International Conference on Computer, Communication and Electrical Technology ICCCET2011, 18th & 19th March, 2011

WAVELET BASED PULSE RATE AND BLOOD

PRESSURE ESTIMATION SYSTEM FROM ECG
AND PPG SIGNALS
Ashima Sahoo

P.Manimegalai

Dr.K.Thanushkodi

Electronics and Instrumentation

Karunya University
Coimbatore, India
ashimabmi@gmail.com

Electronics and Instrumentation

Karunya University
Coimbatore, India
manimegalai.vairavan@gmail.com

Electronics and electrical

Akshaya College of Engg. &Tech.
Coimbatore, India

Abstract Conventional blood pressure meters are based on cuff

and are considered being inconvenient for daily monitoring and
these are very sensitive to artifacts due to the presence of cuff. By
having a cuff, the efficiency of the device will be reduced in terms
of power consumption, restriction of frequency and also ease of
use. This paper aimed at designing a noninvasive cuff-less blood
pressure and pulse rate estimation system based on pulse transit
time (PTT) technique using wavelets. This paper presents a novel
algorithm and its real time implementation details for denoising
ECG and PPG signals which are highly non stationary signals
and finding out pulse rate and PTT using discrete wavelet
transform (DWT) in MATLAB and LabVIEW. Hence two
important parameters of cardiology i.e. Pulse rate and Blood
pressure can be easily measured at a time from ECG and PPG
signals that can continuously monitor the health of an individual
with a cardiac disorder.
Keywords- ECG, PPG, DWT, PTT, REALTIME, LabVIEW,
MATLAB

I.

when the pressure in the arterial circulation falls to its lowest

level. Blood Pressure measurement techniques are basically
put into two classes: direct and indirect. [1] The direct method
can obtain BP continuously and accurately, but it is hard to set
and in the danger of infection. The indirect techniques are
non-invasive, with improved patient comfort and safety, but
needs cuff at the expense of accuracy. They measure the
pressure only at a single point of time and are not accurate
enough. In 1922, Bazzett had found that pulse wave velocity
(PWV) is related not only to BP, but also to the cubage and
flexibility of the arteries [2]. Blood pressure (BP) is the
function of cardiac output which is the amount of blood
volume output per cycle. BP is directly proportional to the rate
at which the blood travels in arteries and flow rate is
calculated from pulse wave velocity. Thus, PWV could be a
useful and convenient parameter for continuous monitoring of
calculating blood pressure. However, it is hard to measure
PWV in clinic. So people replace PWV by pulse transit time
(PTT) and thus describe the relation to BP. In 1957, Lansdown
put forward that the relation between PTT and BP in
physiology is linear, and the relation is likely steady for
individual during certain time [3]. Generally speaking, if BP
increases, PTT decreases and vice versa. As pulse rate is
approximately same as that of heart rate, it can be measured
from PPG signal which is a simpler process than measuring
heart rate from ECG which has more complexity in electrode
placement.
The Biokit Physiograph System is used for capturing ECG
and PPG signals from individual persons which are highly non
stationary biomedical signals. The signals are captured at 1000
samples/second/ channel. This paper focuses on the initial
work to develop a pulse rate monitoring system and measure
PTT by MATLAB and Based on the results, systolic blood
pressure (SBP) is coarsely estimated. For real time application
work can be done in LabVIEW. LabVIEW (Laboratory
Virtual Instrumentation Engineering Workbench) is a platform
and development environment for a visual programming
language from National Instruments [5]. LabVIEW helps
create flexible and scalable design, control, and test
applications. With the help of LabVIEW, we can interface

INTRODUCTION

In modern world, demands to improve living styles causes

most people not to really concerned about their healthiness
however since the awareness of high blood pressure is the
biggest known cause of disability and premature death through
stroke, heart attack and heart disease, medical doctor
recommended a regular self monitoring of blood pressure and
pulse rate to make sure of the necessassity to control blood
pressure and prevent it from taking the shape of either
hypertension or hypotension. Blood pressure is the
measurement of the force applied to the blood vessels during
blood circulation which decreases as it moves away from the
heart through arteries and capillaries, and toward the heart
through veins. It represents one of the principal vital signs
often measured and the most intensively studied parameter in
medical and physiological practice.
For each heartbeat, blood pressure measurement varies
between systolic and diastolic pressures. The highest pressure
occurs when blood is travels through the arterial circulation by
the contraction of the heart which known as the 'systolic' blood
pressure (SBP), while 'diastolic' blood pressure (DBP)
measurement is taken during the heart relaxes between beats

978-1-4244-9394-4/11/$26.00 2011 IEEE

285

International Conference on Computer, Communication and Electrical Technology ICCCET2011, 18th & 19th March, 2011

with real-world signals; analyze data for meaningful

information; and share results through intuitive displays and
reports.
II.

