You are on page 1of 6

Photoplethysmographic Based Heart Rate Variability

for Different Physiological Conditions


Mansoor Hussain Shah, Syed Absar Kazmi, Khairul Azami Sidek, Sheroz Khan and Fatema-tuz-Zohra Iqbal
Department of Electrical and Computer Engineering
International Islamic University Malaysia
Kuala Lumpur, Malaysia
engrmansoorshah@gmail.com

Abstract-This

paper

investigates

the

feasibility

of

using

photoplethysmographic (PPG) signal for heart rate variability


(HRV) using different physiological conditions. In this paper, we
have analyzed four physiological conditions i.e.; sitting, standing,
laying and jogging. The Easy Pulse sensor module was used to
pass the signal sensed by the optical sensor through a series of
high and low pass filters which later generates a conditioned PPG
signal at its output. The Arduino processing module was used for
the digitization and processing of the PPG signals. The Arduino

Pulse oximetry uses PPG method of blood oxygen


saturation measurement which is based on the idea of light
absorption variation measurements according to the cardiac
induced blood changes [4].
The working principle of pulse oximetery is that the
concerned body part is subjected to the red and infra-red light
rays. As the result, the following outputs are achieved [5].

program was then developed to capture the PPG data and


waveforms. The Kubios HRV software was used to process the

i)

Red rays are reflected back to the photodetector


placed on the same side after very little absorption of
light as hemoglobin of the blood absorbs light rays in
the concerned body part. The rest of the light is
reflected back as shown in Fig.1 (a).

ii)

Infra-red rays of the light from the source are


penetrated through the body part (such as finger and
earlobes) which reaches to the photodetector placed
on the other side for detection purposes as shown in
Fig. l(b).

PPG data and manipulate it into a HRV format. The PPG signal
was then analyzed in time and frequency domain parameters.
Later, report sheets were generated based on these analyses. The
results showed that the PPG signal as well as HRV changes
depending

upon

the

physiological

conditions.

It

was

also

examined that the low and high frequency components of PPG


signal also varied

according

to the change in

physiological

condition.

Keywords-PPG; HRV; optical sensor; Kubios HRV.


I.

INTRODUCTION

In biomedical science, signal detection has become


extensively important as they are used to assess the health of
the lungs, heart and other parts of body under observation.
Blood oxygen measurement in this regards also plays a very
important role as it is carried and delivered to different parts of
the body. Besides that, non-invasive methods using
photoplethysmography (PPG) signals have also gained
significant attention by researchers. [t is an electro-optical
technique used to measure physiological parameters non
invasively such as heart rate, respiratory-induced intensity
variations (RIIV), blood oxygen saturation levels and arterial
stiffness [1].
A wide variety of diseases like high blood pressure,
diabetes, arrhythmia and detections of the health disorders can
be assessed by monitoring, diagnostics and an array of
assessment tools. The assessment of cardiovascular health is
being monitored in clinical routine using PPO signal. Another
non-invasive quantitative marker towards the cardiovascular
health is HRV which can be easily extracted from the
acquisition of PPG and ECG signals [2].

978-1-4799-6428-4/14/$31.00 2014 IEEE

(a)

(b)

Fig.l.PPG sensor heads (a) Reflectance type (b) Transmission type[6].

During the systole process, blood volume is increased in


the arteries which results in the decrease of the amount of light
passing through the body tissues. On the other hand, in the
end-diastole process, the amount of light passing through the
body tissues and reaching to the detector is increased. [t
results in producing the maximum output during the end
diastole and minimum output during the systole process
[7].This is obvious from the waveform during the systole and
end-diastole processes shown in Fig. 3 below.

Reference
No.

end-diaslole

[2]

Discussed Issue

Transmission!
Reception of the
PPG signal through
a wireless device to
examineHRV

BL
BL-AM

13J
TIME

Fig. 2. A PPG signal over time. The base line is the maximum of the
pulse,and AM is the pulse amplitude [7].

