Content uploaded by Ramon Pallas-Areny
Author content
All content in this area was uploaded by Ramon Pallas-Areny on Apr 14, 2014
Content may be subject to copyright.
Abstract—Electronic bathroom scales are an easy-to-use,
affordable mean to measure physiological parameters in
addition to body w eight. They have been proposed to obtain the
ballistocardiogram (BCG) and derive from it the heart rate,
cardiac output and systolic blood pressure. Therefore, weighing
scales may suit intermittent monitoring in e-health and patient
screening. Scales intended for bioelectrical impedance analysis
(BIA) have also been proposed to estimate the heart rate by
amplifying the pulsatile impedance component superimposed on
the basal impedance. However, electronic weighing scales
cannot easily obtain the BCG from people that have a single leg
neither are bioimpedance measurements between both feet
recommended for people wearing a pacemaker or other
electronic implants, neither for pregnant women.
We propose a method to detect the heart rate (HR) from
bioimpedance measured in a single foot while standing on an
bathroom weighting scale intended for BIA. The electrodes
built in the weighing scale are used to apply a 50 kHz voltage
between the outer electrode pair and to measure the drop in
voltage across the inner electrode pair. The agreement with the
HR simultaneously obtained from the ECG is excellent. We
have also compared the drop in voltage across the waist and the
thorax with that obtained when measuring bioimpedance
between both feet to compare the possible risk of the proposed
method to that of existing BIA scales.
I. INTRODUCTION
EART rate monitoring can be used to predict acute
coronary events including sudden cardiac death,
cardiovascular mortality, development of myocardial
infarction and acute coronary syndrome. Therefore, heart
rate monitoring on patients with or without any ischaemic
heart disease record should be a periodic task, with the same
importance as other risk factors such as high blood pressure
and cholesterol [1]. Heart rate supervision is normally
carried out in health care centers, but the interest in
developing monitoring systems suitable to non-clinical
environments is increasing. The electrocardiogram (ECG)
can be non-invasively recorded by embedding passive or
active electrodes in different household items [2], for
example, bath tubes [3] [4] and showers [5]. The ECG has
also been acquired for a fully dressed person by placing
capacitive electrodes in sheets and pillows in a bed [6] [7],
Manuscript received June 25, 2010. This work was supported by the
Spanish Ministry of Science and Innovation under Grant TEC2009-13022
and the European Fund for Regional Development. Ms. Delia H. Díaz was
supported by CONACYT (México).
The authors are with the Instrumentation, Sensors and Interfaces Group,
Castelldefels School of Technology (EPSC), Universitat Politècnica de
Catalunya (UPC), 08860 Castelldefels (Barcelona), Spain. (e-mail:
ddiaz@eel.upc.edu; jocp@eel.upc.edu; ramon.pallas@upc.edu).
and the heart rate variability (HRV) analyzed [8]. However,
remote ECG detection using these methods is somewhat
inconvenient if the subject has to use them on purpose.
In a recent work [9], the heart rate has been monitored
through impedance plethysmography (IPG) by placing four
electrodes on a chair’s seat and detecting blood volume
changes in the thighs [9] but the signal recorded strongly
depends on body position relative to electrodes.
Bathroom scales are simple to use and inexpensive, and
their use implies a fixed position for the feet although some
body tremor can be present. Electronic scales have been
recently proposed to obtain the ballistocardiogram (BCG)
and derive from it the heart rate [10], cardiac output [11] and
blood pressure [12], but these methods cannot be easily
applied to people that has a single leg. Bathroom scales
intended for bioelectrical impedance analysis (BIA) [13]
have been used to obtain the heart rate too [14] but they
cannot be applied to people with a single leg nor people that
wear electronic medical implants, neither are they
recommended for pregnant women because they inject an ac
current between both feet.
In this work we propose: a) to detect the heart rate by
impedance plethysmography (IPG) in a single foot by plantar
measurements while the subject stands on a bathroom scale,
and b) to measure the drop in voltage across the waist and
across the thorax when injecting current in a single foot as
proposed here or between both feet, as usually done in BIA
scales. This second set of measurements should help in
comparing the possible risk of single-foot versus foot-to-foot
impedance measurements.
