is calculated using equation (A2), T (0) is calculated using equation (A3), and U is calculated using equation (A8). The values necessary for these calculations are calculated in the initial process and stored in the memory. SUM_sq_R, SUM_sq_IR, and SUM_prdct_R_IR are read, dkv_SUM_prdct_R_IR is calculated using equation (A9), and dkv_SUM_sq_IR is calculated using (A10) Then, sq_dkvSUM_sq_IR is calculated using Expression (A11).

  Next, q (0, i) is read from the memory (step S42). From the second measurement, q_min (i) corresponding to k_v_min is stored in step S62 described later, and q_min (i) is set as the initial value q (0, i). Thus, an increase in calculation amount can be prevented.

  Next, the process of step S43 to step S50, that is, the process of calculating variation by adding k_v by Δk_v is repeated M times.

  First, Q (m) is calculated by equation (B2), and a threshold value used in binarization processing is calculated (step S43). In this embodiment, a threshold value thresh_1 and a threshold value thresh_2 obtained by the following equations are set from Q (m). Note that f_1 and f_2 are, for example, 0.2 and −0.2, respectively.

Sqr_Q = Q (m) ^0.5
thresh_1 = f_1 * Sqr_Q
thresh_2 = f_2 * Sqr_Q
If m = 0, Q (0) obtained in step S41 is used.

  Next, binarization processing of q (m, i) is performed (step S44). When m = 0, q (0, i) obtained at step S42 is used, and when m> 0, q (m, i) obtained at step S50 is used.

  Next, the variation index Var_PW is calculated from the information obtained by the binarization process (step S44). The binarization process will be described in the later section <Binarization process>, and the variation index calculation process will be described in the later section <Variation index calculation process>.

  The variation index Var_PW obtained in step S45 is compared with the minimum variation index Var_PW_min, and the smaller value is set as the minimum variation index Var_PW_min. If the minimum variation index Var_PW_min is set, k_v of the set Var_PW is set to kv_min (step S46).

  If the processing of step S43 to step S50 has not been repeated M times (step S47: Yes), k_v is added to Δk_v to obtain k_v used in the next processing (step S48).

  Then, Q (m) and T (m) are calculated using the equations (B2) and (B3) (step S49). Further, q (m, i) is calculated using the formula (B1) (step S50).

  When the processing of step S43 to step S50 is repeated M times (step S47: No), the process is changed to the minus side by Δk_v and the process proceeds to the process of calculating the variation.

  First, the SvO2 estimation SpO2 determination unit 19 sets −1 to a counter m indicating the number of times k_v has been changed (step S51), and uses equation (B5) to calculate Q (-1) and equation (B6). To calculate T (-1) (step S52), and calculate q (-1, i) using equation (B4) (step S53).

  Next, the process of step S54 to step S61, that is, the process of calculating variation by subtracting k_v by Δk_v by M times is repeated.

  The processing from step S54 to step S57 is the same as the processing from step S43 to step S46.

  When the processing of step S54 to step S61 has not been repeated M times (step S58: Yes), Δk_v is subtracted from k_v to obtain k_v used in the next processing (step S59).

  Then, Q (m) and T (m) are calculated using the equations (B5) and (B6) (step S60). Further, q (m, i) is calculated using equation (B4) (step S61).

  When the processing from step S54 to step S61 is repeated M times (step S58: No), q (m, i) at the time of the minimum variation index Var_PW_min is stored as q_min (i) (step S62). At this time, the minimum variation index Var_PW_min is set to the minimum variation index Var_PW, and kv_min is set to k_v of the minimum variation index Var_PW_min.

  In this embodiment, Δk_v that is a fixed value is used. First, kv_min_1 that minimizes the variation in the pulse width of binarized q (m, i) is obtained in a wide k_v range using a large Δk_v. Then, kv_min that minimizes the variation in the pulse width of binarized q (m, i) may be obtained by using a small Δk_v in a narrower k_v range around kv_min_1. This makes it possible to determine kv_min with high accuracy without increasing the amount of calculation.

<Binarization processing>
Here, the binarization process in step S44 will be described with reference to FIGS.

  FIG. 6 is a flowchart of the binarization process. FIG. 7 is an enlarged view of the waveform of q for explaining the binarization process.

