KR101361578B1 - Apparatus for PPG signal processing - Google Patents

Abstract

The present invention relates to a heartbeat signal processing apparatus that removes signal noise of a heartbeat signal detected by a PPG sensor, corrects a detection error caused by motion noise, and detects a more accurate blood flow signal to provide accurate heartbeat information and stress information. PPG measuring unit for detecting a heartbeat signal; A pre-processing unit for removing signal noise in a high frequency band mixed with the heartbeat signal detected by the PPG measuring unit using a low pass filter; A PPI detector detecting a PPI signal from the heartbeat signal from which the signal noise is removed from the preprocessor; A post processor which corrects a stress detection error of the PPI signal detected by the PPI detector; And the post-processing unit processes the PPI signal in which the detection error is corrected, and determines whether or not the stress is detected. If the PPI signal is larger than the reference value, it is determined to be in a relaxed state. Contains wealth.

Description

Heart rate signal processing device {Apparatus for PPG signal processing}

The present invention relates to a heartbeat signal processing apparatus, and more particularly, to a heartbeat signal processing apparatus that detects an accurate blood flow signal and provides accurate heartbeat information and stress information.

As the development of clinical medical measurement equipment and the performance of sensors capable of measuring bio signals in various body parts have been improved, various and accurate bio signals can be measured. With the development of these technologies, more and more researches are being conducted in the medical field to detect and treat diseases, and to recognize and apply human response or sensitivity to specific stimuli. In particular, in the case of autism or depression patients, a lot of studies are being conducted to determine the patient's condition by measuring and analyzing various bio signals instead of consulting a patient to determine the patient's condition. Similarly, research is being actively conducted on the development of algorithms that can quantitatively evaluate how stressed or how people are recovering from a stress by using biosensors.

Research on specific stimuli that induce or relieve human stress generally involves analyzing the body's response through stress-stimulating and relaxing stimuli. Of these, the most representative and easy-to-measure biosignal response that can assess the stress level on external stimuli is the change in heart rate.

The cardiovascular center of the brain regulates the heart rate through sympathetic, parasympathetic, and hormones as well as changes in physical activity and stress stimulation. In a stress state, the sympathetic nerve is excited and epinephrine is secreted to increase the rate of depolarization, thereby increasing the heart rate. On the contrary, in the relaxed state, parasympathetic nerve is activated to reduce the heart rate by inhibiting depolarization. Therefore, a technology for detecting a person's stress by detecting a heart rate that changes according to a stress stimulus through an optical blood flow meter (Photo-PlethysmoGraphy: hereinafter referred to as PPG) and analyzing a heartbeat signal using a signal processing algorithm is used. have.

In general, PPG detects blood flow signals related to heart rhythm by using a predetermined number of LEDs and photodetectors. A simple sensor module contacts a part of the body (for example, a finger and an ear) to provide blood flow information through a contact point. To extract. Since the PPG can be easily performed by an examiner compared to an electrocardiogram detection method using an electrodiogram or ECG, which requires attaching two or more electrodes, an inspection device for identifying an examiner's condition for medical or non-medical purposes. It is widely used in.

That is, the heart rate (hereinafter referred to as HR) and heart rate variability (hereinafter referred to as HRV) are calculated using blood flow signals obtained through PPG, and the mentality of the inspector obtained by analyzing HR and HRV. And / or a stress test device that measures the degree of stress with information on the physical state.

On the other hand, the PPG has a disadvantage of causing a large amplitude of dynamic noise even with a slight movement, which is a noise signal generated by the shaking and small movement of the examiner during the detection of blood flow signals, and accurately detects HR and HRV. There was a problem that became a big obstacle to the.

In order to solve this problem, US Patent No. 5,662,106 (name of the invention: OXIMETER WITH MOTION DETECTION FOR ALARM MODIFICATION) proposed previously by setting a predetermined threshold value obtained through the noise-induced signal waveform If the noise is confirmed, the examiner is alert.

That is, as shown in FIG. 1, since the ratio of the positive signal value to the negative signal value (A / B) is greater than 1 to 1.4 in the DERIVATIVE SIGNAL induced dynamic noise, the threshold is 1 to 1.4. The value less than 1 ~ 1.4 is set as the value to judge the noise.

In the prior art, if the noise signal is determined while the timer is operated for a predetermined time to detect the blood flow signal through the body part of the examiner, the total measurement time is increased because the timer must be reset and the blood flow signal must be detected again. There was a problem causing a burden on the blood flow detection.

