You are on page 1of 8

ISSN (Online) 2581-9429

IJARSCT ISSN (Print) ABCD-PQRS

International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 9, Issue 2, September 2020

Heart-Rate Variability Estimation Using


Photoplethysmography Signal
Anita Patil1 and Sayli Hulwan2
Associate Professor, Department of Electronics and Telecommunication1
Post-graduate Student, Department of Electronics and Telecommunication2
MKSSS’S Cummins College of Engineering, Pune, India
anita.patil@cumminscollege.in1, saylih2011@gmail.com2

Abstract: Monitoring heart-rate and heart-rate variability is attaining compliance using probable non-
invasive techniques such as photoplethysmography. The Photoplethysmography (PPG) is a non-intrusive
approach, mostly utilized in pulse oximetry and fitness bands to monitor the heart’s activity.
Photoplethysmography is an optical approach to detect volumetric changes in the blood. This approach
operates at 660nm and 880nm wavelength of red and infrared led respectively. This technique is good for
the estimation of heart-rate and heart-rate variability non-invasively. The heart-rate and heart-rate
variability are estimated using statistical parameters in the time domain. The signal is acquired using
MAX30102 pulse sensor. The analyzed signal data is compared with the Pulse Oximeter. The proposed
method provides the heart-rate data with the mean absolute error of 0.67.

Keywords: Photoplethysmography, pulse sensor, heart-rate, heart-rate variability, mean, standard deviation

I. INTRODUCTION
The heart is one of the most important organs. It continuously and reliably pumps blood throughout the body. Due to
the busy and stressful lifestyle of people nowadays, the monitoring of the heart is essential. Photoplethysmography has
become popular for monitoring the heart-rate.
Photoplethysmography (PPG) is a mechanism that uses an optical procedure to perceive the volumetric distortion in
the blood and operates at 660nm and 880nm wavelength. This technique utilizes optoelectronic components like a
photodetector to measure the small variabilities in light intensity corresponding with deformation in the peripheral
blood volume, and a source of light to illuminate the target tissue. The basic operation of a PPG sensor is measuring the
volume inside an organ by evaluating the reaction between the light source and the body tissue. The heartbeat is
synchronous to the volumetric changes in the blood, which is utilized to determine the heart-rate. When this light
moves through the target tissues, it’s absorbed by arterial and venous blood. PPG measures the volumetric changes
occurring in the arterial blood with each pulse beat. These changes can be detected in peripheral parts of the body such
as finger-tip, wrist, or earlobe using the PPG technique.
The variations in the PPG signal are caused as a result of the transitions occurring in the arterial blood volume
simultaneous with the heartbeat. The amplitude variations depend on the amount of blood rushing into arteries and
ventricular. The time-period of each beat in the signal is prescribed by pulse rate and amplitude is dictated by the
changeable concentration of constituents of arterial blood. The range of PPG signal pulse wave is 0.5 – 4.0 Hz [1]. The
PPG signal is comprised of Alternating Coupling (AC) component and Direct Coupling (DC) component. The AC -
component is superimposed on the DC component, which has to be removed as shown in Fig. 1. The AC component
contains the pulse variation of blood volume in arteries; therefore, it is utilized to compute heart-rate. The cardiac
rhythm correlates with the periodicity of the PPG signal that is utilized for the approximation of heart-rate (HR) and
heart-rate variability (HRV). Heart-rate and heart-rate variability are important factors in order to provide an in-depth
knowledge on the volumetric changes in the heart. The HRV is the measurement of dissimilitude of the average HR. It
provides knowledge viable to balance of the sympathetic - parasympathetic autonomic system which in return provides
intelligence on the jeopardy of unforeseen cardiac demise.
Copyright to IJARCST DOI: XX.082020/IJARSCT 67
www.ijarsct.co.in
ISSN (Online) 2581-9429
IJARSCT ISSN (Print) ABCD-PQRS

International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 9, Issue 2, September 2020

In this research work, a pulse sensor is used to acquire the PPG data of resting subject, to gauge the HR and HRV
using Python programming language. There are two types of photoplethysmography i.e. reflectance and transmittance.
This proposed methodology uses a reflectance PPG due to the advantage of the light-dissipating response of body
tissue. Using reflective PPG, we can acquire the signal by placing the sensor on distinctive body locations, in particular,
the forehead, the outer ear canal, the areas around the bicep or calve muscles, and the area around a wrist. The
reflectance mode PPG sensor guarantees flexibility to the user, especially during physical activities.

