JP5562805B2 - Pulse rate measuring method and blood oxygen saturation measuring method - Google Patents

Description

  The present invention relates to a digital processing method of a light reception signal in a pulse meter and a pulse oximeter for measuring blood oxygen saturation using LED transmitted light or reflected light from a vascular tissue.

As a conventional pulse meter, in the light reception signal waveform for LED transmitted light or reflected light from vascular tissue, the peak and bottom timing in each cycle is detected, and the pulse rate is estimated from the time width between each peak or bottom. There is a way.
Further, in the conventional pulse oximeter, the blood oxygen saturation is obtained from the ratio of both by reading the peak value and the bottom value for the signal waveforms of red light and infrared light.

  Both are based on the premise that there is no external noise such as body movement, and even if a slight amount of noise is mixed, the periodicity of the signal waveform will be disturbed, and the peak and bottom of the waveform that should be synchronized with the pulse will appear randomly. There is also a large error in blood oxygen saturation that is not estimated and calculated based on it.

  In addition to these methods, there is a conventional method using a harmonic spectrum structure in which a pulse component forms a sharp peak at a position that is an integral multiple of the fundamental frequency in the frequency spectrum distribution. In this method, blood oxygen saturation is calculated from the ratio of each peak value to red light and infrared light using the fact that the position of the harmonic peak represents the pulse frequency.

  However, with this method, when the frequency of the noise component is significantly different from the frequency of the pulse component, it shows good resistance to noise, but when both are close or coincide with each other, it is difficult to avoid the influence of noise.

Japanese Patent No. 2816944 JP 2008-220722 A JP 2002-17694 A JP 2009-22484 A

In response to the problem that it is difficult to avoid the influence of noise when the frequency of the noise component and the frequency of the pulse component are close or coincident with each other, body motion is detected using an acceleration sensor, etc., and the frequency spectrum distribution and the pulse signal Focusing on the difference from the overtone spectrum structure, a method for reducing the influence of noise due to body movement has been proposed (Patent Document 1).
However, this method requires a separate acceleration sensor, leading to an increase in the cost of the device, and the signal detected from the acceleration sensor does not always accurately represent the body movement. There are problems such as limited effect of suppression.

On the other hand, a method of removing the influence of body movement by performing an analog filter process on the received light signal without using another sensor has been proposed (Patent Document 2).
In this method, the blinking frequency of the LED emission is set sufficiently higher than the pulse frequency, and a filter that extracts a frequency component in the vicinity of the blinking frequency and a filter that extracts a lower frequency component are used. The pulse information is obtained by obtaining the component and removing it from the former.

  However, this method is effective in removing noise caused by the invasion of extraneous light due to body movement, but transmitted light that has passed through the vascular tissue due to deviations in the irradiation position of LED light or disturbance in blood flow. Alternatively, it is difficult to separate noise components mixed in reflected light.

There has also been proposed a method for removing the influence of body movement by processing the frequency spectrum of the received light signal.
For example, when there are a plurality of spectrum peaks of a certain level or more, it is determined that there is an influence of body movement, and a method for selecting a frequency spectrum and peak value close to the pulse spectrum peak estimated at the previous time (Patent Document 3) ) And a method of distinguishing from the frequency spectrum peak of body motion using the fact that the frequency spectrum peaks of the pulse signal are in a harmonic relationship with each other (Patent Document 4) has been proposed.
However, any of these methods has a problem that once the influence of body movement becomes large and the spectrum peak of the pulse cannot be traced, it is difficult to return to the correct pulse peak and stable pulse estimation becomes difficult.

  Therefore, the present invention has been made in view of the above problems, and has a robustness to body motion and changes in pulse with respect to measurement of pulse rate and measurement of blood oxygen saturation using LED transmitted / reflected light. It is an object of the present invention to provide a digital signal processing system that can be followed.