In addition, since the inspector may cause a stopping or shaking operation in the process of alerting the examiner to the noise, there is a problem that the noise is not efficient for more accurate blood flow signal detection.

U.S. Patent No. 5,662,106 (1997.09.02)

The present invention is to overcome the problems of the prior art described above, by removing the signal noise of the heartbeat signal detected by the PPG sensor, correcting the detection error due to the operation noise to detect a more accurate blood flow signal and accurate heart rate information and It is an object of the present invention to provide a heartbeat signal processing apparatus for providing stress information.

The present invention to achieve the above object, PPG measuring unit for detecting a heartbeat signal; A pre-processing unit for removing signal noise in a high frequency band mixed with the heartbeat signal detected by the PPG measuring unit using a low pass filter; A PPI detector detecting a PPI signal from the heartbeat signal from which the signal noise is removed from the preprocessor; A post processor which corrects a stress detection error of the PPI signal detected by the PPI detector; And the post-processing unit processes the PPI signal in which the detection error is corrected, and determines whether or not the stress is detected. If the PPI signal is larger than the reference value, it is determined to be in a relaxed state. It provides a heart rate signal processing apparatus comprising a portion.

The PPG measuring unit may measure a PPG, which is a change in blood flow through a blood vessel, using a PPG sensor having an optical characteristic, and convert it into a discrete electrical signal to detect a heartbeat signal.

The low pass filter is preferably a hamming filter.

The preprocessor includes a parameter determiner which determines a parameter of the low pass filter by applying an energy density analysis method using a spectrogram to determine a parameter adapted to a different heartbeat signal for each individual.

The parameter determining unit determines a cutoff frequency in order to remove signal noise of a high frequency band existing above an active frequency band before determining a parameter of the low pass filter, wherein the cutoff frequency detects a heartbeat signal through the PPG sensor. The energy density of the PPG signal can be determined by spectrogram for the total detection time.

The parameter determining unit analyzes a user's heartbeat signal with a frequency spectrum over time, and determines a cutoff frequency for a section less than 98% of the frequency spectrum by an energy density method.

The post-processing unit may be separated from the PPG measuring unit and the user's body end to correct the stress detection error caused by the operation noise in which the stress state and the relaxation state are repeated within a shorter time of the reference time by applying a hysteresis algorithm.

When the post-processing unit detects a stress in real time to determine the stress detection error in real time, if a detection error occurs within a reference time, the detection result of the stimulus judgment unit is fed back to the post-processing unit to correct the stress and correct the detection error. It is desirable to.

The post-processing unit detects the stress while repeating the transition between the stress state and the relaxation state, but when it is determined that the relaxation state is maintained below the reference time while determining that the stress state is due to the operation noise of the PPG measurement unit. If it is determined that the error is, and the relaxation state is maintained for more than the reference time, it is more preferable to determine that the normal transition from the stress to the relaxation state.

The post-processing unit may have the same reference time of transition from the stress state to the relaxed state and the reference time of transition from the relaxed state to the stress state.

The present invention may further include a display unit that provides information on the physical state obtained through the stimulus judging unit on a GUI screen, and a memory unit that stores information on the physical state and the stress determination result according to the selection of the inspector.

According to the heartbeat signal processing apparatus according to the present invention configured as described above, whether or not to determine the stress based on the change in the PPI detected according to the change in the heart rate, it effectively reduced the signal noise through the preprocessing unit using a low pass filter In addition, there are various effects of improving the detection error caused by the operation noise in the post-processing unit using the hysteresis algorithm.

1 is a graph showing a signal waveform induced for determining the conventional dynamic noise.
Figure 2 is a block diagram showing a heartbeat signal processing apparatus according to a preferred embodiment of the present invention.
3 is a graph representing a discrete electrical signal of the heartbeat of the present invention.
Figure 4 is a graph showing the hysteresis algorithm applied in the post-processing unit to correct the operation noise of the present invention.
5 is a graph of a time series of the PPI of the present invention.
Figure 6 is a detection graph before the post-treatment of the present invention.
Figure 7 is a detection graph after the post-treatment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

Figure 2 is a block diagram showing a heartbeat signal processing apparatus according to a preferred embodiment of the present invention, Figure 3 is a graph representing the heartbeat of the present invention as a discrete electrical signal, Figure 4 is a correction of the operation noise of the present invention This is a graph showing the hysteresis algorithm applied in the post-processing unit.