Figure 1: Representation of PPG signal characteristics


In reflectance Photoplethysmography, the source of light and the photodetector are placed adjacent to each other on
the target body part, as shown in Fig. 2 [2]. In reflectance photoplethysmography, the finger is illuminated by the light
source as shown in Fig. 2. relative to the volume of blood or blood flow into the interior of the fingertip, an assortment
of light gets consumed, some quantity of light is reflected and some is transmitted. The intenseness of light fluctuates
along with the volume of blood into the interior of the fingertip, that transforms simultaneously with the heartbeat. The
higher the volume of blood, lower the intensity of reflected light and vice versa.

Figure 2: Reflectance-mode photoplethysmography, light-emitting diode (LED) and photodetector (PD).


The PPG signal is obtained for estimating both HR and HRV. The acquired signal is processed and used for
calculating statistical parameters in the time domain. These statistical parameters are used to further calculate valid HR
and HRV from the signal. The estimated result is compared with Pulse Oximeter to calculate the accuracy of the
proposed methodology. This methodology conceptually and computationally provides accuracy to a good extent.

Copyright to IJARCST DOI: XX.082020/IJARSCT 68


www.ijarsct.co.in
ISSN (Online) 2581-9429
IJARSCT ISSN (Print) ABCD-PQRS

International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 9, Issue 2, September 2020

II. LITERATURE REVIEW


Photoplethysmography technique has become popular amongst biomedical researchers and developers in the past
few years for monitoring of the heart. This technology is used in various wearable devices for continuous monitoring of
the heart. Some related works have been mentioned below.
Lee et al. [3], proposed a methodology to cancel motion artefacts from the PPG signal acquired using a single green
LED. The two reflective signals acquisition is carried out by switching the light intensities between high and low. To
achieve motion artefact removal the use of differential amplifiers with common-mode rejection is used the proposed
methodology.
Saquib et al. [4], proposed a pulse rate monitoring device using photoplethysmography. The data is acquired
commonly from fingertip and earlobe using TCRT1000 optical reflective sensor. The acquired sensor data is processed
to remove the DC component using a passive high-pass filter and amplify the AC component using an active low-pass
filter. They have employed PIC16F648A microcontroller with a timer module to calculate the heart-rate from filtered
data.
Srinivas et al. [5], proposed a methodology of evaluation concerning heart-rate variability from circumferential heart
wave obtained employing photoplethysmography sensor. The signal is acquired using red led and photodiode. The
transmitted signal is amplified using a transimpedance amplifier. The amplified signal is filtered using a third order
Butterworth filter with cut-off frequencies of 0.4Hz and 5Hz. The HR and HRV are calculated using a versatile
algorithm.
Bolanos et al. [6], discussed the development of a PDA-based ECG and PPG signal acquisition system to do a
comparative study of HRV features obtained from both signals. They have used autoregression (AR) modelling,
Poincare’ plots and other mathematical parameters to derive comparative measures from both ECG and PPG signals.
They discuss the advantages of a PPG signal for the estimation of HRV.
Moraes et al. [1], discussed the photoplethysmography technique. They have discussed the type of sensors available
using this technique. The methods for studying and analyzing the PPG signal are described such as linear (time and
frequency domain) and non-linear methods (Scaled Amplified Analysis, Displacement Floatation Analysis, etc.). They
have also discussed latest technologies used in medical diagnosis tools for example, the Internet of Things, Internet of
Health, genetic algorithms biosensors and artificial intelligence (AI) to conclude that the PPG is a crucial mechanism
for prognosis of heart related diseases.
Longmore et al. [7], discussed the best anatomical location to collect PPG signal for an accurate measure of Heart-
rate, SpO2 and Respiration rate simultaneously. They used eight anatomical locations out of which finger gives
accurate results for all three features at rest.
Ghamari et al. [8], proposed the design and development of PPG based wristband-type instrument for HRV analysis.
The use of reflective mode PPG sensor is used for data acquisition. The acquired signal is filtered and amplified to
isolate the AC component from the DC components. The first and second derivative of the signal is calculated to
analyze arterial stiffness. The time-domain analysis is used to extract heart-rate and frequency-domain analysis is used
for HRV.
In this paper, section 1 describes the brief introduction to PPG technique, a brief explanation of the mode of PPG and
advantages of reflective PPG sensor. In section 2, the literature review used in this paper. Section 3 describes the data
acquisition in order to obtain the estimation of heart-rate and heart-rate variability. It includes the basic block diagram
for all the steps. It described the materials used and methods used for data acquisition.
In section 4, the pre-processing of the signal and methodology is described. It includes feature extraction of different
parameters. Section 5, illustrated the results obtained from the extracted features. It also a comparative study of reading
obtained from the proposed methodology and an existing Pulse Oximeter. This method provides to be accurate to a
good extent.