In order to make the analysis simplified, the general


representation of the PPO signal is inverted such that increase
in the arterial blood (during the systole process) should
correspond to the increase in the PPO signal and vice versa as
shown in Fig. 3.

[4]

PPG Pulses

15J

[6]
o

2
TIMES

..

(s)

Fig. 3. The inverted PPG signal. The increase in PPG signal corresponds to
increase in tissue blood volume [7].

The PPO signal spectras are different for the person who is
normal and the one which is a cardiovascular patient. PPO
signal can also be utilized to analyze HRV [8].

[7]

Along with other several advantages of the PPO it is


simple, inexpensive and easy to do maintenance works. The
compactness and portability has increased its important up to
such an extent where it can be used in all types of
environments [8].
TABLE I
Reference
No.

[I]

RELATED LITERATURE REVIEW.

Discussed Issue

PPG measurements
focusing on
assessment of the
cardiovascular
systems.

Merits

It gives concepts
of the
measurement of
the oxygen
saturation and the
detection of the
disease using PPG
signal

[8]

Deficiency

Lack of
establishing
comprehensive
normative data
ranges for
comparison
with patients.

19J

Examining the
coherence of the
pulse rate
variability (PRV)
obtained from the
PPG andHRV
obtained from EeG
for a subject before
and after the
exercise

Merits

A very good idea


towards PPG
signal
transmissionirecep
tion to a far place
for monitoring or
other purposes.

The two PRY and


HRV are
coherence for a
subject in rest
conditions

Deficiency

The paper did


not cover the
effect of
physiological
condition on
PPG signal.
The behavior of
HRV on the
basis of PPG
signal for
diflerent
physiological
conditions was
not examined.

Pulse oximetry and


PPG signal
acquisition using
non-lllvaSlve
technique

A very good
explanation of the
PPG signal and its
acquisition.

The research
never covered
the variation of
PPG signal
according to
physiological
conditions.

Development of the
pulse oximeter and
the steps in getting
the PPG signal in
the industrial and
medical
app lications

Very useful to
understand the
basics of pulse
oximetry

The author gave


a very basic
idea about pulse
oximetry and
PPG signal.

Development of the
model for the
computation of the
SPO, and the
expression of the
attenuation of the
light sensor used for
the detection of the
PPG signal.

Model can be
utilized to derive
the values of the
patient regardless
of their conditions
and sensor related
parameters

The values
acquired cannot
be set as the
average
standard values
for patients
with the
different
physical
conditions.

Technique can be
used to measure
venous blood
oxygen
measurements

Determination
of the oxygen
saturation of the
blood cannot be
measured
through the
tissues under
examination

PPG signal signals


were of different
shape and values
for the normal
healthy person
than that of
cardiovascular
disordered person.

Research does
not cover the
effect of
physiological
conditions on
PPG signal.

Suggested the
prolongation in the
light path due to
scattering effect
through the tissues
produces
comparatively
higher values of
PPG signals
PPG signal for
cardiovascular
patients in the
supplement ofHRV
signal.

The compare the


accuracy of Peak to
Peak Intervals (PPI)
and Valley to
Valley Intervals
(VVI) from PPG in
sitting conditions
forHRV

VVI forHRV is
more accurate and
reliable then PPI

It did not
compare the
PPG signal and
HRV for
different
physiological
conditions.

Reference
No.

Discussed Issue

Invented to extract
unique features of
PPG signal on the
bases of which the
subject can be
discriminated from
others.

[101

Merits

It is due important
for biometric
recognition.

Deficiency

This paper did


not cover the
features of the
PPG signals for
different
physiological
conditions.

A. Easy Pulse Analyzer.

Fig. S shows the window of the Easy Pulse Analyzer.


This is the representation of the wave form of the generated
PPG signal. The window of the Easy Pulse analyzer also
shows three parameters corresponding to the detected PPG
signal. These parameters are,
111-

METHODOLOGY

I.
A.

111-

PPG signal Acquisition and Conditioning

The optical sensing device is placed on the fmger which then


converts physical condition into an electrical signal. In our
experiment, we used HRM-2SIIE optical sensor. This signal is
then fed to the input of the Easy Pulse sensor where it is passed
through a series of low and high pass filter to obtain a
conditioned PPG signal available at its analog output terminal.
B.