Wavelet theory can be understood and developed only by

using such digital filters. This is the meeting
point
between wavelets and sub band coding and the origin
of two different nomenclatures for the same concepts. In
fact, wavelet transform and sub band coding are so closely
connected that both terms are often used interchangeably.
Filter banks are structures that allow a signal to be
decomposed into sub signals through digital filters, typically
at a lower sampling F ig. 2 show a two-band filter bank.

PULSE TRANSIT TIME

Pulse transit time is the time interval for the arterial pulse
pressure wave to travel from the aortic valve where it is
ejected from the left ventricle to a peripheral site. This
peripheral site can be anywhere along the brachial artery
where the pulse can be felt the most. The brachial artery is the
major blood vessel of the upper arm, so for this reason it is the
artery that gets occluded when measuring blood pressure using
the traditional mercury sphygmomanometer. PTT is the time
interval between two pulses detected on the same artery.
Usually, PTT is defined as the time interval between two
characteristic points the R peak of the electrocardiogram
(ECG) and the peak of the pulse at finger, ear lobe, or toe [4].
So PTT is measured by monitoring ECG and Pulse wave in
this study. To determine the time at which the blood leaves the
aortic valve an ECG is used. R peak is used as a marker for the
time when the blood leaves the aortic valve. With the increase
in blood velocity BP increases which means blood reaches the
peripheral site from the aortic valve in a smaller time that
means transit time is less and decrease in BP corresponds to
longer PTT.

Fig.2 One-level two b and perfect reconstruction filter bank

The DWT analyses the signal at different resolution

(multiresolution) through the decomposition of the signal into
several successive frequency bands. The DWT utilizes two set
of functions (t) and (t), each associated with the low pass
and the high pass filters respectively [9].These functions have
a property that they can be obtained as the weighted sum of
the scaled (Dilated) and shifted version of the scaling function
itself:

--- (1)

Fig.1 The definition of PTT

III.

--- (2)

METHODOLOGY
Here, h[n] and g[n] are the half band low pass filter and high
pass filter respectively.

A. Wavelets
Wavelets are a powerful tool for the representation and
analysis of such physiologic waveforms because a wavelet
has finite duration (compact support) as

B. Block diagram
The ECG and PPG signals are taken from individual
persons using Biokit at a sample rate of 1000 samples/second.
The frequency response of the ECG signal shows its main
concentration in the lower frequency range 2- 40 Hz and for
PPG it is 0.05-10Hz. After removal of noise using DWT, the
peaks of the QRS complex with its high dominated amplitude
in the signal is detected. Similarly the peak point of the
peripheral pulse is detected. From the peaks of the PPG, the
pulse rate is estimated. Pulse rate = 300/number of large

contrasted with Fourier methods based on sinusoids of

infinite duration. A main goal of wavelet research is to
create a set of expansion functions and transforms that give
informative, efficient, and useful description of a function or
signal. In applications working on discrete signals, one never
has to directly deal with expansion functions. Discrete
wavelet transform (DWT) is obtained simply by passing a
discrete signal through a filter bank.

286

International Conference on Computer, Communication and Electrical Technology ICCCET2011, 18th & 19th March, 2011

squares between successive P waves. The normal pulse rate

varies within the range 60-120 beats/min Then PTT is
calculated from the R peak of ECG to the characteristic peak
point of PPG. After calculation of PTT regression analysis is
done in MATLAB to calculate blood pressure.

Fig.4 A4, A6 Approximations and D1, D2, D3, D4, D5, D6 Details

Fig.3 Block diagram for blood pressure and pulse rate estimation

C. Wavelet approximation coefficient method

Wavelet transform, especially the orthogonal wavelet
transform has the character of strong decorrelation, which can
make the transformed signal energy concentrate in the larger
wavelet coefficients in the wavelet domain, and the energy of
noise distribute in the whole wavelet domain. Therefore, the
wavelet coefficients of useful signal is larger than that of noise
through the wavelet decomposition, and then the wavelet
transform can be regarded as a filter processing.
The ECG and PPG signals were decomposed into 8 levels
by discrete wavelet transform (DWT) using daubechies 6
(db6) wavelet. The level 4 and 8 approximations are
reconstructed from the approximation coefficients. Since most
of the noise lie in high frequency detail coefficients,
discarding these detail coefficients and reconstructing the
signal from appropriate approximation coefficients will give a
denoised signal. By subtracting A8 level from A4 level and
cubing the result QRS complex can be obtained and in this
process the base line drift is also removed. Similarly A6
approximation is reconstructed for PPG signal for getting a
denoised and baseline corrected signal.
After preprocessing the signals, corresponding peaks of the
signals i.e. R peak and pulse peak are estimated and pulse rate
is calculated from the peaks of the PPG signals. Using the
difference in index values of the peak onset of the two signals,
the pulse transit time is calculated which is inversely
proportional to BP.