Copyright to IJARCST DOI: XX.082020/IJARSCT 69


www.ijarsct.co.in
ISSN (Online) 2581-9429
IJARSCT ISSN (Print) ABCD-PQRS

International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 9, Issue 2, September 2020

III. DATA ACQUISITION


This section illustrates the acquisition of the heart signal from PPG. This section describes the sensor configuration
for data acquisition. The PPG signal is acquired using optical reflectance mode PPG sensor MAX30102. This sensor
comes with the advantages of built-in in ambient light cancellation circuit and signal conditioning circuit. The mode of
data acquisition from MAX30102 is set to be red led plus infrared led. A longer pulse width changes the amount of
time the sensor led radiate into the fingertip before establishing the absorption of light. Before the collection of data, th
the
fingertip is fully illuminated which results in high-resolution
high data. Due to this reason, the sensor has limited time to
collect the required data. Therefore, the pulse width is set to 69µs. The data is captured at a sampling rate of 400Hz.
The ADC range is set to 14bits (16384). The data is processed using the Python script and displayed using the plot
function. To approximate the heart-rate and heart-rate variability and calculation of the statistical signal parameters.
The proposed method is compared with the existing Pulse oximeterdevice. The proposed method uses the workflow
shown in Fig. 3.

Figure 3: Block diagram for heart-rate and heart-rate variability estimation.


The heart signal is acquired using MAX30102 from the fingertip of the subject/case at rest. The raw PPG signal with
a timestamp is stored in the form of data--frame using Python Script. The signal storedred signal is of 1m duration. This
raw signal is very low and contains unwanted noise. It is observed that this noise has a high frequency.

IV. METHODOLOGY
The PPG signal consists systolic peak and diastolic peak as shown in Fig. 1. The heart-rate rate is extracted using
systolic peaks. The acquired PPG signal is very low and noisy. The acquired signal shows a large amplitude variation
due to which there are complications in analysis of the signal. For extracting the heart-rate,
heart rate, peak placement is not
crucial
ucial but for HRV calculation it is critical. The need to amplify and smoothen the signal is necessary to estimate HR
and HRV.
The data (A) is scaled (AS) by estimating the minimum (Amin) and the maximum (Amax) values in the data-frame
data as
shown in equation 1. To enhance the peaks the data is squared and scaled with the lower (AL) and upper (AU) threshold,
0 and 1024 respectively [3].


= ( − ) ∗ + (1)

Copyright to IJARCST DOI: XX.082020/IJARSCT 70


www.ijarsct.co.in
ISSN (Online) 2581-9429
IJARSCT ISSN (Print) ABCD-PQRS

International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 9, Issue 2, September 2020

In the proposed method the iteration for scaling


scaling is done twice as to get the enhanced signal. The scaled data is
filtered using Butterworth bandpass filter of third order with cutoff frequencies of 0.75 and 3.5 Hz i.e. all the
frequencies below 45 BPM and above 210 BPM are removed from the signal to get get accurate results. The frequencies
from the data-frame
frame that are within 0.7 and 3.5 Hz. The filtered signal data (AF) is processed to find peaks. The
filtered data is shown in Figure 4.

Figure 4: Filter PPG signal


The moving average is calculated to detect the heartbeat. The average mean to obtain the peaks. The mean of the
filtered (A ) data points are calculated using equation 2.
1
= (2)

where N is number of data points and i is each value of the data point. The average mean uses a sliding window
function with the window size of 0.75 for each segment. The mean of data points achieved by the sliding window
function gives the average mean needed for peak detection.
For peak detection, consideration of both x and y coordinates is to take place. The point on x coordinate that is less
than moving average mean and the data point on y coordinate that is the greater value for the x coordinate is taken. The
peak is detected using these coordinate points. T
The
he array of these peak points is stored and mapped on the filtered data.
The calculation of standard deviation is necessary for accurate peak detection using equation 3.

