JASC: Journal of Applied Science and Computations ISSN NO: 1076-5131

Noninvasive Measurement of Blood Pressure from

PPG Signals- An Overview
Jean Effil N#1, Rajeswari R#2
1
Department of Computer Science, Manonmaniam Sundaranar University Constituent College of Arts and
Science,Kadayanallur
2
Department of Computer Applications, Bharathiar University,Coimbatore
[email protected]

2
[email protected]

Abstract— Hypertension is a degenerative medical condition which slowly downgrades one’s physical health and leads to very many
complications at an uncontrolled stage. Hence in order to detect hypertension at its early stage, blood pressure must be monitored
regularly. The aim of this work is to provide an overview of techniques used for noninvasive measurement of blood pressure from
photoplethysmography (PPG) signals. To measure blood pressure noninvasively, photoplethysmography signals are extracted from
fingertip images and the variations in the light signals are analyzed to determine the blood pressure levels. Photoplethysmography
signals can be captured using PPG sensors and blood pressure is estimated by analysing the features extracted from PPG signals. In
the recent years, much research work has been performed in measuring blood pressure from PPG signals captured from fingertip
images using smartphone camera. This paper explores the different methodologies used to measure blood pressure from PPG signals
captured using PPG device and also from smartphone camera. The advantages and limitations have been discussed. But these
existing methods do not guarantee 100 percent accuracy. The intention of this work is to motivate the research community to exhibit
their contributions in this area to achieve 100 percent accuracy in noninvasive blood pressure measurement using PPG signals.
Keywords— Blood pressure, Photoplethysmography, Smartphone, Fingertip images, Regression

I. INTRODUCTION
Blood pressure is the force exerted by the blood against the arteries as it circulates throughout the body. As the heart beats
(contracts and expands), it creates pressure that forces blood in and out of the heart and eventually through arteries. As per the
laws of nature, for every action there is an equal and opposite reaction, hence the blood circulation is possible only by the heart
exerting pressure to push the blood throughout the arteries to carry oxygen and energy, this pressure is known as blood pressure.
Hence it is clear that the blood pressure is created as a result of two continuous and periodic forces : The first force called
systolic pressure is the pressure within the arteries when the heart pumps out blood. The second force called diastolic pressure
is the pressure in the arteries while the heart muscle is resting between beats and refilling the blood. Blood pressure is measured
in millimeters of mercury (mm Hg) and recorded with the systolic reading first, followed by the diastolic reading. Pulse
pressure is the difference between the systolic and the diastolic pressures. If the difference between the systolic and diastolic
pressure reading is less than 45 mm Hg, then the individual is in a safe condition and there is no abnormality in their blood
pressure.

American Heart Association classifies Blood pressure readings as Normal, Elevated, Hypertensive stage 1 and stage 2 and
Hypertensive crisis. When the systolic BP is equal to 120 mm Hg and diastolic BP is equal to 80 mm Hg, then the Blood
Pressure is considered to be normal. At the elevated stage, the systolic BP reading is between 120 mm Hg - 129 mm Hg and
diastolic BP reading is less than 80 mm Hg. At this stage, there is a need to alert patients to be aware of the increasing health
risks and to make changes to their lifestyle. When the systolic BP increases and is between 130 mm Hg – 139 mm Hg and the
diastolic BP is between 80 mm Hg – 89 mm Hg, then BP is considered to be High (Hypertensive stage 1). When the systolic
BP is higher than 140 mm Hg and diastolic BP is more than 90 mm Hg, then BP is considered to be Higher (Hypertensive stage
2 ), at this stage patient requires medication and drastic lifestyle changes. When the Blood Pressure reading exceeds 180mm
Hg/120 mm Hg, then it is the stage of hypertensive crisis, the patient requires immediate medical attention [14]. The blood
pressure of a human varies continuously due to physical activity, medication, anxiety and emotions [12]. High blood pressure
is a life threatening medical condition, at elevated stages it may lead to cardiovascular diseases such as stroke and chronic
kidney disease and consequently is a risk factor for life [2]. Hence it is necessary to monitor one’s blood pressure regularly and
in case of variations from the normal level lifestyle modifications and medications are necessary.

The common devices used for blood pressure measurement are sphygmomanometer, stethoscope and digital blood pressure
monitors. Measuring blood pressure from photoplethysmography signals have become an area of interest among the
researchers in the last few years. Photoplethysmography is a non invasive method, that uses low intensity infrared light to
measure blood volume changes in peripheral circulation. When light passes through biological tissues, it is absorbed by blood,
bones and skin pigments. Blood absorbs light stronger than the surrounding tissues, hence it is easy to detect the changes in
blood flow.