In order to solve the above problems, the pulse rate measuring method of the present invention measures the time width between the peaks or the bottoms of the waveform of the received light signal by the light receiving element that receives the light transmitted or reflected through the tissue of the living body, and responds to it. A time width measurement process for generating a harmonic spectrum distribution to be performed, a spectrum distribution calculation process for calculating a spectrum distribution by frequency analysis of a received light signal, a harmonic spectrum distribution at the current time, a spectrum distribution at the current time, and a previous time A spectrum distribution update step of updating the spectrum distribution by taking a weighted average of the obtained harmonic overtone spectrum distribution and the spectrum distribution, and sequentially executing the time width measurement step, the spectrum distribution calculation step, and the spectrum distribution update step After obtaining the spectral distribution of the pulse wave at each time, It is characterized by calculating the pulse rate from the wave number.

  In the pulse rate measurement method of the present invention, when it is difficult to measure the time width in the time width measurement process due to body movements, the current time spectrum distribution and the estimated value of the previous time calculated in the spectrum distribution calculation process It is preferable to calculate the pulse rate using.

  In the pulse rate measurement method of the present invention, it is preferable that a peak whose frequency is not related to harmonics is determined to be a body motion component and excluded from the tracking target in peak selection from the spectrum distribution of the pulse wave.

  In the pulse rate measuring method of the present invention, in peak selection from the spectrum distribution of the pulse group, the peak located near the frequency corresponding to the pulse rate estimated at the previous time has a value compared to other peaks. Even if it is small, it is preferable to select it when it is larger than a predetermined level.

  The blood oxygen saturation measurement method of the present invention is obtained by frequency analysis of a light reception signal by a light receiving element that receives red light and infrared light with reference to the pulse rate calculated by any one of the above-described pulse rate measurement methods. The spectrum peak closest to the pulse rate frequency in the obtained spectrum distribution is selected, the first blood oxygen saturation is obtained from the ratio of the spectrum peak values for red light and infrared light, and the peak corresponding to the spectrum peak value is obtained. The second blood oxygen saturation is estimated from the ratio of the spectral peak values of the red light and the infrared light with respect to the peak having a harmonic overtone with respect to the frequency, and the first blood oxygen saturation and the second blood oxygen saturation are estimated. The final blood oxygen saturation is calculated from the average of the above.

  In the method for measuring blood oxygen saturation according to the present invention, when the second blood oxygen saturation takes a value lower than a predetermined level, the peak related to overtone is affected by body movement. It is preferable to exclude it from the calculation of the average.

The nth blood oxygen saturation is estimated from the ratio of the spectral peak values of the red light and the infrared light for the peak having an n overtone relationship with the peak frequency corresponding to the spectral peak value, and the first to nth. The final blood oxygen saturation may be calculated by averaging up to blood oxygen saturation or by selection.

  In the blood oxygen saturation measurement method of the present invention, when the nth blood oxygen saturation takes a value lower than a predetermined level, the peak related to the harmonic overtone is affected by body movement. Judge and exclude from average calculation.

  According to the pulse rate measuring method and the blood oxygen saturation measuring method of the present invention, it is robust against body movement and can follow changes in the pulse, thereby improving the measurement accuracy of the pulse meter and pulse oximeter. be able to.

It is a block diagram which shows the flow of the pulse measurement which concerns on embodiment of this invention. It is a flowchart which shows the flow of the pulse measurement which concerns on embodiment of this invention. It is a graph which shows the example of the output waveform of the downsampling shown in FIG. It is a graph which shows the example of the spectrum distribution shown in FIG. It is a graph which shows the example of the overtone spectrum distribution shown in FIG. It is a graph which shows the example of the spectrum distribution produced | generated by addition of the spectrum distribution shown in FIG. It is a block diagram which shows the flow of the blood oxygen saturation measurement which concerns on embodiment of this invention. It is a flowchart which shows the flow of the blood oxygen saturation measurement which concerns on embodiment of this invention. It is a graph which shows the example of the spectrum distribution shown in FIG. 8 which concerns on embodiment of this invention.

Hereinafter, a pulse rate measuring method and a blood oxygen saturation measuring method according to an embodiment of the present invention will be described in detail with reference to the drawings.
The present invention provides a signal that enables stable and highly accurate measurement even under body movement in measurement of pulse and blood oxygen saturation using transmitted light and / or reflected light from vascular tissue of a living body irradiated with an LED. A processing method is provided.

  In the embodiments described below, a waveform peak / bottom detection method for detecting a peak / bottom of a pulse wave signal obtained from the received light signal of the transmitted light and / or reflected light and a short-time frequency analysis of the received light signal And a spectrum analysis method for calculating a wave spectrum.