Fig.5 subtraction of various approximation levels to obtain QRS complex

In discarding all the high-frequency information, we've also

lost many of the original signal's sharpest features. Optimal
de-noising requires a more subtle approach called
thresholding. This involves discarding only the portion of the
details that exceeds a certain limit.
D. Wavelet thresholding and multiresolution method
Denoising and baseline drift removal:
Wavelet thresholding de-noising methods deals with
wavelet coefficients using a suitable chosen threshold value in
advance. The wavelet coefficients at different scales could be
obtained by taking DWT of the noisy signal. Normally, those
wavelet coefficients with smaller magnitudes than the preset
threshold are caused by the noise and are replaced by zero,
and the others with larger magnitudes than the preset threshold
are caused by original signal mainly and kept (hardthresholding case) or shrunk (the soft-thresholding case). Then
the denoised signal could be reconstructed from the resulting
wavelet coefficients. These methods are simple and easy to be
used in de-noising of ECG signal. But hard thresholding denoising method may lead to the oscillation of the reconstructed
ECG signal and the soft thresholding de-noising method may
reduce the amplitudes of ECG waveforms, and especially
reduce the amplitudes of the R waves. To get better result an
improved thresholding de-noising method can be used which
is between hard and soft thresholding. Similarly in PPG Low
frequencies (which constitute baseline wander) appear at high
scales (details D6 and above). Removal of such details

287

International Conference on Computer, Communication and Electrical Technology ICCCET2011, 18th & 19th March, 2011

Where, E0 is Youngs Modulus of arterial wall at zero

pressure, h is the brachial artery wall thickness, d is inner
radius of brachial artery, is the blood density, L is distance
of heart from fingertip and PTT is the pulse transit time.[8]

corrects baseline wander. High frequency burst of noise is

captured at the smallest scales (which correspond to high
frequencies), namely D1 and D2. Removals of such details
correct the noise.

IV.

RESULTS AND ANALYSIS

In this paper wavelet approximation and thresholding, both

methods are performed and compared. Discarding all the highfrequency information in wavelet approximation method,
many of the original signal's sharpest features may get lost
hence wavelet thresholding method proved to be a better one
which involves discarding only the portion of the details that
exceeds a certain limit.
The simulation results from MATLAB are shown in
Fig.6,Fig.7,Fig.8, Fig.9,Fig.10 and Fig.11.After denoising and
baseline correction, R peaks and pulse peaks are detected and
then PTT is calculated. Since PTT is different for each
individual,the physiologic parameters from 5 healthy people
are collected from Biokit. We tested our system and pulse rate
is estimated from the PPG signal which is approximately same
as heart rate calculated from ECG and which can be clearly
seen in Fig.10. SBP can be calculated from PTT using
regression analysis in Matlab. The calculated SBP is compared
with measured automatic BP meter and is shown in Fig.11.

Fig.6 denoised and baseline correct ECG and PPG signals using wavelet
thresholding

In this method db6 wavelet is used and the ECG and PPG
signals are decomposed into 8 levels. As a result approximate
coefficients and detail coefficients are obtained. Then hard and
soft thresholding is applied to the detailed coefficients. After
then reconstruction of the de-noised signal is done from detail
and approximation coefficients by inverse discrete wavelet
transform to get a denoised signal which is then subtracted
from original signal to get baseline correct signal

TABLE 1. THE MEASURED AND ESTIMATED HEART RATE, PULSE

RATE AND SBP.
SL.
HEART
PULSE
PTT
SBP
SBP
NO
RATE
RATE
(calculated)
(measured)
(ms)
(bpm)
(bpm)

R Peaks and Pulse peak estimation:

Peaks of the R waves have the largest amplitudes. In
order to detect the peaks, specific details of the signal
were selected. Details 2^3 2^5 were kept and all the details
were removed. This procedure r e m o ve d low frequencies
and h i g h frequencies. The attained signal samples were
then squared. High a m p l i t u d e transitions of the signal were
then more noticeable, even if R peaks are deformed. Then a
practically lower limit is applied on the signal to remove
unrelated noisy peaks. Because no subsequent beats
happen less than .25 second, pseudo-beats are also removed.
The two frequencies that constitute the PPG signal complex
can be extracted from details D3, D4 and D5. The signal
obtained by retaining these details was appropriately cubed to
enhance peaks which ensured better detection.