As shown, the heartbeat signal processing apparatus according to an embodiment of the present invention, PPG measuring unit 10 for detecting a heartbeat signal; A preprocessing unit 20 for removing signal noise mixed in the heartbeat signal detected by the PPG measuring unit 10 using a low pass filter 21; A PPI detector (30) for detecting the PPI signal from the heartbeat signal from which the signal noise is removed from the preprocessor (20); A post processor 40 for correcting a stress detection error of the PPI signal detected by the PPI detector 30; The post-processing unit 40 processes the error-corrected PPI signal to determine whether or not stress. If the PPI signal is larger than the reference value, it is determined to be in a relaxed state. The unit 60 is included.

The present invention configured as described above first detects the heartbeat signal from the PPG sensor 11 of the PPG measuring unit 10 to create a stress detection algorithm based on the change in heart rate.

The PPG sensor 11 measures the PPG, which is a change in the blood flow rate in blood vessels that are concentrated at the end of the body by using optical characteristics, and converts the heartbeat as shown in FIG. 3 into discrete electric signals. The PPG signal measured by the PPG sensor 11 can be distinguished from the change in heart rate through signal processing.

In the present invention, it is determined whether the stress based on this, when the body is subjected to stress stimulation increases the heart rate. That is, while the time between peak and peak (PPI: PPI) in the PPG signal decreases, the PPI increases because the heart rate decreases in the relaxed state. Based on the change in the PPI detected by the change in the heart rate, it is possible to determine whether the stress.

According to a preferred embodiment of the present invention, it is possible to determine the increase or decrease of the heart rate through the PPI analysis in the PPG signal, and at this time, it indicates the degree of activation for each frequency band of the PPG signal according to the stress. When stressed, the low frequency (LF) is activated from 0.04 Hz to 0.15 Hz, which is associated with the sympathetic nervous system. In the relaxed state, the high frequency (HF: 0.15 Hz to 0.4 Hz) associated with the parasympathetic nervous system is activated. The stress can be detected through the analysis of the active frequency (LF and HF) band.

On the other hand, since the heartbeat signal detected by the PPG measurement unit 10 is mixed with the signal noise of the high frequency band according to the heart activity, the effect of the signal noise detected with the heartbeat signal for signal processing of the PPG signal should be considered.

Therefore, according to an exemplary embodiment of the present invention, since the signal noise of the high frequency band detected together with the heartbeat signal is a disturbing factor in PPI detection, it is mixed with the PPG signal by using a low pass filter 21. Eliminate signal noise. When applying the low pass filter 21, the characteristics of the heartbeat should be taken into account. The average heart rate in adults at rest is 70 beats per minute, but the normal range of heart rates is wide. At rest, athletes can keep their heart rate below 50 bpm. When excited or anxious, your heart rate may increase to 125 bpm or higher. If the low pass filter 21 has a narrow main lobe, distortion may occur in an active frequency band, and if the low pass filter 21 has large sidelobes, signal noise may be included in the preprocessed PPG signal. . Therefore, in consideration of the characteristics of the heartbeat, it is necessary to determine the low pass filter 21 which does not generate distortion of the active frequency and can effectively remove the signal noise, and also determines the parameter of the filter. Therefore, the following describes a method for selecting an appropriate low pass filter for effectively eliminating signal noise and detecting a parameter of the filter before detecting the PPI.

In order to effectively remove the signal noise of the high frequency band, it is desirable to select a filter having a small size of the side lobes. Filters capable of effectively eliminating signal noise in pretreatment due to filter characteristics having small side lobes include a Bartlett filter, a Hanning filter and a Hamming filter. Table 1 shows the characteristics of each filter.

According to a preferred embodiment of the present invention, the Hamming filter showing the largest stopband attenuation among the commonly used filters was able to effectively remove the signal noise of more than the cutoff frequency. Therefore, in a preferred embodiment of the present invention, a Hamming filter is used as the low pass filter 21 of the preprocessing unit 20 to remove signal noise mixed in the heartbeat signal. It can be defined together.

As such, even when the Hamming filter capable of effectively removing the signal noise is selected as the low pass filter 21, if the parameters of the filter are inappropriately determined, the general low pass filter 21 for removing the active frequency band may be applied. have. At this time, before determining the parameter of the filter, the optimum cutoff frequency should be determined to remove the signal noise of the high frequency band existing above the active frequency band. The cutoff frequency is determined by spectrogram analysis of the energy density of the PPG signal with respect to the total detection time for detecting the heartbeat signal through the PPG sensor 11. The energy density spectrum of the discrete signal is defined as in Equation 2 according to Parseval theory.