1
= − (3)
−1

It was found that by raising the rolling mean by by 5 percent, accurate peaks positions are detected (peak-list).
(peak To
further analyze the obtained peak positions (P), the data is checked for outliers. The outlier is a data point that exceeds
1.5 times the interquartile range higher than the third quartile (Q3) or lower than the first quartile (Q1) [9]. Interquartile
range (IQR) is the difference between the 25th and 75th percentile of the data [10]. The outliers are detected by
defining new decision range and the data point lying outside this range will be considered an outlier. Equation 4 is used
for the calculation of the decisive range.
= −

= − (1.5 ∗ ) (4)

= + (1.5 ∗ )

The median is set and indices are replaced for data points that are marked as outliers. The peak to peak interval (PPI)
is calculated using equation 5.

1
= ( − ) ∗ 1000 (5)
_

where N is number of peak points and i is the value of each peak position. The accurate peaks for heart-rate
estimation are calculated. The peak to peak intervals for the signal has been calculated. The time-domain
domain measurements
Copyright to IJARCST DOI: XX.082020/IJARSCT 71
www.ijarsct.co.in
ISSN (Online) 2581-9429
IJARSCT ISSN (Print) ABCD-PQRS

International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 9, Issue 2, September 2020

to be calculated are the standard deviation of PPI (SDPP), the standard deviation of consecutive differences (SDCD)
and root mean square of consecutive differences (RMSCD). The heart-rate variability is calculated using equation 6.
The heart-rate (BPM) is calculated using equation 7.

1
= (6)

1
= ∗ 60000 (7)

The standard deviation of PPI is the golden standard for medical hierarchy of heart risk when recorded over a 24 h
period [11]. The SDPP is calculated using equation 8.

1
= ( − ) (8)
−1

The standard deviation of consecutive differences (SDCD) is calculated using equation 9.

1
= −

1
= − (9)
−1

The RMSCD reflects the heart beat-to-beat variance and is applied to approximate the vagally mediated changes in
HRV [12]. The RMSCD is calculated using equation 10.

1
= ( − ) (10)

V. RESULTS AND DISCUSSION


The proposed method used in order to obtain the determination of heart-rate was also compared with existing Pulse
Oximeter device. For comparing the estimated heart-rate using the proposed method with the existing Pulse Oximeter
device, both the readings were taken simultaneously. The acquired readings of heart-rate were stored in a CSV file. The
mean absolute error was calculated to check the accuracy of the proposed methodology. The Fig. 5. shows the graphical
representation of the comparison of the heart-rates obtained by pulse oximeter vs the proposed method.
The calculation of mean absolute error shows that the proposed method has a mean absolute error of 0.67. For the
accuracy of good extant for estimation of heart-rate, the signal should at least be of 2 – 4 minutes. This method can also
be further developed to calculate the breathing rate. This proposed method is the estimation of heart-rate and HRV of a
person at rest to reduce the motion artefacts. Stress can be detected using HRV for a 24 hrs recorded PPG signal. The
RMSCD will be the decisive factor for detection of stress as it provides information on vagally mediated changes.
Copyright to IJARCST DOI: XX.082020/IJARSCT 72
www.ijarsct.co.in
ISSN (Online) 2581-9429
IJARSCT ISSN (Print) ABCD-PQRS

International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 9, Issue 2, September 2020

Figure 5: Comparative study of the proposed methodology vs Pulse oximeter.

VI. CONCLUSION
The heart rate and heart rate variability were computed with the help of time-domain
t domain statistical parameters from the
acquired PPG such as standard deviation and mean. The signal was processed using a bandpass filter. The time
time-domain
measurement such as standard deviation of PPI (SDPP), the standard deviation of consecutive differences
diffe (SDCD) and
root mean square of consecutive differences (RMSCD) were calculated for better understanding of HRV. The proposed
system has a good accuracy of mean absolute error 0.67. From the time-domain
domain analysis it is observed that a very short
duration
ion signal (< 1 min) can give accurate heart-rate.
heart rate. The proposed methodology will work for 24 hrs recorded PPG
signal to detect stress using HRV.