Easy PUlse PPG An.IY"or VI . a

103.9 BPM

Signal Processing and Communicating Unit

We used Arduino board as the processing and


communicating device between Easy Pulse unit and the Pc.
An Arduino program was developed to read the data from the
Easy Pulse circuit which is then digitized and displayed in the
form of PPG wave forms and ADC samples data.
We used two graphical user interfaces (GUI) i.e. the Easy
Pulse analyzer to display waveforms for a specific PPG signal
and Cool Term software to display data on the PC. The block
diagram of the process is shown in Fig. 4.

Easy Pulse
/

Sample time (which is Sms)


Range of ADC samples (which is Sms)
Blood Pressure Measurement (which is currently
103.9)

sensor

1&

Mod.

Arduino
PC,wave
----JIo Processing ----JIo shap e, data
Board

Kubios

Fig. 4. Block diagram of the data acquisition and processing.

Finally, we fed these PPG data to the Kubios HRV software


to generate HRV analysis, mean RR interval, and low and high
frequency components.
II.

EXPERIMENT AND RESlJLTS

We used two main modules to perform our experiments


that are Easy Pulse sensor module and Arduino processing
board. Four volunteers were involved where we managed to
get their PPG signals in four different physiological conditions
which consist of sitting, standing, laying and jogging. We used
three majors graphical user interface (GUI) based softwares to
receive waveform, capture the data and to analyze the results
which are Easy Pulse analyzer, Cool Term software and
Kubios HRV software.

Fig. 5. Easy Pulse displaying waveform window


B.

Cool Term

This software was used to get PPG data by connecting


the output of the Arduino board with one of the
communication port of the PC. Fig. 6 shows the window
of this software. In this example, the PPG data was
obtained for one minute duration for each physiological
condition per subject.
-.

...._

...._Hot.I
..
..
_

C<_ I I v!. l

401
496
50S
484
44S
419
415
416
419
425
436
446
44S
439

C/_II;-N-l.
C...-.erled(lOOOcl.

tx
GolD:

IR"S
EoII OTR
.C'fSeOSR

DCO
.R:II

Fig. 6. CoolTerm displaying data window

C.

Kubios HRV

When we fed these data for different physiological


conditions of the volunteers to the Kubios HRV, it generated
time and frequency domain analysis. It also generated low
frequency (LF), high frequency (HF) components and HRV
for different physiological conditions. The same process was
applied to the PPG data acquired from each volunteer and for
four different physiological conditions (sitting, standing,

laying and jogging). The following results were obtained as


in Table 2 and the subject's details are shown in Table 3.
Kubios HRV generated the following report for PPG signal
for physiological conditions.

Time-Domain lesults
VarBbe

Mean RR"
STO RR (SONN)
Mean HR"
STOHR

RMSSO
NN"
pNNOO

Distributions"

Units

Vau
l e

(m
(m
(lImi
(lImi

717.7
257.7
192.37
&0.34
;!;7.3
3.
40D
1.536
-1200.0

(m
(count)

RRlrBllJUBI irrl3x

liNN

(m,

DO
RR)

Frequency-Domain lesults
m IIK"-II (I'l'el::'s p3rb:lcgram: 2569 wind:rllwnh 5OJ.Q\ISIEp)
Time-Domain lesults
Units

Value

Im
(o
(1Ioi
(lloi
(.

194.1

V,rBb.
Magn RR'
STO RR (SONN)

!.lEOn HR'
STDHA
AMSSO
NNill

19'3.5
85.02

53.52
',J1l.l

(COUlrt)

pNN'"

25

32.5

RRlrsr{luBIird:9:

1J4,

(o

liNN

-1365.0

[] D
)

HA ,,",1SIoi

AA

Frequency-Domain IesuHs
m (W9k:bBp3fi:dOJr.3m:25Sswimwrrb50%i8rBp)

AR

t'-Amco.'orc!;I.16,iDllacbl.,d)

0.4

.-:r

s
@

"

0.2

@ 0.2

"
0

0.2

01

Q.