Furthermore, pulse estimation is performed in each of the waveform peak / bottom detection method and spectrum analysis method described above, and the results (estimation results at the current time) and the estimation results up to the previous time are integrated, so The final pulse value corresponding to is obtained.
The specific processing flow is as follows.

  First, pulse estimation will be described with reference to FIGS. FIG. 1 is a block diagram showing the flow of pulse measurement. FIG. 2 is a flowchart showing the flow of pulse measurement. FIG. 3 is a graph showing an example of an output waveform of the downsampling (step S100) shown in FIG. FIG. 4 is a graph showing an example of the spectrum distribution (step S103) shown in FIG. FIG. 5 is a graph showing an example of the overtone spectrum distribution (step S106) shown in FIG. FIG. 6 is a graph showing an example of the spectrum distribution generated by the addition of the spectrum distribution shown in FIG. 2 (step S108).

The flow up to the downsampling output will be described with reference to FIG.
A pulse signal from the pulse generator 101 is amplified by an amplifier 102, and an LED (Light Emitting Diode) 103 is driven by the pulse signal. The LED 103 emits light toward the measurement object S side at a constant frequency (for example, 200 Hz). The light receiving element 104 receives light that has passed through or reflected through the in-vivo vascular tissue S such as a finger or earlobe. Since the blood vessel cross-sectional area changes with the pulse and the light attenuation changes accordingly, the light received by the light receiving element 104 has an amplitude corresponding to the pulse wave in the vascular tissue S in the living body.

  The light receiving element 104 outputs a light receiving signal corresponding to the received light. This received light signal is amplified by the amplifier 105. The amplified and synchronized received light signal is converted into a digital signal by the A / D converter 106. This conversion is synchronously controlled with an output signal from the pulse generator 101 to the LED 103 so that the circuit operates only during light emission for power saving.

An unnecessary DC (direct current) component is cut from the digital signal by the DC cut circuit 107 by the A / D converter 106. Further, this signal has a frequency component (for example, 50 Hz or more) unnecessary for pulse estimation cut by an LPF (Low Pass Filter) 108, and a down-sampled signal 109 is output. The frequency of the downsampling output signal 109 (step S100 in FIGS. 3 and 2) is, for example, 25 Hz.

The signal converted into a digital signal by the A / D converter 106 has high frequency noise removed by the LFP 120 for peak / bottom detection. The LFP 120 performs, for example, an 8-sample moving average.

The downsampling output signal 109 is processed by the pulse detector 110. This processing step will be described with reference to FIGS.
The window function calculation circuit 111 multiplies the down-sampled signal 109 (step S100) by the window function (step S101). For this calculation result, the FFT circuit 112 cuts out a time width of, for example, 5 seconds, and then executes FFT (fast Fourier transform) (step S102). Subsequently, the spectrum distribution calculation circuit 113 calculates a frequency spectrum distribution (spectrum distribution 1) from the fast Fourier transformed signal (step S103, spectrum distribution calculation step). An example of the frequency spectrum distribution calculated by such short-time frequency analysis is shown in FIG. The spectrum distribution calculation circuit 113 calculates a pulse estimated value based on the calculated spectrum distribution.

  On the other hand, the time width measurement circuit 114 detects the timing of the peak and the bottom in each cycle of the waveform of the signal from which the high frequency noise has been cut by the LFP 120, and calculates the time width between the peaks or between the bottoms (step S104, Time width measurement process). The spectrum distribution calculation circuit 115 determines whether or not the pulse can be estimated from the calculated time width (step S105). When it is determined that the pulse can be estimated (YES in step S105), the spectrum distribution calculation circuit 115 calculates a pulse estimated value and a pulse frequency based on the time width, and a peak appears at the harmonic frequency position of the pulse frequency. A harmonic spectrum distribution (synthetic spectrum distribution) is generated (step S106). An example of the generated overtone spectrum distribution is shown in FIG.

  When it is determined that the pulse cannot be estimated (NO in step S105), the spectrum distribution calculation circuit 115 does not generate a spectrum distribution (step S107).