Volume VI, Issue IV, April/2019 Page No:1064

JASC: Journal of Applied Science and Computations ISSN NO: 1076-5131

Thus by measuring the blood volume changes, blood pressure can be assessed.PPG device that is used to capture PPG
signals consists of a light source and a photo detector. Infra red LED is the most commonly used light source. Accuracy of
results depends on the factors influencing the environment of measurement. The most commonly used measuring areas are the
fingertip, earlobe and forehead as the blood flow can be easily measured in these areas [1][15]. Light source illuminates the
skin and the photodetector captures the intensity of light variations. These light variations in conjunction with time form the
signal which is used to calculate blood pressure.

In the recent years, there have been many research works carried out in measuring blood pressure from PPG signals using
smartphone. PPG signals are captured by illuminating skin using camera flash light and the changes in skin illuminated light
variations are measured using smartphone camera. This paper describes some of the existing techniques available to measure
blood pressure from PPG signals captured using PPG device and smartphone. The rest of the paper is organized as follows:
Section 2 describes the approaches used to calculate blood pressure from PPG signals captured using PPG device. Section 3
describes the approaches used to calculate blood pressure from PPG signals captured using smartphone camera. Section 4 gives
the conclusion and the possibilities for future works.

II. MEASURING BLOOD PRESSURE FROM PPG SIGNALS CAPTURED USING PPG DEVICE
Hassan et al. [1] have proposed a method in which systolic blood pressure (SBP) is calculated from pulse wave transit time
(PWTT). Time interval between ECG R-wave and the peak of peripheral pulse in the same cardiac cycle is selected as PWTT.
Peripheral pulses are recorded at the finger using the technique of photoplethysmography. Measure of ECG R-wave and
peripheral pulses are recorded from each subject. At the same time, blood pressure is measured using standard BP monitor.
The measurement is taken for a person with three trials (during resting, during exercise and after exercise). The procedure is
repeated for more persons. It is observed in [1] using linear regression that SBP varies inversely with PWTT. Using least
square algorithm, linear equation, SBP=a.PWTT+b is derived, which shows that SBP is linearly related to PWTT, a and b are
the constant coefficients. The derived equation can be used to estimate the blood pressure of any person if the PWTT is known.

To estimate blood pressure, Teng et al. [3] have captured PPG signals from fingertip of 15 young healthy subjects and
examined few features such as width of 2/3 pulse amplitude, width of ½ pulse amplitude, systolid upstroke time and diastolic
time. The correlation analysis is performed on blood pressure with each of these features. Correlation of diastolic time with BP
is found to be higher than BP with other features. Regression line is derived to estimate BP.

Rohan et al. [2] have estimated Systolic blood pressure(SBP) and Diastole blood pressure(DBP) from specific features
extracted from PPG signals. The features include systolic upstroke time, diastolic time and the time delay between the systolic
and diastolic peak of the waveform. It is observed in [2] that there exists negative correlation between BP and each of the three
features. Among the three features, it is observed that time delay and diastolic time correlates well with the diastolic blood
pressure measured using OMRON digital BP monitor. SBP is found to be best correlated with diastolic time.

Yan and Zhang [4] have calculated BP from normalized harmonic area (NHA) that is extracted from PPG signals using
discrete period transform. NHA is calculated from the changes in the distribution of harmonic components of PPG waveform.
It is demonstrated that NHA has a good correlation with blood pressure with NHA and smaller BP estimation error when
compared to pulse transit time and diastolic time.

In [5], Myint et al, have developed a mathematical model in which BP is calculated from PTT. PTT is calculated from
PPG waveforms generated from finger and wrist tissues in the same cardiac cycle. To measure the PPG waveforms, finger PPG
sensor and wrist PPG sensor are used. The analog signals are converted to digital signals. PTT is calculated from resultant
digital signals. The study shows that PPG varies inversely with blood pressure. Positioning the PPG sensor on the wrist is a
challenging factor. Positioning wrongly may produce incorrect signals which may lead to inaccurate results.

Foo et al. [7] have formulated a regression equation that relates SBP and vascular transit time (VTT) changes. They have
used phonocardiography and photoplethysmography to measure VTT changes. VTT is the transmission delay between the first
heart sound of the phonocardiogram (PCG) and the upstroke of the corresponding photoplethysmogram (PPG). Electronic
stethescope was used to measure PCG. DS-100 finger probe was used to measure PPG readings. Linear regression analysis
was done to determine the correlations between SBP and VTT changes, DBP and VTT changes and heart rate and VTT changes.
It was observed that SBP has the strongest correlation with VTT changes.