.<$.

0.1

Fl9:juen::y

Illrd

VlF (O-OD4Hz)
LF (0.0+-0.15 Hz)
HF."-{).4Hz)

10ia1

0.'

0.1

P .. k

lWoe

(Hz)

(m5\

Q,COg1
0.1289
0,m17

136
12119
19107
31353
o.S34

oJ

02

0.4

0.'

Frraquercy (Hz)

lWoe
('10)
0.4
38.6

00.9

Po..,

FfI:o:juelq

(n.u)

Bard

3B.8
612

VLF(O-o.04Hz)
LF (O,04-0.15Hz)
HF."-{).4Hz)

10ia1

POlk
(Hz)

IW"

003g1
Q.14e4
0.1641

e;!J

(m5\
6111
21002
287<
0.179

LHF

FUNsr
('10)
1.4
21.3
IB.3

p"""
(n.u.)
21 ,
7B.2

Fig. 9. Subject.3 KubiosHRV report sheet for laying condition

0.1

01

/\, I.-

oj

0.2

04

0.1

0.5

0.5
F"'Iu'IIHz)

F"'IU'1IH

PEek
1Hz)

IW"
(m5\

IW"
1'

Po""
In.u)

VLF(0-004 Hz)
LF .04-o.I5 Hz)

0.((191
0.1445

15
3213

02
36.5

366

HF.I5-{).4Hz)

0.2852

1!i02

63.3

63.4

FU8n::

Illrd

10ia1

LFIHF

F U9

rtr

p..,
1Hz)

IW"

lWei

P"""

Im5\

1'

In.u.)

VLF(O-O.04Hz)
LF .04-0.15 Hz)

0((191
0.1484

1 .5
14.B

15.1

HF.I5-{).4Hz)

0.1719

I iii
l ill5
141

1l.6

84B

Bard

B8",

10ia1

0.577

WHF

Time-Domain IlesuHs
VarEt.

Valu,
467.3
1.5

1m)
1m)

MOlnAA'

111mi
111mi

M."HA'

SID HA
RMSSD
NNill
pNNill

395.61
112095
197.9

Ims)
lcourrt)
1'

13

679
10.154

RRtrilrguBrim

liNN

-1110.0

Ims)

10146

II

AR

,
.-I

"

"

s
Cl 0, 5
00

...--r1

/\1\

0.2

oj

04

...

Il.

Amoillorc!;I.16,iDllacb""d)

Vau
l e

(m
(m
(1lmir\l
(1Imif\l
(m

004.B7

5B2.6
301.2

5BB4JJ
Jro.6
70
66.7
3o<ll
-1200.0

(court)
N
(m

D0
RR)

AR

HR lb=atsimirV

'"

(A,Rmocl3lordir.16,rotlactJrized)

FlE<1uerty

P.,

IW"

Illrd

I Ilz)

I.s

0.2

0.3

0.4

0.5

F"'Iu'jIHz)
lWer

1'

Po""

In.u)

Fre:juarr:y

Bard

0.0234
0.1484

143
3355

0.4

BJ

8.3

VlF (O-o.D4Ilz)
lF.04-o.l5Hz)

0.1875

311157
40455

91.4

91.7

HF.I5-{).4Hz)

0.031

0.1

lotal

lRHF

lWer

0((191
0.1414

340
3519

0.8
1.1

1.1

0.1797

36310
401111

lilA

912

oIiJl

Fig. 8. Subject.3 KubiosHRV report sheet for standing condition.