  In the binarization process, the threshold value thresh_1 and the threshold value thresh_2 obtained in step S43 are used.

  Then, i is stored in the table t1_l_to_h (kr_1) when the q (m, i) exceeds the value below the threshold thresh_1, and i is stored in the table t1_h_to_l (kf_1) when the q (m, i) is below the value exceeding the threshold thresh_1. Similarly, i is stored in table t2_l_to_h (kr_2) when q (m, i) exceeds a value below threshold thresh_2, and i is stored in table t2_h_to_l (kf_2) when it is below a value exceeding threshold thresh_2. (See FIG. 7). kr_1, kf_1, kr_2 and kf_2 are indices starting from 0.

  First, variables used in the binarization process are initialized (step S70). Here, status_1 is a flag that is set to 1 when q (m, i) exceeds the threshold value thresh_1, and is set to 0 in the following cases. status_2 is a flag that is set to 1 when q (m, i) exceeds the threshold thresh_2, and is set to 0 in the following cases. This flag indicates the state of the value of q (m, i-1) immediately before q (m, i) currently being processed. The threshold value is calculated in step S43 in FIG.

  When the flag status_1 is 0 and q (m, i) is larger than the threshold thresh_1, that is, when the threshold thresh_1 is exceeded from a value below the threshold thresh_1 (step S71: Yes), i is stored in the table t1_l_to_h (kr_1). (Step S72). In other cases (step S71: No), nothing is stored.

  When the flag status_1 is 1 and q (m, i) is less than or equal to the threshold thresh_1, that is, when the value exceeding the threshold thresh_1 is less than or equal to the threshold thresh_1 (step S73: Yes), i is stored in the table t1_h_to_l (kf_1). (Step S74). In other cases (step S73: No), nothing is stored.

  Then, it is determined whether q (m, i) exceeds the threshold thresh_1 or below, and 0 or 1 is set to the flag status_1 (step S75).

  Next, similarly, when the flag status_2 is 0 and q (m, i) is larger than the threshold thresh_2, that is, when the threshold thresh_2 is exceeded from a value lower than the threshold thresh_2 (step S76: Yes), i is stored in the table t2_l_to_h. (kr_2) is stored (step S77). In other cases (step S77: No), nothing is stored.

  When the flag status_2 is 1 and q (m, i) is less than or equal to the threshold value thresh_2, that is, when the value exceeds the threshold value thresh_2 and becomes less than or equal to the threshold value thresh_2 (step S78: Yes), i is stored in the table t2_h_to_l (kf_2). (Step S79). In other cases (step S78: No), nothing is stored.

  Then, it is determined whether q (m, i) exceeds the threshold value thresh_2, and 0 or 1 is set to the flag status_2 (step S80).

  The processes in steps S71 to S80 are repeated from i to 1 to N (steps S81 and S82), and the number of i stored in the table such as t1_l_to_h (kr_1) is stored in kr_1_max and the like (step S83).

<Dispersion index calculation process>
Here, the variation index calculation process in step S45 will be described with reference to FIG.

  FIG. 8 is a flowchart of the variation index calculation process.

  In this embodiment, the variation index Var_PW is obtained from the tables t1_l_to_h (kr_1), t1_h_to_l (kf_1), t2_l_to_h (kr_2), and t2_h_to_l (kf_2) obtained by the binarization process.

  Specifically, as shown in FIG. 7, the lengths of a1, a2, b1, and b3 of peaks and valleys of all waveforms are obtained, and the minimum and maximum values of a1, a2, b1, and b3 are obtained. And the harmonic average is calculated | required and it is set as a variation index.

  First, the size of i registered in the table t1_l_to_h (0) and the table t1_h_to_l (0) is compared. This is to determine whether the waveform starts from a mountain or a valley.

  If t1_l_to_h (0)> t1_h_to_l (0) (step S90: Yes), it is determined that it starts from the valley, the index of the table t1_h_to_l is shifted by 1, and a1 is obtained. A1 of all the peaks is calculated, and the maximum value t_1_h_max and the minimum value t_1_h_min are obtained (step S93).

  Further, b1 of all the peaks is calculated, and the maximum value t_1_l_max and the minimum value t_1_l_min are obtained (step S94).