Volume VI, Issue IV, April/2019 Page No:1065

JASC: Journal of Applied Science and Computations ISSN NO: 1076-5131

Kurylyak et al. [6] have used artificial neural network to estimate BP. Training data was extracted from multiparameter
intelligent monitoring in intensive care waveform database.from this database, PPG heartbeats, 15000 in number and the
corresponding BP values were extracted. 21 parameters extracted from each of the 15000 PPG heartbeats, forms the input data.
Multilayer feed-forward back propagation ANN with 21 input neurons and 2 output neurons were used to estimate SBP and
DBP. It is proved that this method show better accuracy than the linear regression method.

Xing et al. [13] have estimated blood pressure from PPG signals captured from fingertip alone. To derive their model they
have obtained PPG signals and corresponding BP reading from freely distributed MIMIC II (Multiparameter Intelligent
Monitoring in Intensive Care) database. As a first step, normalization of PPG signals have been made. Key features are
extracted from normalized PPG signals using fast fourier transform. Artificial neural network is trained with these key features
as input data and BP reading as output data. The results are compared with sphygmomanometer. Bland altman analysis shows
limit of agreement to be 4.1 mmHg, that is acceptable in clinical settings.

III. MEASURING BLOOD PRESSURE FROM PPG SIGNALS USING SMARTPHONE

Hosseini, et al. [9] have proposed a BP model in which BP is calculated from pulse wave velocity (PWV). Pulse wave
velocity which is inversely propotional to pulse transit time (PTT) is calculated from product of BDC and height of the subject
and the resultant value is divided by PTT. PTT is calculated as the time interval between the peak of ECG and the arrival of
peak wave of the pulse oximeter in the same cardiac cycle. Using linear regression model, BP is calculated from PWV.

In [8] Atef et al. have used the multiple linear regression model in which blood pressure is calculated from pulse transit
time and heart rate. Pulse transit time is calculated from ECG signals and PPG signals. ECG signals and PPG signals are
obtained using ECG and PPG sensors respectively. These signals are sent to Smartphone/PC through serial port or Bluetooth
and BP computation is performed using Android application.

Aman et al. [11] have proposed an artificial neural network (ANN) model using PPG signal from which BP can be
calculated. To train and validate ANN model, 1,750,000 pulses have been extracted from a public database. This method is
based on PPG signal obtained from fingertip alone. PPG signal magnitude and temporal features (8), APG signal derived
features(19), non linear derived features (8) and heart rate variability based features (11), a total of 46 features have been
extracted. ANN model is trained with these 46 features as input and diastolic pressure as a output. In addition to 46 features,
diastolic pressure is also added as input to train the ANN model for the estimation of systolic pressure. The accuracy is found
to be significantly higher compared to previous studies.

Matsumura, et al. [10] have presented a model in which BP is estimated from heart rate and modified normalized pulse
volume which can be measured using PPG signals captured using smartphone. They have derived this model using the evidence
from circulation physiology and psychophysiology which shows that mean arterial pressure can be calculated as the product of
cardiac output and total peripheral resistance. Cardiac output and total peripheral resistance is found to be correlated with heart
rate and modified normalized pulse volume, respectively. Hence, using the exponential transformation of linear polynomial
equation, mean arterial pressure, systolic BP, diastolic BP are estimated. This method requires only smartphone and no
additional device. Also it is proved in [10] that their proposed method provides more accuracy compared to combined PWV–
PPG-based method with machine learning using PPG feature extraction. The participants were only young people and does not
involves measurement under various stressful tasks.

Vikram et al. [12] have proposed a model in which systolic BP is calculated from Vascular transit time (VTT). VTT is
defined as the time taken by the blood to travel from the heart to an extremity of the body for one stroke of the heart. VTT is
calculated by recording heart sound using microphone/another smartphone and the arrival of pulse wave at the fingertip in the
same cardiac cycle which is measured from peak of PPG signal captured from fingertip using smartphone camera. The time
interval between heart sound and the arrival of corresponding pulse wave at fingertip is calculated as VTT. The relation
between systolic BP and VTT is obtained using linear curve fitting. Diastolic BP is calculated from pulse pressure and systolic
BP. Pulse pressure is computed from stroke volume, heart rate, weight and age of the person using the equation given in [12].
To compute stroke volume, ejection time is measured. The time difference between two consecutive heart sounds recorded
using microphone is calculated as ejection time. Stroke volume is computed from ejection time, body surface area, heart rate
and age of the subject using the equation proposed in [12]. Then the diastolic pressure is calculated by subtracting pulse
pressure from systolic pressure. To validate their proposed system, the estimated blood pressure values are compared with
spygmomanometer, and have showed 95-100 percent accuracy.