II. PROCEDURE AND MEASUREMENT SYSTEM
A. Measurement System
Fig. 1 shows the block diagram of the IPG system
designed. A 50 kHz voltage was applied between two
electrodes (A) in contact with the same foot (10 V) or
different feet (1 V). Voltage detecting electrodes B were
connected to voltage buffers followed by an ac-coupled fully
differential amplifier with 0.5 Hz high-pass cut-off frequency
[15] and an instrumentation amplifier (INA111) with G1 = 5.
A synchronous demodulator based on a switched-gain
amplifier controlled by a square wave signal obtained from
the 50 kHz carrier detected impedance variations, mostly
attributable to cardiovascular activity. This signal was further
amplified (G2 = 300) and band-pass filtered with a second-
Heart Rate Detection from Single-Foot Plantar Bioimpedance
Measurements in a Weighing Scale
Delia H. Díaz, Óscar Casas, Member, IEEE, and Ramon Pallas-Areny, Fellow, IEEE
H
32nd Annual International Conference of the IEEE EMBS
Buenos Aires, Argentina, August 31 - September 4, 2010
978-1-4244-4124-2/10/$25.00 ©2010 IEEE 6489
order active filter (0.5 Hz – 10 Hz) to remove offset and
limit noise bandwidth. Gains and bandwidth were the same
for single-foot and foot-to-foot measurements.
To measure the carrier amplitude across the waist and the
thorax, a 20 V, 50 kHz voltage was applied between either
two plantar electrodes on the same foot or between one
plantar electrode for each foot. The voltage at the output of
the instrumentation amplifier (vcarrier(t), Fig 1.) was measured
with an oscilloscope. G1 was tuned for each subject to obtain
similar amplitudes for vcarrier(t) for all subjects, thus
compensating for the different basal and electrode contact
impedance between subjects.
B. Measurement procedure
A bathroom scale intended for BIA was used that has
multiple-contact dry electrodes for each foot (Fig. 2) instead
of the customary single-contact electrodes (usually two
injecting and two detecting electrodes). Factory connections
between contacts were rearranged to perform four-electrode
plantar bioimpedance measurements on a single foot or
between both feet. Figs. 2a and 2b show the respective
position of injecting (A) and detecting (B) electrodes.
Volume changes at each heart beat result in pulsatile
impedance variations. When measuring in a single foot,
impedance variations can be mostly attributed to volume
changes in foot plantar arteries (Fig. 3). However, foot-to-
foot impedance measurements encompass a larger sample
volume that includes major arteries in both legs and the
abdomen.
The IPG was simultaneously recorded along with the ECG
(lead I) obtained with a custom-made circuit and electrodes
at each wrist. Both signals were sampled at 1 kHz and
recorded with a 12 bit digital scope (DL 750, Yokogawa).
Signals were also digitally processed by MATLAB® version
7.1 and the signal processing toolbox employing Welch’s
method and a Hamming rectangular (rectangular, segment
length 20000, overlap percent: 75).
To measure voltages across the waist and thorax when
applying a 20 V (peak-peak), 50 kHz voltage between
electrodes A in the scale, a respective pair of pre-gelled,
disposable electrodes were placed along the midaxillary line.
In each instance, G1 was tuned until the peak-to-peak voltage
across the waist (lower electrode pair) was about 8 V; then,
the carrier amplitude across the thorax (upper electrode pair)
was measured. Next, the same 20 V were applied to a single
foot and the voltage across the waist and across the thorax
were measured without modifying G1.
Measurements were performed in seven males and three
females whose weight was (77.3 ± 22.7) kg and height
(1.75 ± 0.11) m.
III. EXPERIMENTAL RESULTS AND DISCUSSION
Figs. 4 and 5 show the ECG (top) and IPG (bottom) for
single-foot and foot-to-foot plantar bioimpedance
measurements for two of the ten subjects measured. Results
for the other subjects were similar regardless of gender,
height, complexion and feet position on the bathroom scale.
Plantar arch
Posterior tibial
artery
Lateral plantar
artery Medial plantar
artery
Fig. 3. Foot plantar arteries.
B
A
A
B
A
B
B
A
(a) (b)
Fig. 2. Commercial bathroom weighing scale used for plantar
bioimpedance measurements. (a) Placement electrodes for single foot.
(b) Placement electrodes for both feet. A = Injection electrodes.
B = Detection electrodes.