PaNel

P"""

p..,k
IIlz)

m
I s

005 0,1 0.15 02 0.25 03 035 0.4 0.45 0.5


Frraquerq(Hz,l
Frraquercy

Ill,"

VLF(0-004 Hz)
LF (0.04--0.15 Hz)
HF (0.15-{).4Hz)

10ia1

LFIHF

POlk
(Hz)

IW"

0.0195
0.1405
o.ZleS

1970
52710
::OB27s
rus5
0.301

(m5\

1W9I
('10)
6.B
21.5
71.5

Po""

0.2

0.1
Frequen::y

(n.a'

Ba,"

.I
769

VLF ,,"0.04Hz)
LF (0.04-0.1 5Hz)
HF.15-{).4Hz)

10ia1

LHF

0.3
Frequert:y(Hz)

POlk

lWei

(Hz)

(0.\

OJll9 1
0.1484
oJ9B4

53305
IO
72:931
97464s
0.278

0.4

lW ei
('10)
,.5
aJ.6
74.0

0.'

P"""
(n.u.)
21.S
782

Fig. 10. Subject.3 KubiosHRV report sheet for jogging condition.

0
0.5

F"'Iu'jIH

VlF (0-004Hz)
IF .04-o.I5Hz)
HF.15-{).4Ilz)

Distributions'

Units

Frequency-Domain lesults
FR ryiel::h'S p3100cgram: 2565 wimbHwnh 5OJ. Cll'erBp)

s
Cl 0,5
00

01

RMSSO
NN",
pNNEO
liNN

HA ,,",1SIml

AA )

,
.-I

M93n RR'
STO RR (SONN)
Me:en HR'

RRtrarguarim

0.177

WD

Frequency-Domain IlesuHs
FFr [lisch' pali:dQJ.m: 156..inciJN.15(JJ, ",,,Ilp)

VarBbB

Distnbions'

Un

SID AA ISDNN)

Time-Domain lesults

STOHR

Fig. 7. Subject.3 KubiosHRV report sheet for sitting condition

lFIHF

04

005
0
0

lotal

oj

Frequercy(Hz,l

0.15

a
00

fARmocl3lorri;lra 16,rotiactJrilsd)

0.4

.-:r

LFIHF

015

AA

Distnbulions'

1'

In.u.)

Fig. 7 to 10 represent the Kubios HRV software windows,


where the bar graphs showing the Respiratory Rate (RR) and
Heart Rate (HR) for each PPG signal in different physiological
conditions. The Area graphs in each figure (Fig.7 to 10)
represent the Power Spectral Density (PSD) estimation and
Autoregressive (AR) estimation for each PPG signal in all four
different physiological conditions, however our main focus in
this research are the values of HRV and LF/HF components,
hence, in this paper we will limit our concern to these values
for four physiological conditions (sitting, standing, laying and
jogging) for different subjects.

TABLE 2

SRNO

LFiHF & HRV VALUES

SITTING

STANDING

LAYING

CONDITIONS

CONDITIONS

CONDITIONS

JOGGING CONDITIONS

LFiHF

HRV

LFiHF

HRV

LFiHF

HRV

LFiHF

HRV

Subject. 1

0.22

227.59

0.04

353.79

0.007

148.5

0.24

967.39

Subject. 2

0.06

717.39

0.24

361.74

0.04

371.95

0.09

235.95

Subject. 3

0.577

85.02

0.091

359.62

0.634

192.37

0.301

5884.33

Subject. 4

0.224

143.00

4.180

150.85

0.290

852.13

0.372

267.94

Subject 5

0.469

420.83

0.073

162.68

0.099

206.33

0.099

487.78

Subject 6

0.426

562.51

2.065

758.65

0.296

602.80

0.022

617.60

Subject 7

0.092

313.78

0.063

432.49

0.059

254.86

0.447

271.63

Subject 8

0.284

190.79

0.375

522.63

0.395

109.51

0,130

201.75

Subject 9

0.072

166.1

0.403

232.16

0.108

369.85

0.402

370.26

Subject 10

0.606

166.13

0.753

218.77

0.130

186.20

0.355

129.04

TABLE 3
SRNO.