  On the other hand, if t1_l_to_h (0) ≦ t1_h_to_l (0) (step S90: No), it is determined that it starts from the mountain, a1 of all the mountains is calculated, and the maximum value t_1_h_max and the minimum value t_1_h_min are obtained. (Step S91).

  Further, the index of the table t1_h_to_l is shifted by 1, and b1 of all the peaks is calculated, and the maximum value t_1_l_max and the minimum value t_1_l_min are obtained (step S92).

  Next, the size of i registered in the table t2_l_to_h (0) and the table t2_h_to_l (0) is compared. This is to determine whether the waveform starts from a mountain or a valley.

  If t2_l_to_h (0)> t2_h_to_l (0) (step S95: Yes), it is determined that it starts from the valley, the index of the table t1_h_to_l is shifted by 1, and a2 is obtained. The a2 of all the peaks is calculated, and the maximum value t_2_h_max and the minimum value t_2_h_min are obtained (step S98).

  Further, b2 of all the peaks is calculated, and the maximum value t_2_l_max and the minimum value t_2_l_min are obtained (step S99).

  On the other hand, if t2_l_to_h (0) ≦ t2_h_to_l (0) (step S95: No), it is determined that it starts from the mountain, a2 of all the mountains is calculated, and the maximum value t_2_h_max and the minimum value t_2_h_min are obtained. (Step S96).

  Further, the index of the table t2_h_to_l is shifted by 1, and b2 of all the peaks is calculated, and the maximum value t_2_l_max and the minimum value t_2_l_min are obtained (step S97).

  Next, a variation index Var_PW is obtained from the maximum value t_1_l_max, the minimum value t_1_l_min, the maximum value t_1_h_max, the minimum value t_1_h_min, the maximum value t_2_l_max, the minimum value t_2_l_min, the maximum value t_2_h_max, and the minimum value t_2_h_min (step S100).

  Specifically, first, an average harmonic t_l_max_min between (t_1_l_max-t_1_l_min) and (t_2_l_max-t_2_l_min) is obtained by the following equation, for example.

t_l_max_min = {(t_1_l_max-t_1_l_min) * (t_2_l_max-t_2_l_min)}
/ {(t_1_l_max-t_1_l_min) + (t_2_l_max-t_2_l_min)}
Similarly, an average harmony t_h_max_min between (t_1_h_max-t_1_h_min) and (t_2_h_max-t_2_h_min) is obtained.

  The larger one of t_l_max_min and t_h_max_min is set as a variation index Var_PW.

<Pulse rate PR calculation process (step S21)>
After the above-described blood oxygen saturation detection process (steps S11 to S20 in FIG. 2) is completed, the pulse rate calculation unit 20 executes the pulse rate PR calculation process.

  FIG. 9 is a flowchart of the pulse rate PR calculation process.

  In the embodiment, the average T_ave of the state change intervals of q_min (i) stored in step S62 of FIG. 4 is calculated to obtain the pulse rate PR.

  First, binarization processing is performed on q_min (i) (steps S110 to S118). The threshold value is set to 0 (zero), and the same process as the binarization process in FIG. 6 is performed. The threshold used in this case may be the threshold used when determining k_v.

  Specifically, variables used in the binarization process are initialized (step S110). Here, status is a flag that is set to 1 when q_min (i) exceeds the threshold value, and is set to 0 when the value is equal to or less than the threshold value. This flag indicates the state of the value of q_min (i-1) immediately before q_min (i) currently being processed.

  When the flag status is 0 and q_min (i) is larger than the threshold value, that is, when the threshold value is exceeded from a value below the threshold value (step S111: Yes), i is stored in the table t_l_to_h (kr) (step S112). . In other cases (step S111: No), nothing is stored.

  When the flag status is 1 and q_min (i) is less than or equal to the threshold value, that is, when the value exceeding the threshold value is less than or equal to the threshold value (step S113: Yes), i is stored in the table t_h_to_l (kf) (step S114). In other cases (step S113: No), nothing is stored.

  Then, it is determined whether q_min (i) exceeds the threshold value or less than the threshold value, and 0 or 1 is set to the flag status (step S115).

  Steps S111 to S117 are repeated until i is 1 to N (steps S116 and S117), and the numbers of i stored in the tables t_l_to_h (kr) and t_h_to_l (kf) are stored in kr_max and kf_max (steps). S118).