76

75

122

127

120

67

63

130

122

117

78

80

79

153

144

93

95

98

141

135

94

96

93

144

133

E. PTT Calculation
PTT is defined as the time interval between the R wave
peak of ECG and the peak of PPG in the same cardiac cycle.
With sampling rate known, using the difference in index
values of the peak onset of the two signals, we are able to
calculate the Pulse transit time. BP increases with decrease in
PTT and vice- versa.
Blood pressure is calculated by using the following formula:
Equation:

Fig.7 (a) ECG signal, (b) baseline correct signal, (c) denoised signal,
(d) ECG peaks with estimated Heart Rate=76beats/min

--- (3)

288

International Conference on Computer, Communication and Electrical Technology ICCCET2011, 18th & 19th March, 2011

V.

CONCLUSION

In this paper, Electrocardiogram (ECG) and Peripheral Pulse

(PPG) signals are processed using Discrete Wavelet
Transform. The Wavelet based signal processing allows taking
advantage of characteristics of time-scale analysis. This
wavelet approach proved to be an efficient method for ECG
and PPG signal processing and the pulse rate is accurately
obtained from PPG signal in MATLAB. PTT is a
physiological parameter which can be obtained from ECG and
PPG signals that indicates the change of BP, and is inversely
proportional to pulse wave velocity and BP. So it is possible to
monitor BP continuously by PTT. Hence pulse rate and BP
can be measured at one time continuously. Although BP is a
function of PTT which is acceptable, more study is required to
increase the accuracy.
To make the system more accurate and to implement in real
time, signals are processed and initial works has done in
LabVIEW for better results in future.

Fig.8 (a) PPG signal, (b) baseline correct signal, (c) denoised
signal, (d) PPG peaks with estimated Pulse Rate= 75beats/min

REFERENCES
[1]

Yuan-ting Zhang, Blood Pressure Devices, IEEE/EMBS International

Summer School on Medical Devices and Biosensors (ISS- MD), 2004.
[2] Bazzet HC, Dfeyer NB. Measurement of pulse wave velocity [J]
.Am.J.Physiol, 1922, 63:94-115
[3] Lansdown M. A method of using induced waves to study pressure
Propagation in human arteries, [J]. Circ.Res, 1957, 5:594-601
[4] GU-Young Jeong, Kee-Ho Yu, and Nan-Gyun Kim, Continues Blood
Pressure Monitoring using PulseWareTransit Time, ICCAS2005, June
2-5, KINTEX, Gyenonggi-Do, Korea.
[5] Zhang Jun-feng,Wang Yan-yan,Design for high data acquisition system
based on PCL-816 and LabVIEW, Machinery Design
&Manufacture,2007 Vol7, pp.64-66
[6] Lass J, Meigas K, Karai D, Kattai R, Kaik J, Rossmann M. Continuous
Blood Pressure Monitoring During Exercise Using Pulse Wave Transit
Time Measurement. Proceedings of the 26th Annual International
Conference of the IEEE EMBS 2004; September: 2239-2242.
[7]
Meir Nitzan, Amikam Patron, Zehava Glik, Abraham T Weiss,
Automatic noninvasive measurement of systolic blood pressure using
Photoplethysmography, biomedical Engineering Online, 2009
[8] Kalju Meigas, Rain Kattai and Jaanus Lass, Continuous Blood Pressure
Monitoring Using Pulse Wave Delay
[9] J.Y.A. Foo,, S.J. Wilson, G.R. Williams, M.A. Harris, D.M. Cooper,
Measurement of pulse transit time using AT90S8535 microcontroller,
Elsevier Measurement Section, 2006
[10] Leslie Cromwell , Fred J. Weibell, Erich A. Pfeiffer, Biomedical
instrumentation and Measurements, Second Edition, Prentice hall of
India, New Delhi, 1997
[11] Andras Mersich, A k os Jobbagy, A advanced Blood Pressure
Measurement at Home
[12] D .J. Pitson, A. Sandell , R. v a den Hout , J.R. Stradling, Use of
Pulse transit time as a measure of inspiratory effort in patients
With obstructive sleep apnoea, ERS Journals Limited, 1995
[13] R. S. Khandpur, Handbook of Biomedical Instrumentation, 2nd
ed.NewDelhi: Tata McGraw-Hill Publishing Company Limited, 2003,
pp. 113.
[14] Dai Junwei, Wang Boliang, A New Design of PPG Pulse Sensor and
Noise Analysis, Modern Electronics Technique, 2006, Vol.2, pp.78-80.
[15] Jin Yinbin, Yu Youli, Jin Jie, Huang Yecho, Development on Pacing
ECG Monitoring System, proceeding of the 20th Annual nternational
Conference of the IEEE Engineering in Medicine and Biology Society,
Vol.20, No1, 1998

Fig.9 Estimated PTT=122ms and SBP=127.

Fig.10 Heart rate and pulse rate w.r.t subjects

Fig.11.SBP by automatic and by PTT w.r.t subjects

289