  Next, the adding circuit 116 adds the original spectral distribution calculated in step S103, the harmonic overtone distribution generated in step S106, and the spectral distribution of these two spectral distributions at the previous time (step S109) at a constant rate. Then, a spectrum distribution at the current time is generated and updated, and the pulse rate is calculated from the spectrum distribution (step S108, pulse rate calculating step, spectrum distribution updating step). The spectrum distribution at the previous time used for the addition is the spectrum distribution 1 calculated by the spectrum distribution calculation circuit 113 and the harmonic overtone distribution generated by the spectrum distribution calculation circuit 115, which is stored in time series by the delay circuit 117, and is more than the current time. The spectrum distribution at the previous time is output to the adder circuit 116. Further, the addition at a certain ratio is performed by, for example, performing weighted averaging on the spectrum distribution 1, the harmonic overtone spectrum distribution, and the spectrum distribution at the previous time. An example of the generated corrected spectral distribution is shown in FIG.

  Subsequently, the peak detection circuit 118 selects a peak having a frequency closest to the estimated pulse value at the previous time in the corrected spectrum distribution (step S110). At this time, the peak detection circuit 118 has a peak value of a reference abnormality in the vicinity of the selected peak and whether there is a peak (overtone peak) having a value that is an integer multiple of twice or more of the peak frequency. It is determined whether or not (step S111). If such a harmonic peak exists (YES in step S111), the harmonic peak is selected (step S112), and the peak frequency is set as the final pulse value at the current time (step S113).

  Here, in the peak selection from the pulse wave spectrum distribution in step S110, the peak located near the frequency corresponding to the pulse rate estimated at the previous time is smaller than the other peaks, If it is greater than the set level, select it.

On the other hand, if there is no overtone peak (NO in step S111), it is determined that the selected peak is a body motion component and is excluded from the tracking target, and the final pulse at the current time with the peak frequency selected in step S110. A value is set (step S113).
Therefore, in step S113, the pulse rate is calculated from the maximum peak frequency of the spectrum distribution of the pulse wave.
By repeatedly calculating the above processing at regular time intervals (for example, every 0.5 seconds), a time-series pulse estimated value and a pulse rate can be obtained as time-series data.

  Next, estimation of blood oxygen saturation will be described with reference to FIGS. FIG. 7 is a block diagram showing a flow of blood oxygen saturation measurement. FIG. 8 is a flowchart showing the flow of blood oxygen saturation measurement. FIG. 9 is a graph showing an example of the spectrum distribution (step S203) shown in FIG.

The flow up to the downsampling output will be described with reference to FIG.
From the pulse generator 101, pulse signals for infrared light and red are output and amplified by the amplifiers 202a and 202b, respectively. The amplified pulse signal drives the LED 203a for infrared light emission and the LED 203b for red light emission. These LEDs 203a and 203b emit light toward the measuring object S at a constant frequency (for example, 200 Hz). The light receiving element 204a for infrared light and the light receiving element 204b for red light respectively receive infrared light and red light that have passed or reflected through the in-vivo vascular tissue S such as a finger or earlobe.

  The light receiving elements 204a and 204b each output a light reception signal corresponding to the received light. These received light signals are amplified by amplifiers 205a and 205b, respectively. The amplified and synchronized received light signals are converted into digital signals by the A / D converter 106, respectively. This conversion is synchronously controlled with the output signal from the pulse generator 101 to the LEDs 203a and 203b so that the circuit operates only during light emission for power saving.

The signal converted into a digital signal by the A / D converter 106 is separated into an AC (alternating current) component and a DC (direct current) component by an AC / DC separator 207. The AC / DC calculation circuit 208 uses the ratio of the distributed AC component and DC component for each of the received light signals of infrared light and red light as a signal for measuring blood oxygen saturation.

The signal converted into a digital signal by the A / D converter 106 has high frequency noise removed by the LFP 120 for peak / bottom detection. The LFP 120 performs, for example, an 8-sample moving average.

  Further, for each of the red light signal and the infrared light signal, an unnecessary frequency component is cut through the LPF 108 and a down-sampled signal 210 is output. The frequency of the downsampling output signal 210 is, for example, 50 Hz.