Next, the pulse rate calculation unit 20 calculates an average value T_ave of the period T (j) within a predetermined time of q_min (i) by the following formula (step S119).
T_ave = ((Σ _{k = 0} ^kr_max-1 | t_l_to_h (k) -t_l_to_h (k + 1) |)
+ (Σ _{k = 0} ^kf_max-1 | t_h_to_l (k) -t_h_to_l (k + 1) |)) / (kr_max + kf_max)
Next, the pulse rate is calculated (step S120). The pulse rate is obtained as the reciprocal of the period. That is, PR = 60 / T_ave, where the pulse rate is PR (times / min).
<Modification>
In the embodiment, when the binarization process of q (m, i) is performed in the kv_min calculation process (FIG. 4), all data stored as measurement data, i.e., i = 1 to i = N. A binarization process is performed for each data. In the modified example, the amount of calculation is reduced by thinning out this data and performing binarization processing using only data having an even number i, for example.

  In this case, in order not to reduce the accuracy of pulse width detection by thinning out the data, when the signal q crosses the threshold value, binarization processing is performed on the data thinned between the crossed data, and the threshold value is set. It is assumed that the time straddling is accurately acquired. It should be noted that the data thinning is performed within a range in which variation in the pulse width of the signal q can be detected and an appropriate kv_min can be calculated.

  In the embodiment, when the heart rate PR is obtained from q_min (i) in the pulse rate PR calculation process (FIG. 9), all data from i = 1 to i = N are used. Similar to the binarization process, the amount of calculation is reduced by thinning out this data and performing the binarization process using only data having an even number i, for example.

  Hereinafter, the binarization process and the pulse rate PR calculation process of the modification will be described.

<Binarization processing>
FIG. 10 is a flowchart of a binarization process according to a modification. Here, every other data is thinned out.

  The binarization process of the modification is almost the same as the binarization process of the embodiment described with reference to FIG. The difference is that 2 is first added to the index i of q (m, i) (step S214). This is because every other data is processed. In the embodiment, when q (m, i) crosses the threshold value, i is stored in the table. However, in the modified example, when q (m, i) crosses the threshold, it is checked whether the data q (m, i-1) skipped at that time crosses the threshold, and i or i-1 is stored in the table. (Step S202 to Step S205, Step S207 to Step S209, Step S213 to Step S216, Step S218 to Step S221).

  First, variables used in the binarization process are initialized (step S200). This variable is the same as in FIG.

  When the flag status_1 is 0 and q (m, i) is larger than the threshold thresh_1, that is, when the threshold thresh_1 is exceeded from a value lower than the threshold thresh_1 (step S201: Yes), the q (m, i-1) A value is calculated (step S202). That is, the skipped data is calculated.

  When q (m, i-1) is larger than the threshold thresh_1, that is, when the skipped data exceeds the threshold thresh_1 from a value lower than the threshold thresh_1 (step S203: Yes), i-1 is stored in the table t1_l_to_h ( kr_1) (step S204), and if not exceeded (step S203: No), i is stored in the table t1_l_to_h (kr_1) (step S205).

  Next, when the flag status_1 is 1 and q (m, i) is less than or equal to the threshold thresh_1, that is, when the value exceeds the threshold thresh_1 and becomes less than or equal to the threshold thresh_1 (step S206: Yes), q (m, The value of i-1) is calculated (step S207).

  When q (m, i-1) is equal to or less than the threshold thresh_1, that is, when the skipped data becomes less than the threshold thresh_1 from a value exceeding the threshold thresh_1 (step S208: Yes), i-1 is stored in the table t1_l_to_h. (kr_1) is stored (step S209), and if not (step S208: No), i is stored in the table t1_h_to_l (kf_1) (step S210).

  Then, it is determined whether q (m, i) exceeds the threshold value thresh_1 or below, and 0 or 1 is set to the flag status_1 (step S211).

  Similarly, when the flag status_2 is 0 and q (m, i) is larger than the threshold value thresh_2, that is, when the threshold value thresh_2 is exceeded from a value lower than the threshold value thresh_2 (step S212: Yes), q (m, The value of i-1) is calculated (step S213). That is, the skipped data is calculated.