A mplitude (V)
-0.1 3
-0.0 8
-0.0 3
0.02
0.07
0.12
0.17
0.22
12345678910
Time (s)
0
12345678910
-0.1
-0.0 5
0
0.05
0.1
0.15
0.2
Time (s)
0
A m p litu d e ( V )
Fig. 4. ECG (top) and IPG (bottom) for single-foot plantar impedance
measurements. Markers show the IPG peaks obtained with a software
algorithm.
A
A
B
BHPF
fL = 0.5 Hz
G1 = 5 Synchronous
D em od ula tor BPF
fL = 0.5 Hz
fH = 10 Hz
G2= 300
vcarrier(t)
Fig. 1. Block diagram of the system for plantar bioimpedance measurement.
6490
The baseline impedance was stable provided the subject was
standing still, but low-frequency components attributable to
breathing were more apparent in single-foot than in foot-to-
foot measurements.
IPG waveforms were similar to those obtained by
conventional methods. IPG and ECG peaks could easily be
identified for each heart beat by using a maxima detection
algorithm based on comparisons between adjacent points to
detect peaks and later confirmation of these peaks as maxima
through calculation of differences and their comparison with
a predefined delta value. Hence, volume changes in foot
arteries are large enough to be detected by bioimpedance
measurements.
The agreement between RR time intervals detected from
the ECG and from the single-foot plantar IPG was assessed
through the Bland-Altman test. Mean bias for differences in
RR intervals was 1.33ms and the 95% confidence interval
was ± 30.65 ms (Fig. 6).
Fig. 7 shows the Power Spectral Density (PSD) of the IPG
measured on a single foot (top) and between both feet case
(bottom), in different subjects. Both spectra show
components in the 1-1.3 Hz band that can be attributed to
cardiac activity and components in the 0.2-0.4 Hz band that
may be related to respiration.
Table I shows the mean and standard deviation of the
peak-to-peak amplitude of the carrier measured across the
waist and thorax when applying 20 V between both feet
Average RR time interval from ECG and single foot-plantar IPG (seconds)
MEAN+2STD
MEAN
MEAN-2STD
-0.4
-0.2
0
0.2
0.4
0.6
0.72 0.73 0.74 0.75 0.76 0.77 0.78 0.79 0.8 0.81
RRECG - Pulse wavepeak-peak time difference (seconds)
Fig. 6. Bland-Altman plot of RR time intervals detected from the ECG
and single foot plantar IPG for one volunteer.
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
01234567891
0
Time (s)
A m p litud e (V )
-0.0 4
-0.0 2
0
0.02
0.04
0.06
0.08
012345678910
Time (s)
A m p litud e (V )
Fig. 5. ECG (top) and IPG (bottom) for foot-to-foot plantar impedance
measurements. Markers show IPG peaks obtained with a software
al
g
orithm.
Frequency (H z)
Pow er/frequency (dB/Hz)
-5 0
-4 5
-4 0
-3 5
-3 0
-2 5
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
C ardiac com ponent
R espiratory com ponent
Pow er/frequency (dB/Hz)
x 10
-3
Frequency (H z)
-40
-35
-30
-25
-20
-15
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
R espirato ry com po nent C ardiac com ponent
Fig. 7. Power spectral density of the IPGs measured from two different
subjects for a single foot case (top) and between both feet case (bottom).
Both
p
lots show com
p
onents related to the breathin
g
and heart rates.
TABLE I.
MEAN VALUES FOR VOLTAGES MEASURED ACROSS THE WAIST AND THORAX, ATTENUATION FACTOR, α RATIO AND GAIN G1 FOR THE TWO BIOIMPEDANCE
MEASUREMENT METHODS. CONFIDENCE INTERVALS CORRESPOND TO ONE STANDARD DEVIATION.
Measurements between both feet Measurements in a single foot
vcarrier (V) vcarrier (V)
Waist Thorax Attenuation Waist Thorax Attenuation
vout1 v
out2 A
1 = vout1/ vapp vout3 vout4 A2 = vou t3/ vapp α = A1/ A2 G1
8.05 ± 0.13 2.06 ± 0.14 0.4 ± 0.01 1.70 ± 0.62 1.46 ± 0.71 0.09 ± 0.03 5.32 ± 1.92 64.23 ± 22.67
B
AB
A
(a) (b)
Fig. 8. Current lines in IPG measurements (a) In a single foot and (b)
between feet. Current flowing through the torso is larger in this second
case. A : Injecting electrodes. B : Detecting electrodes.