The energy density spectrum according to the time change is analyzed through Equation 2, and a section less than 98% of the energy density may be determined as the cutoff frequency. The Hamming filter, which is the low pass filter 21, removes signal noise in the high frequency band higher than the cutoff frequency detected through the preprocessing. The parameter determiner 22 determines the window size using the cutoff frequency detected through the spectrogram and the energy density analysis. The equation for determining the window size is derived from the main lobe width of the Hamming filter shown in Table 1 and defined as Equation 3.

The low pass filter 21 may be a general low pass filter 21 using a spectrogram and an energy density analysis method in addition to the Hamming filter.

On the other hand, due to the mechanical and physical characteristics of the PPG sensor 11, the user's finger and the PPG sensor 11 can be separated instantly. In the section in which the heartbeat signal is not detected due to the poor contact of the PPG sensor 11, since there is no input signal, the PPI is detected very large. This PPI can be determined as a relaxed state in the detection of stress, but the low pass filter 21 is not supplemented, and thus an error may occur in the stimulus determination. Therefore, the present invention can correct the stress detection error due to the operation noise that can not be removed in the preprocessor 20 in the post processor 40.

That is, in the algorithm for determining the stress using the PPG sensor 11, it is possible to repeatedly determine the stress state and the relaxation state within a short time, but the heart rate increased quickly because the increased heart rate is not possible due to the excitability of the sympathetic nervous system. Cannot be stabilized to a relaxed state. In other words, even if you apply a relaxation stimulus to a person in a stress state, the heart rate slowly decreases and changes slowly. When the PPG sensor 11 is separated from the finger while being detected as a stress state, the PPI increases and is detected as a relaxed state. When the PPG sensor 11 is in contact with the finger, the PPI decreases to detect the stress state.

Therefore, a stress detection error occurs that repeats the stress state and the relaxation state several times within a short time due to operation noise such as poor contact. In the preferred embodiment of the present invention, the hysteresis algorithm as shown in FIG. 4 is applied by the post-processing unit 40 to correct the operation noise causing the stress detection error. The operation is as follows.

In FIG. 4, a is a section for maintaining stress by judging as an error due to an operation noise due to poor contact of the PPG sensor 11 when it is determined that the relaxation state maintained for less than 5 seconds is determined as stress, b The relaxation state is maintained for more than 5 seconds, indicating that the judgment shifts from the stress to the relaxation state. c and d indicate stress-to-relaxation, as opposed to a and b. As such, the detection error may be corrected through the post-processing unit 40 proposed in the present invention for a stress detection error that may be generated due to operation noise such as poor contact of the PPG sensor 11.

Therefore, according to the hysteresis algorithm according to an embodiment of the present invention, when the detection error occurs within a reference time while determining whether the stress in real time to correct the stress detection error, the detection of the magnetic pole determination unit 60 The feedback may be fed back to the post-processing unit 40 to determine the stress to correct the detection error.

5 is a graph obtained by converting the PPI of the present invention into a time series, FIG. 6 is a detection graph before post-processing, and FIG. 7 is a detection graph after post-processing.

In order to determine the stress, PPIs representing the peak-to-peak intervals were detected in the preprocessed PPG signal and implemented in time series. The graph of converting the PPI into a time series is shown in FIG. 5.

In the stimulus judgment, a section in which a relatively large PPI is generated based on the PPI in the neutral state is determined to be a relaxed state, and a section in which a relatively small PPI is generated based on the PPI in the neutral state is determined to be a stress state. However, when the PPG sensor 11 is separated from the finger of the test subject during the detection of the heartbeat signal, a PPI having a very large change range occurs as shown in FIG. 5. The PPI, which has a very large change range, repeatedly determines the stress state and the relaxation state in a short time in an algorithm for determining stress. In order to correct this detection error, the hysteresis algorithm was used as described above in the post processing.

Therefore, the operation noise in which the PPG sensor 11 is separated from the finger to generate an error in the PPI signal causes an error in the stress detection as shown in FIG. 6. In order to correct the operation noise, the PPI may be corrected using a hysteresis algorithm of the post-processing unit 40, and the detection error may be corrected as shown in FIG. 7 by redetecting the stress.

In addition, the stimulus determination unit 60 determines the stress by processing the error-corrected PPI signal in the post-processing unit 40. In this case, the stimulus judging unit 60 may be determined to be in a relaxed state when the PPI signal is larger than the reference value, and may be determined to be in a stress state when the PPI signal is smaller than the reference value.

To this end, the magnetic pole determination unit 60 may be a conventional microprocessor or a microcomputer having a control, arithmetic and judgment functions.