Processing steps after the downsampling output will be described with reference to FIGS.
The window function calculation circuit 111 multiplies the downsampled signal 210 (step S200) by the window function (step S201). For this calculation result, the FFT circuit 112 cuts out to a time width of, for example, 5 seconds, and then performs FFT (fast Fourier transform) (step S202). Subsequently, the spectrum distribution calculation circuit 113 calculates a frequency spectrum distribution from the fast Fourier transformed signal (step S203). An example of the calculated frequency spectrum distribution is shown in FIG. In FIG. 9, the IR spectrum distribution is a spectrum distribution of infrared light, and the R spectrum distribution is a spectrum distribution of red light.

  On the other hand, in the pulse detection unit 110, similarly to the above-described estimation of the pulse, the time width measurement circuit 114 detects the timing of the peak and the bottom in each cycle of the waveform of the signal 210 from which the high frequency noise has been cut, and between the peaks. Alternatively, the time width between the bottoms is calculated (step S104 in FIG. 2). The spectrum distribution calculation circuit 115 determines whether the pulse can be estimated from the calculated time width (step S105 in FIG. 2). When it is determined that the pulse can be estimated (YES in step S105 in FIG. 2), the spectrum distribution calculation circuit 115 calculates a pulse estimated value and a pulse frequency based on the calculated time width (step S205 in FIG. 8). .

  Based on the pulse frequency estimated in this way, the IR overtone peak selection circuit 214a and the R overtone peak selection circuit 214b respectively select the spectrum peak closest to the frequency for infrared light and red light (step S204). ). The R / IR ratio calculation circuit 215 calculates the ratio of the peak values of both the infrared light and red light selected in step S204, and obtains the blood oxygen saturation (SpO2) from this ratio (first blood oxygen Saturation, step S206). Here, “SpO2” is an abbreviation of “percutaneous oxygen saturation” or “oxygen saturation by pulse oximetry”.

  The harmonic overtone selection / averaging circuit 216 determines the blood oxygen saturation (SpO2) based on the ratio between the peaks of harmonics with respect to the peak frequencies of the infrared light and red light selected in step S204, as in step S206. (Second blood oxygen saturation).

The overtone selection / averaging circuit 216 determines whether or not there is a peak close to the blood oxygen saturation at the previous time for the first blood oxygen saturation and the second blood oxygen saturation (step S207).
If there is a peak close to the blood oxygen saturation at the previous time (YES in step S207), the harmonic overtone selection / averaging circuit 216 selects only one or two peaks (step S208).
If there is no peak close to the blood oxygen saturation level at the previous time (NO in step S207), the harmonic selection / averaging circuit 216 determines that the peak related to harmonics is affected by body movement, and blood Peaks with high intermediate oxygen saturation are selected for infrared light and red light, respectively (step S211).

Similarly, the harmonic overtone selection / averaging circuit 216 also applies the blood oxygen saturation (from the ratio of both in step S206 to the peak in the nth harmonic relationship with respect to the peak frequency of the infrared light and red light selected in step S204. SpO2) may be determined (nth blood oxygen saturation). In this case, the harmonic overtone selection / averaging circuit 216 selects 1 to n peaks (step S208). Here, n is an integer greater than 1.

The overtone selection / averaging circuit 216 performs weighted averaging on the peak selected in step S208 or S211 (step S209). In this weighted average, the blood oxygen saturation at the previous time is referred to (step S210).
The overtone selection / averaging circuit 216 determines the final blood oxygen saturation from the calculated average. However, if the blood oxygen saturation estimated from each overtone peak takes a value lower than the specified level, it is determined that the overtone peak is affected by body movement, and the calculation of averaging is performed. Exclude from Since the calculation up to the spectrum distribution is almost the same as the above-described pulse estimation, for example, the red light may be shared for the pulse estimation and the blood oxygen saturation estimation.

The pulse rate measurement method and blood oxygen saturation measurement method of the present embodiment described above are highly sensitive to the pulse rate and blood oxygen saturation, which are not easily affected by body movements and are capable of dealing with sudden pulse changes. Measurement is possible. This makes it possible to perform measurement during exercise, which was difficult in the past.
Further, since the measurement is realized only by the digital signal processing, the above-described high performance can be achieved only by replacing the software of the existing apparatus, and the cost is low. Furthermore, since the algorithm is simple, it can be implemented by a fixed-point CPU.