Volume VI, Issue IV, April/2019 Page No:1066

JASC: Journal of Applied Science and Computations ISSN NO: 1076-5131

IV. CONCLUSION
This work provides an overview of techniques used for noninvasive measurement of blood pressure using
photoplethysmography (PPG) signals. Some researchers have used PPG signals captured from fingertip and ECG signals to
estimate blood pressure. Few other researchers have used PPG signals captured from two different sites such as fingertip and
wrist to estimate blood pressure. This method requires two PPG sensors or two smartphone to capture PPG signals. Few
researchers have used microphone to record heart sound and PPG device or smartphone to capture PPG signals and from these
measurements, blood pressure is calculated. While there have been studies where researchers have used only PPG signals
captured from fingertip alone to estimate blood pressure. These methods do not provide 100 percent accuracy as of now. Our
future work will focus on developing an improved methodology to measure blood pressure noninvasively from PPG signals
with enhanced accuracy.

ACKNOWLEDGMENT
The first author would like to thank Anna University, Regional Campus, Coimbatore and Manonmaniam Sundaranar
University Constituent College of Arts and Science, Kadayanallur, Tirunelveli District. The second author would like to thank
Bharathiar University, Coimbatore for providing the necessary support to carry out the research work.

REFERENCES
[1] Hassan, M. K. B. A., Mashor, M. Y., Nasir, N. M., & Mohamed, S., “Measuring blood pressure using a photoplethysmography approach.” In
4th Kuala Lumpur International Conference on Biomedical Engineering 2008, Springer, Berlin, Heidelberg, pp. 591-594, 2008.
[2] Samria, R., Jain, R., Jha, A., Saini, S., & Chowdhury, S. R., “Noninvasive cuff'less estimation of blood pressure using Photoplethysmography
without electrocardiograph measurement.” IEEE REGION 10 SYMPOSIUM, pp. 254-257, 2014.
[3] Teng, X. F., & Zhang, Y. T., “Continuous and noninvasive estimation of arterial blood pressure using a photoplethysmographic approach.” In
Proceedings of the 25th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (IEEE Cat. No. 03CH37439)
Vol. 4, pp. 3153-3156, 2003.
[4] Yan, Y. S., & Zhang, Y. T., “Noninvasive estimation of blood pressure using photoplethysmographic signals in the period domain.” In 2005 IEEE
Engineering in Medicine and Biology 27th Annual Conference, pp. 3583-3584, 2006.
[5] Myint, C. Z., Lim, K. H., Wong, K. I., Gopalai, A. A., & Oo, M. Z., “ Blood pressure measurement from photo-plethysmography to pulse transit
time.” IEEE Conference on Biomedical Engineering and Sciences (IECBES), pp. 496-501, 2014.
[6] Kurylyak, Y., Lamonaca, F., & Grimaldi, D., “A neural network-based method for continuous blood pressure estimation from a PPG signal.” IEEE
International instrumentation and measurement technology conference (I2MTC), pp. 280-283, 2013.
[7] Foo, J. Y. A., Lim, C. S., & Wang, P., “Evaluation of blood pressure changes using vascular transit time.” Physiological measurement, 27(8), 685,
2006.
[8] Atef, M., Xiyan, L., Wang, G., & Lian, Y., “PTT based continuous time non-invasive blood pressure system.” In 2016 IEEE 59th International
Midwest Symposium on Circuits and Systems (MWSCAS), pp. 1-4, 2016.
[9] Hosseini, H. G., Baig, M. M., Mirza, F., & Luo, D., “Smartphone-based continuous blood pressure monitoring application-robust security and
privacy framework.” IEEE Region 10 Conference (TENCON), pp. 2939-2942, 2016.
[10] Matsumura, K., Rolfe, P., Toda, S., & Yamakoshi, T., “Cuffless blood pressure estimation using only a smartphone.” Scientific reports, 8(1), 7298,
2018.
[11] Gaurav, A., Maheedhar, M., Tiwari, V. N., & Narayanan, R., “Cuff-less PPG based continuous blood pressure monitoring—A smartphone based
approach.” 38th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), pp. 607-610, 2016
[12] Chandrasekaran, V., Dantu, R., Jonnada, S., Thiyagaraja, S., & Subbu, K. P., “Cuffless differential blood pressure estimation using smart phones.”
IEEE Transactions on Biomedical Engineering, 60(4), 1080-1089, 2013.
[13] Xing, X., & Sun, M., “Optical blood pressure estimation with photoplethysmography and FFT-based neural networks.” Biomedical optics express,
7(8), 3007-3020, 2016
[14] American Heart Association News. Nearly half of U.S. adults could now be classified with high blood pressure, under new definitions. Published
November 13, 2017.
[15] Hassan, M. A., Aamir Saeed Malik, David Fo_, Naufal Saad, B. Karas_, Yasir Salih Ali, and Fabrice M_eriaudeau. "Heart rate estimation using
facial video: A review." Biomedical Signal Processing and Control,No.38, pages: 346-360, 2017.

Volume VI, Issue IV, April/2019 Page No:1067