GENDER

SUBJECT DETAILS

AGE

REGION

PHYSIOLOGICAL CONDITION

Subject. 1

Male

21

South Indian

All four states

Subject. 2

Male

24

South Indian

All four states

Subject. 3

Male

25

South Indian

All four states

Subject. 4

Male

26

Bangladesh

All four states

Subject 5

Male

27

Afghanistan

All four states

Subject 6

Female

23

Malaysian

All four states

Subject 7

Female

23

Malaysian

All four states

Subject. 8

Female

23

Malaysian

All four states

Subject. 9

Female

20

Malaysian

All four states

Subject. 10

Female

23

Malaysian

All four states

CONCLUSION

[n this research the PPG signal was analyzed in terms of


HRV using Kubios HRV software for four different
physiological conditions (sitting, standing, laying and jogging)
for different male and female volunteers. Table 2 represents
the values of the LF/HF components and HRV for the 10
volunteers (5 males and 5 females) within the age group of 2030 years. The values of HRV obtained from the PPG signal for
different physiological conditions conclude that the HRV
increases whenever the transition of the physiological
condition changes from sympathetic to parasympathetic
condition and vice versa. [t is also to be concluded that the LF
component representing sympathetic condition and decreases
during transition towards parasympathetic condition. Likewise
HF component of PPG signal increases and hence LF/HF ratio
decreases as well. The value Table 3 represents the details of
the subjects in terms of their age, gender and region for the
comparison purposes.

REFERENCES

[1] Allen, "Photoplethysmography and its application in


clinical
physiological
measurement,
Physiological
Measurements", pp R[-R39, 2007.
[2] I. Reyes, H. Nazeran, M. Franco, and E. Haltiwager,
"Wireless Photoplethysmographic Device for Heart Rate
Variability Signal Acquisition and Analysis ", 34th
Annual International Conference of the IEEE EMBS, San
Diego, California USA,pp.2092-2095, 2012.
[3] W.H. Lin, D. WU,c. Li, H. Zhang, Y. T. Zhang
"Comparison of Heart Rate Variability from PPG
withThat from ECG". The International Conference on
Health Informatics , IFBME Proceedings 42, Springer
[nternational Publishing Switzerland, pp. 2[3-215, 2013
[4] O. Wieben, "Light absorbance in pulse oximetry," in
"Design of Pulse Oximeters", J.G.Webster, ed., Bristol:
Institute of Physics Publishing, pp. 40-55, [997.
[5] G. D. Baura, "System theory and practical applications of
biomedical signals," in IEEE Press Series on Biomedical
Engineering. New York Wiley-IEEE Press, 2002.
[6] K. A. Reddy, B. George, N. M. Mohan and V. J. Kumar;
"A Novel Calibration-Free Method of Measurement of
Oxygen Saturation in Arterial Blood", IEEE Transaction
on Instrumentation and Measurement, vol. 58, no. 5,
pp.1699-1705,2009.
[7] M. Nitzan and H. Taitelbaum, "The measurement of
oxygen
saturation
in
arterial
blood,"
IEEE
Instrumentation and Measurement Magazine., vol. [[, no.
3, pp. 9-15,2008.
[8] V.S. Murthy ,S. Ramamoorthy, N.Srinivasan, S.Rajagopal
and M. M. Rao, "Analyses of Photoplethysmographic
Signals for Cardiovascular Patients", Proceeding of the

23rd Annual EMBS International Conference, Istanbul,


Turkey, vol.3, pp.2204-2207, 200l.
[9] X. Chen, T. Chen, F. Lou and J. Li," Comparison of
Valley to-Valley and Peak-to-Peak [ntervals from
Photoplethysmographic Signals to obtain Heart Rate
Variability in the Sitting Position", 6 th [nternational
Conference on Biomedical Engineering and Informatics
(BME[), pp.2[4-218, 2013.
[[0] G.R. Salanke, A. Samraj, Meshawari N and S.
Sadhasiwam, "Enhancement in the Design of Biometric
Identification system based on the Photolethysmography
Data", Proceeding of 2013 International Conference on
Green High performance Computing, pp. 1-6, 2013.

You might also like