  When q (m, i-1) is larger than the threshold thresh_2, that is, when the skipped data exceeds the threshold thresh_2 from a value lower than the threshold thresh_2 (step S214: Yes), i-1 is stored in the table t2_l_to_h ( kr_1) (step S215), and if not exceeded (step S214: No), i is stored in the table t2_l_to_h (kr_1) (step S216).

  Next, when the flag status_2 is 1 and q (m, i) is less than or equal to the threshold thresh_2, that is, when the value exceeds the threshold thresh_2 and becomes less than or equal to the threshold thresh_2 (step S217: Yes), q (m, The value of i-1) is calculated (step S218).

  If q (m, i-1) is smaller than the threshold value thresh_2, that is, if the skipped data falls below the threshold value thresh_2 from a value exceeding the threshold value thresh_2 (step S219: Yes), i-1 is stored in the table t2_h_to_l. (kf_1) is stored (step S220), and if not (step S219: No), i is stored in the table t2_h_to_l (kf_1) (step S221).

  Then, it is determined whether q (m, i) exceeds the threshold value thresh_2 or below, and 0 or 1 is set to the flag status_2 (step S222).

  The processing of step S201 to step S222 is repeated from i to 1 to N (steps S213 and S214), and the number of i stored in the table such as t1_l_to_h (kr_1) is stored in kr_1_max and the like (step S215).

<Pulse rate PR calculation process>
FIG. 11 is a flowchart of a pulse rate PR calculation process according to a modification.

  The average T_ave of q_min (i) state change intervals stored in step S62 in FIG. 4 is calculated to obtain the pulse rate PR.

  First, binarization processing is performed on q_min (i) (steps S220 to S234). The threshold value is set to 0 (zero), and the same process as the binarization process in FIG. 10 is performed. The threshold used in this case may be the threshold used when determining k_v.

  Specifically, variables used in the binarization process are initialized (step S220). This variable is the same as in FIG.

  When the flag status is 0 and q_min (i) is larger than the threshold value, that is, when the threshold value is exceeded from the value below the threshold value (step S221: Yes), the value of q_min (i-1) is calculated (step S222). ). That is, the skipped data is calculated.

  If q_min (i-1) is larger than the threshold, that is, if the skipped data exceeds the threshold from a value below the threshold (step S223: Yes), i-1 is stored in the table t_l_to_h (kr). (Step S224) When not exceeding (Step S223: No), i is stored in the table t_l_to_h (kr) (Step S225).

  Next, also when the flag status is 1 and q_min (i) is less than or equal to the threshold, that is, when the value exceeding the threshold is less than or equal to the threshold (step S226: Yes), the value of q_min (i-1) is also set. Calculate (step S227).

  If q_min (i-1) is equal to or smaller than the threshold value, that is, if the skipped data falls below the threshold value from the value exceeding the threshold value (step S228: Yes), i-1 is stored in the table t_h_to_l (kf). However, if not the following (step S228: No), i is stored in the table t_h_to_l (kf) (step S230).

  Then, it is determined whether q_min (i) exceeds the threshold value or below, and 0 or 1 is set to the flag status (step S231).

  Next, the average value T_ave of the period T (j) within a predetermined time of q_min (i) is calculated in the same manner as Step S119 in FIG. 9 (Step S235).

  Next, the pulse rate PR is calculated in the same manner as in step S120 in FIG. 9 (step S236).

  According to the embodiment and the modification as described above, since the signal component having the best periodicity is obtained using the recurrence formula, the arithmetic processing can be simplified as compared with the conventional technique, and the arithmetic processing amount is reduced. can do. In the modified example, since the processing is further performed on a smaller amount of data, the calculation processing amount can be further reduced. Therefore, it is possible to suppress power consumption associated with arithmetic processing. As a result, a battery can be used for the biological information measuring device 40, and a small and lightweight biological information measuring device convenient for carrying can be provided. For this reason, for example, even if a mountain climber or a home oxygen therapy patient always carries, a physical burden can be reduced.

  Moreover, since the biological information measuring apparatus 40 according to the present embodiment determines whether or not to extract a noise component based on an index having information on the ratio between the intensity of the signal component and the intensity of the noise component. When the noise component is negligibly small, the calculation for removing the noise component can be omitted, so that the calculation time for obtaining biological information is shortened, and the battery operating time is lengthened by suppressing power consumption. In addition, the biological information can be obtained more quickly.