  Although the present invention has been described with reference to the above embodiment, the present invention is not limited to the above embodiment, and can be improved or changed within the scope of the purpose of the improvement or the idea of the present invention.

  As described above, the pulse rate measuring method and the blood oxygen saturation measuring method according to the present invention are useful for a pulse oximeter that measures the pulse rate and blood oxygen saturation transcutaneously.

101 Pulse generator 102, 105 Amplifier 103 LED
104 light receiving element 106 A / D converter 107 DC cut circuit 108 LPF
DESCRIPTION OF SYMBOLS 110 Pulse detection part 111 Window function arithmetic circuit 112 FFT circuit 113, 115 Spectral distribution calculation circuit 114 Time width measurement circuit 116 Adder circuit 117 Delay circuit 118 Peak detection circuit 202a, 202b Amplifier 203a, 203b LED
204a, 204b Light receiving element 205a, 205b Amplifier 207 AC / DC separator 208 AC / DC calculation circuit 214a IR overtone peak selection circuit 214b R overtone peak selection circuit 215 R / IR ratio calculation circuit 216 Overtone selection / averaging circuit

Claims (8)

A time width measuring step for measuring a time width between peaks or bottoms of a waveform of a light reception signal by a light receiving element that receives light transmitted or reflected through a living tissue, and generating a harmonic spectrum distribution corresponding to the time width; and
A spectral distribution calculation step of calculating a spectral distribution by frequency analysis of the received light signal;
A spectrum distribution updating step of updating the spectrum distribution by taking a weighted average of the harmonic spectrum distribution at the current time, the spectrum distribution at the current time, and the harmonic spectrum distribution obtained up to the previous time and the spectrum distribution; ,
Have
The pulse width is calculated from the maximum peak frequency after obtaining the pulse wave spectrum distribution at each time by sequentially executing the time width measuring step, the spectrum distribution calculating step, and the spectrum distribution updating step. A pulse rate measuring method characterized by the above. If it is difficult to measure the time width in the time width measurement step due to body movement of the living body, the pulse rate is calculated using the current time spectrum distribution and the estimated value of the previous time calculated in the spectrum distribution calculation step. The pulse rate measuring method according to claim 1 , wherein the pulse rate is calculated. 2. The pulse rate measuring method according to claim 1 , wherein in selecting a peak from the spectrum distribution of the pulse wave, a peak whose frequency is not related to a harmonic is determined to be a body motion component and excluded from a tracking target. In the peak selection from the spectrum distribution of the pulse group, the peak located near the frequency corresponding to the pulse rate estimated at the previous time is less than the predetermined level even if the value is smaller than the other peaks. 2. The pulse rate measuring method according to claim 1 , wherein when it is large, it is selected. Obtained by frequency analysis of the received light signal by the light receiving element that receives red light and infrared light with reference to the pulse rate calculated by the pulse rate measuring method according to any one of claims 1 to 4. Selecting the spectral peak closest to the frequency of the pulse rate in the given spectral distribution;
Obtain the first blood oxygen saturation from the ratio of the spectral peak values for red light and infrared light,
Estimating the second blood oxygen saturation from the ratio of the spectral peak value of red light and infrared light with respect to the peak in harmonic relation with the peak frequency corresponding to the spectral peak value,
A method for measuring blood oxygen saturation, comprising calculating a final blood oxygen saturation from an average of the first blood oxygen saturation and the second blood oxygen saturation. If the second blood oxygen saturation level is lower than a predetermined level, it is determined that the harmonic-related peak is affected by body movement and is excluded from the average calculation. The method for measuring blood oxygen saturation according to claim 5 . For each peak that has an n overtone relationship with the peak frequency, the nth blood oxygen saturation is estimated from the ratio of the spectral peak values of red light and infrared light,
6. The blood oxygen saturation measuring method according to claim 5 , wherein a final blood oxygen saturation is calculated from an average of the first to nth blood oxygen saturations. If the n-th blood oxygen saturation level is lower than a predetermined level, it is determined that the overtone-related peak is affected by body movement and is excluded from the average calculation. The method for measuring blood oxygen saturation according to claim 7 .