REFERENCES
[1]. Moraes, J. L., Rocha, M. X., Vasconcelos, G. G., Vasconcelos Filho, J. E., De Albuquerque, V. H. C., &
Alexandria, A. R. (2018). Advances in photopletysmography signal analysis for biomedical applications.
Sensors, 18(6), 1894.
[2]. Tamura, T., Maeda, Y., ., Sekine, M., & Yoshida, M. (2014). Wearable photoplethysmographic sensors - past
and present. Electronics, 3(2), 282
282-302.
[3]. Lee, H., Ko, H., Jeong, C., & Lee, J. (2016). Wearable photoplethysmographic sensor based on different LED
light intensities. IEEE
EE Sensors Journal, 17(3), 587-588.
587
[4]. Saquib, N., Papon, M. T. I., Ahmad, I., & Rahman, A. (2015, January). Measurement of heart rate using
photoplethysmography. In 2015 International Conference on Networking Systems and Security (NSysS) (pp.
1-6). IEEE.
[5]. Srinivas, K., Reddy, L. R. G., & Srinivas, R. (2007). Estimation of heart rate variability from peripheral pulse
wave using PPG sensor. In 3rd Kuala Lumpur International Conference on Biomedical Engineering 2006 (pp.
325-328).
328). Springer, Berlin, Heidelberg.
Heidel
[6]. Bolanos, M., Nazeran, H., & Haltiwanger, E. (2006, August). Comparison of heart rate variability signal
features derived from electrocardiography and photoplethysmography in healthy individuals. In 2006
International Conference of the IEEE Engineering
Engineering in Medicine and Biology Society (pp. 4289
4289-4294). IEEE.

Copyright to IJARCST DOI: XX.082020/IJARSCT 73


www.ijarsct.co.in
ISSN (Online) 2581-9429
IJARSCT ISSN (Print) ABCD-PQRS

International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 9, Issue 2, September 2020

[7]. Longmore, S. K., Lui, G. Y., Naik, G., Breen, P. P., Jalaludin, B., & Gargiulo, G. D. (2019). A comparison of
reflective photoplethysmography for detection of heart rate, blood oxygen saturation, and respiration rate at
various anatomical locations. Sensors, 19(8), 1874
[8]. Ghamari, M., Soltanpur, C., Cabrera, S., Romero, R., Martinek, R., & Nazeran, H. (2016, August). Design and
prototyping of a wristband-type wireless photoplethysmographic device for heart rate variability signal
analysis. In 2016 38th Annual International Conference of the IEEE Engineering in Medicine and Biology
Society (EMBC) (pp. 4967-4970). IEEE.
[9]. Renze, John. "Outlier." From MathWorld--A Wolfram Web Resource, created by Eric W. Weisstein.
https://mathworld.wolfram.com/Outlier.html
[10]. Interquartile range. (2020, August 21). Retrieved from https://en.wikipedia.org/wiki/Interquartile_range
[11]. Electrophysiology, T. F. O. T. E. S. O. C. T. N. A. S. O. P. (1996). Heart rate variability: standards of
measurement, physiological interpretation, and clinical use. Circulation, 93(5), 1043-1065.
[12]. Shaffer, F., McCraty, R., & Zerr, C. L. (2014). A healthy heart is not a metronome: an integrative review of
the heart's anatomy and heart rate variability. Frontiers in psychology, 5, 1040.
[13]. van Gent, P., Farah, H., van Nes, N., & van Arem, B. (2019). HeartPy: A novel heart rate algorithm for the
analysis of noisy signals. Transportation research part F: trac psychology and behaviour, 66, 368-378.
[14]. Brumeld, A. M., & Andrew, M. E. (2005). Digital pulse contour analysis: investigating age-dependent
indices of arterial compliance. Physiological measurement, 26(5), 599.
[15]. Marcinkevics, Z., Greve, M., Aivars, J. I., Erts, R., & Zehtabi, A. H. (2009). Relationship between arterial
pressure and pulse wave velocity using photoplethysmography during the post-exercise recovery period. Acta
Univesitatis Latviensis: Biology, 753, 59-68.
[16]. Shriram, R., Wakankar, A., Daimiwal, N., & Ramdasi, D. (2010, April). Continuous cuffless blood pressure
monitoring based on PTT. In 2010 International Conference on Bioinformatics and Biomedical Technology
(pp. 51-55). IEEE.
[17]. Shaffer, F., & Ginsberg, J. P. (2017). An overview of heart rate variability metrics and norms. Frontiers in
public health, 5, 258
[18]. Zainal, N. I., Rodzi, M. Z. M., Khan, S., Habaebi, M. H., & Gunawan, T. S. (2016, December). Design and
development of wireless PPG data acquisition for health monitoring application using Bluetooth module. In
2016 IEEE Student Conference on Research and Development (SCOReD) (pp. 1-6). IEEE.

Copyright to IJARCST DOI: XX.082020/IJARSCT 74


www.ijarsct.co.in

You might also like