  Moreover, since the biological information measuring device 40 according to the present embodiment outputs biological information, signal reliability, and noise reliability by voice, text, light, and the like, it is possible to easily check these. Furthermore, the biological information measuring apparatus 40 according to the present embodiment outputs a warning indicating that the index has exceeded a predetermined value, or a warning indicating a decrease in signal reliability or noise reliability. It is also possible to alert a person or the like regarding the handling of biological information. The handling of the calculated biological information can be changed according to the index, the noise reliability, and the signal reliability. For example, remeasurement or the like can be performed based on the index.

  Moreover, the biological information measuring device 40 according to the present embodiment can reduce erroneous measurement due to, for example, a double peak by using a binarized waveform when calculating the pulse rate.

  In addition, the biological information measuring method according to the present embodiment can calculate biological information in an earlier calculation time with the same calculation resource (hardware resource).

  In the above-described embodiment, as an example, the biological information measuring device 40 that detects a change in the intensity of light having two wavelengths having periodicity and obtains blood oxygen saturation is illustrated. In addition, based on changes in the intensity of light of one wavelength with periodicity, pulse, arrhythmia (atrial fibrillation / extra systole) detection, defibrillation monitoring, autonomic nerve damage, diagnosis of blood vessel age, etc. The present invention can also be applied to a biological information measuring device to be performed. Furthermore, in addition to the pulse wave waveform, the present invention can also be applied to a device that measures biological information such as electrocardiogram.

  In the binarization processing of the embodiment, the period of the peak and valley of the waveform after binarization processing is used as an index for obtaining k_v such that Equation (15) has periodicity, but is not limited thereto. It is not something. As another index in the waveform after binarization processing, only the period in which the peaks are repeated or only the period in which the valleys are repeated may be used. Also, maximum peak period-minimum peak period, maximum peak period-minimum peak period, peak peak standard deviation, peak peak standard deviation, peak peak maximum -Minimum peak width, maximum valley width-minimum valley width, standard deviation of peak width, standard deviation of valley width may be used. Further, the maximum-minimum amplitude or standard deviation of the amplitude obtained from the maximum and minimum values of the waveform of R_signal-k_v * IR_signal may be used as an index.

  In the embodiment, two threshold values are used, but three or more threshold values are provided, and three or more types of binarized pulses that are binarized depending on whether or not the signal corresponding to Expression (15) exceeds each threshold value. The above-mentioned various periodicity indexes may be calculated for each, and k_v that minimizes the average value (including simple average, harmonic average, and geometric average) may be set as the optimal value. Increasing the number of thresholds increases the amount of calculation, but has the advantage of improving the accuracy of periodicity evaluation.

  In any index, k_v that minimizes the variation of the index within a predetermined time may be obtained from the standard deviation of the index, the difference between the maximum value and the minimum value of the index, or the like. When a plurality of indices are used, k_v that minimizes each index may be different. In that case, the harmonic average value of k_v that minimizes each index is set as the optimum value. The average value may be a simple average value or a geometric average value. Alternatively, the closest k_v among the plurality of k_v that minimize the plurality of indices may be employed. Further, an average value of a plurality of indexes (simple average value, harmonic average value, geometric average value, etc.) may be evaluated as an index.

  When a plurality of k_v is obtained according to a plurality of indices, k_a that satisfies the equal sign of Expression (8) may be calculated for each k_v. The method of determining the final k_a from a plurality of k_a may be an average value (simple average value, harmonic average value, geometric average value, etc.) of the plurality of k_a as the final k_a, The one closest to the final k_a may be the final k_a this time, or the average value of the plurality of k_a excluding the maximum and minimum may be the final k_a. Furthermore, the load average may be applied by applying a weight corresponding to the absolute value of the difference from the previous k_a.

  Further, k_v is determined by the above method for 200 sets of R_signal and IR_signal, for example, and when k_a is obtained, 200 sets of data are divided into, for example, 4 to calculate 4 sets of k_a by solving 4 sets of equations (8). Then, those average values (simple average, harmonic average, geometric average value, load average value weighted by the absolute value of the difference from the previous k_a, etc.) may be obtained. You may further average the average value of k_a calculated | required with the whole 200 sets of data, and said 4 k_a.