In addition, the display unit 70 that provides information on the physical condition obtained through the stimulus judging unit 60 on a GUI (Graphic User Interface) screen, and stores the information on the physical condition and the stress determination result according to the selection of the examiner. The memory unit 50 may further include.

Therefore, the stress is judged based on the change in the PPI detected according to the change in the heart rate, but the signal noise is effectively reduced through the preprocessing unit using the low pass filter, and the post-processing unit is applied to the hysteresis algorithm as described above. The detection error caused by noise can be improved.

The embodiments of the present invention described in the present specification and the configurations shown in the drawings relate to the most preferred embodiments of the present invention and are not intended to encompass all of the technical ideas of the present invention so that various equivalents It should be understood that water and variations may be present. Therefore, it is to be understood that the present invention is not limited to the above-described embodiment, and that various changes and modifications may be made without departing from the spirit and scope of the invention as defined in the appended claims. , Such changes shall be within the scope of the claims set forth in the claims.

10: PPG measuring unit 11: PPG sensor
20: preprocessor 21: low pass filter
22: parameter determination unit 30: PPI detection unit
40: post-processing unit 50: memory unit
60: magnetic pole determination portion 70: display portion

Claims (11)

PPG measuring unit for detecting a heartbeat signal;
A pre-processing unit for removing signal noise in a high frequency band mixed with the heartbeat signal detected by the PPG measuring unit using a low pass filter;
A PPI detector detecting a PPI signal from the heartbeat signal from which the signal noise is removed from the preprocessor;
Correct the stress detection error of the PPI signal detected by the PPI detection unit, if it is determined that the relaxation state is kept below the reference time while determining the stress state, it is determined as an error due to the operation noise of the PPG measurement unit, If it is determined that the relaxation state is maintained for more than a reference time, it is determined that the normal state transition from the stress to the relaxation state, the post-processing unit for detecting the stress while repeating the transition between the stress state and the relaxation state; And
The post-processing unit processes the PPI signal in which the stress detection error is corrected to determine whether or not the stress is detected. If the PPI signal is larger than the reference value, it is determined to be in a relaxed state, and if the PPI signal is smaller than the reference value, the stimulus is determined as a stress state. Determination unit; Heart rate signal processing apparatus comprising a. The method of claim 1,
The PPG measuring unit;
Heart rate signal processing apparatus characterized in that for measuring the PPG which is a change in blood flow through the blood vessel using a PPG sensor having an optical characteristic, and converts it into a discrete electrical signal to detect the heart rate signal. The method of claim 1,
The low pass filter is;
Heartbeat signal processing device, characterized in that the Hamming filter. The method according to claim 1 or 3,
The preprocessing unit;
And a parameter determining unit configured to determine parameters of the low pass filter by applying an energy density analysis method using a spectrogram to determine parameters adapted to different heartbeat signals for each individual. 5. The method of claim 4,
The parameter determiner;
Before determining the parameters of the low pass filter, a cutoff frequency is determined to remove signal noise of a high frequency band existing above an active frequency band, and the cutoff frequency is determined at a total detection time of detecting a heartbeat signal through a PPG sensor. Heart rate signal processing apparatus characterized in that for determining the energy density of the PPG signal by analyzing the spectrogram. 6. The method of claim 5,
The parameter determiner;
A heartbeat signal processing apparatus, characterized in that the user's heartbeat signal is analyzed with a frequency spectrum over time and the energy density method determines a section below 98% of the frequency spectrum as the cutoff frequency. The method of claim 1,
The post-processing unit;
Heartbeat signal, characterized in that the stress detection error is generated by applying a hysteresis algorithm, which is separated from the PPG measuring unit and the user's body end and the stress noise and relaxation state is repeated within a shorter time than the reference time Processing unit. 8. The method of claim 7,
The post-processing unit;
When the detection error occurs within the reference time during the determination of the stress in real time to correct the stress detection error, the detection result of the stimulation judgment unit is fed back to the post-processing unit to reconstruct the stress to correct the detection error, characterized in that Heart rate signal processing device. delete The method of claim 1,
The post-processing unit;
Heartbeat signal processing apparatus, characterized in that the reference time transition from the stress state to the relaxed state and the reference time transition from the relaxed state to the same state are formed. The method of claim 1,
A display unit which provides a GUI screen with information about the physical condition obtained through the stimulus determination unit; And
The heartbeat signal processing apparatus further comprises a memory unit for storing the information on the physical condition and the stress determination results according to the selection of the examiner.

Images