  In addition, when multiple k_v are obtained from multiple indicators, the median of the continuous k_v range or the value closest to k_v obtained by the previous arterial blood oxygen saturation calculation is optimal. It may be a value.

In the above description, when the noise component index is equal to or smaller than a predetermined value, k_a is calculated by a conventional method such as Equation (14), but the previous k_v (= k_v_0) is substituted into Equation (8). Then k_a may be obtained. When noise is low, the left side of equation (8) is Σ {IR_signal * k_v-R_signal} * {IR_signal * k_a-R_signal}
≒ (k_v-k_a) * ΣIR_signal * {IR_signal * k_a-R_signal}
≒ (k_v-k_a) * {ΣIR_signal ² * k_a-ΣIR_signal * R_signal}
So, if a value different from k_a is substituted for k_v, the equation is ΣIR_signal ² * k_a-ΣIR_signal * R_signal = 0
Is equivalent to This solution is
k_a ＝ ΣIR_signal * R_signal / ΣIR_signal ²
Is obtained. When noise is small, R_signal ≒ true k_a * IR_signal, so the right side of the above equation is true k_a.

  Therefore, a value sufficiently close to true k_a can be obtained by substituting a value that seems to be different from k_a, such as the previous k_v, into Equation (8).

  In addition, when noise is small, R_signal − k_v * IR_signal has a strong periodicity over a wide range of k_v, but even in that case, the value obtained by multiplying the above various indicators by the absolute value of the difference between the previous k_v and k_v Since the optimal value of k_v is almost the same as the previous k_v, a value sufficiently close to true k_a can be obtained even if the same processing as when the noise index is large is performed. When importance is attached to the simplification of the program rather than the calculation speed, the processing may not be divided depending on the size of the noise index.

  As an example of using periodicity, an example is given in which k_v is obtained so that the left side of Equation (15) satisfies the periodicity, but is not limited to Equation (15), and k_v is expressed as a signal component. If it is a mathematical formula to be used, it may be used instead of the formula (15).

  In order to express the present invention, the present invention has been properly and fully described through the embodiments with reference to the drawings. However, those skilled in the art can easily change and / or improve the above-described embodiments. It should be recognized that this is possible. Therefore, unless the modifications or improvements implemented by those skilled in the art are at a level that departs from the scope of the claims recited in the claims, the modifications or improvements are not covered by the claims. To be construed as inclusive.

DESCRIPTION OF SYMBOLS 1 Data acquisition part 2 Sensor (R) part 3 Sensor (IR) part 10 Measurement part 11 AC / DC (R) conversion part 12 AC / DC (IR) conversion part 13 BPF (R) part 14 BPF (IR) part 15 ΣR_signal * IR_signal calculation unit 16 Σ (R_signal) ² calculation unit 17 Σ (IR_signal) ² calculation unit 18 initial value calculation unit 19 SvO2 estimation SpO2 determination unit 20 pulse rate calculation unit 21 reliability calculation unit 22 difference calculation unit 30 output unit 31 SpO2 output unit 32 Pulse rate output unit 33 Reliability output unit 40 Biological information measuring device

Claims (7)

A first time-series signal including a first signal component having periodicity and a first noise component, a second signal component having a predetermined relationship with the first signal component, and a predetermined relationship with the first noise component And the second time-series signal including the second noise component and the predetermined relationship between the first noise component and the second noise component. Signal generation for generating a plurality of signals including the first signal component from a time-series signal, wherein the predetermined relationship between the first noise component and the second noise component is different by a predetermined difference. And
The predetermined relationship between the first noise component and the second noise component in the first predetermined time range using the periodicity of each signal generated by the signal generation unit in the first predetermined time range. And an estimation unit for estimating
The signal generation unit generates one reference signal, and uses a recurrence formula using the signal and data calculated from the first time-series signal and the predetermined difference, to calculate the plurality of signals. Among them, another signal which is a signal other than the one signal serving as the reference is generated , and a threshold value used for binarization processing of the signal serving as the reference is calculated, and the threshold value and the first time are calculated. Using the series signal or the data calculated using the second time series signal and the predetermined difference, a threshold for use in binarization processing of each of the other signals is calculated by a recurrence formula. Then, a binarized signal is generated by performing binarization processing on the generated signal using the threshold value,
The estimation unit estimates the predetermined relationship by estimating the period of the signal generated by the signal generation unit using the periodicity of the binarized signal . In claim 1 ,
The signal generated by the signal generation unit is a time-series signal,
The signal generation unit performs binarization using data obtained by thinning out a predetermined number from the generated signal. In claim 2 ,
The signal processing apparatus characterized in that the binarization processing is performed on data thinned between two pieces of the thinned data sandwiching the threshold value. A signal processing method comprising:
A first time-series signal including a first signal component having periodicity and a first noise component in a first predetermined time range; a second signal component having a predetermined relationship with the first signal component; and Based on the second time-series signal including the second noise component having a predetermined relationship with the first noise component, and the predetermined relationship between the first noise component and the second noise component, the first A signal including the first signal component from a time-series signal and the second time-series signal, wherein the predetermined relationship between the first noise component and the second noise component is varied by a predetermined difference. A signal generation step for generating a plurality of signals;
Using the periodicity of each signal generated in the generation step in the first predetermined time range, the predetermined relationship between the first noise component and the second noise component in the first predetermined time range is obtained. An estimation step for estimating,
The signal generation step generates one signal as a reference, and uses a recurrence formula using the signal and data calculated from the first time-series signal and the predetermined difference . Among them, another signal which is a signal other than the one signal serving as the reference is generated , and a threshold value used for binarization processing of the signal serving as the reference is calculated, and the threshold value and the first time are calculated. Using the series signal or the data calculated using the second time series signal and the predetermined difference, a threshold for use in binarization processing of each of the other signals is calculated by a recurrence formula. Then, a binarized signal is generated by performing binarization processing on the generated signal using the threshold value,
The signal processing method characterized in that the estimation step estimates the predetermined relationship by estimating the period of the signal generated in the signal generation step using the periodicity of the binarized signal . Based on at least the first measurement data or the second measurement data obtained by irradiating the living body with a plurality of lights having different wavelengths and receiving the light transmitted or reflected through the living body, respectively, the living body of the living body A biological information measuring device for measuring information,
A first measurement unit that measures first measurement data including a first signal component having periodicity and a first noise component;
A second measurement unit for measuring second measurement data including a second signal component having a predetermined relationship with the first signal component and a second noise component having a predetermined relationship with the first noise component;
Based on the first measurement data, the second measurement data, and the predetermined relationship between the first noise component and the second noise component, from the first measurement data and the second measurement data, the A signal generation unit that generates a plurality of signals that include a first signal component and that differ in the predetermined relationship between the first noise component and the second noise component by a predetermined difference;
An estimation unit that estimates the predetermined relationship between the first noise component and the second noise component using the periodicity of the signal generated by the signal generation unit;
The signal generation unit generates one signal serving as a reference, and uses a recurrence formula using the signal and data calculated from the first measurement data and the predetermined difference. Generate another signal which is a signal other than the one signal serving as the reference, calculate a threshold for use in the binarization process of the signal serving as the reference, and the threshold and the first time series Using the signal or the data calculated using the second time-series signal and the predetermined difference, each threshold value to be used for the binarization process of each of the other signals is calculated by a recurrence formula, Then, a binarized signal is generated by performing binarization processing on the generated signal using the threshold value,
The biological information measuring device , wherein the estimation unit estimates the period of the signal generated by the signal generation unit using the periodicity of the binarized signal and estimates the predetermined relationship . In claim 5 ,
A blood oxygen saturation calculating unit that calculates blood oxygen saturation at a predetermined cycle based on the predetermined relationship estimated by the estimating unit,
It said signal producing formation unit, said at a cycle longer than a predetermined period, and generates a plurality of signals to produce a binary signal,
The estimation unit estimates the predetermined relationship between the first noise component and the second noise component based on the periodicity of the binarized signal. In claim 5 or claim 6 ,
A pulse rate calculation unit that generates a binarized signal by performing binarization processing on a signal having a predetermined relationship estimated by the estimation unit, and calculates a pulse rate using the periodicity of the binarized signal. A biological information measuring device comprising:

Images