6491
(vout1, vout2) and on a single foot (vout3, vout4), and G1 had been
tuned until vout1 = 8 V in the first case. A1 and A2 are the
respective attenuation factors with respect to the applied
voltage (vpp, 20 V), excluding the gain G1, to achieve
vout1 = 8 V, and α = A1/A2.
We can see that for voltages applied between both feet,
the voltage measured across the waist is less than 1 %
(A1/G1) of the applied voltage, and that the voltage across the
thorax is about 20 % of that across the waist, whereas for
single-foot measurements the voltages across the waist and
the thorax are about the same (vout3 ≈ vout4) and their peak-to-
peak value is less than 0.2 % (A2/G1) of the voltage applied
to a single foot. This is the expected result because when
measuring in a single foot most current lines flow across that
foot whereas when measuring between both feet all current
lines flow in the torso (Fig. 8).
The range of values needed for G1 is quite large. This
means that there is a large dispersion of values for the basal
impedance and, probably more important, for the electrode
contact impedance, very affected by skin moisture and
contact area.
The possible risk of the voltage or current applied for
bioimpedance measurements in people with electronic
implants or pregnant women should be evaluated from the
estimated interfering voltages (for electronic devices) and
current densities (for excitable tissues). In this work, voltages
much higher than those usual in bioimpedance measurements
have been used for convenience. In practice, IPGs can be
obtained by applying peak voltages smaller than 1 V.
Therefore, the proposed heart rate detection method based
on measurements in a single foot probably has no risk for
those two groups of people and, if there were any risk, it
would be smaller than that for biompedance measurements
between both feet.
IV. CONCLUSION
We propose a simple method to detect the heart rate in a
bathroom scale from the pulsatile component in the plantar
bioimpedance measured in a single foot. That component can
be mostly attributed to cardiac activity, but there are also
components attributable to the respiration. A simple peak
detector based on algorithms implemented with Matlab®
yields the heart rate. The respiratory rate is also present in
the signal recorded.
A clear advantage of this method as compared to those
based on bioimpedance measurements between both feet is
that it can be applied to people with a single leg. Further, an
estimate of the voltage across the waist and the thorax for
both methods shows that single-foot measurements yield a
voltage drop five times smaller than foot-to-foot impedance
measurements.
Finally, although the results presented here refer only to
measurements on people standing on the scale, preliminary
tests on seated people with a foot resting on the scale yield
IPG signals that also show peaks synchronous with the heart
beat. If measurements on a larger group o people confirm
these results, this would mean that the method could be
applied even to people that cannot stand on a scale. This
could be a very simple method for remote checking of the
heart rate on demand. Obviously, this measurement could be
performed by using a flat platform with four electrodes
instead of a scale. BCG-based measurements performed with
a scale, which are an alternative method to obtain the heart
rate, would not work for a seated person.
ACKNOWLEDGMENT
The authors thank the volunteers for their patience and
help, and Mr. Francis López for his technical support.
REFERENCES
[1] A. Hjalmarson, “Heart rate: and independent risk factor in
cardiovascular disease”, Eur. Heart J. Suppl., Supplement F, pp. F3-
F7, Sep. 2007.
[2] I. Korhonen et. al., “Health Monitoring in the Home of the Future”,
IEEE Eng. Med. Biol Mag.., pp. 66-73, May/June 2003.
[3] S. Kwatra and V. Jain, “A new technique for monitoring heart signals
part I: instrumentation design”, IEEE Trans. Biomed. Eng., vol. 33,
no. 1, pp .35-41, 1986.
[4] Y. Lim, K. Kim, K. Park, “The ECG measurement in the bathtub
using the insulated electrodes ”, in Proc. 26th Annu. Int. Conf. IEEE
EMBS, San Francisco, USA, 2004.
[5] M. Ishijima, "Unobtrusive approaches to monitoring vital signs at
home", Med. Biol. Eng. Comput., vol. 45, no. 11, 1137-1141, 2007.
[6] S. Ishida, N. Shiozawa, Y. Fujiwara and M. Makikawa,
“Electrocardiogram measurement during sleep with wearing clothes
using capacitively-coupled electrodes”, Proc. 29th Annu. Int. Conf.
IEEE EMBS, Lyon, France, 2007.
[7] A. Ueno, Y. Akabane, T. Kato, H. Hoshino, S. Kataoka, and Y.
Ishiyama, “Capacitive sensing of electrocardiographic potential
through cloth from the dorsal surface of the body in a supine position:
a preliminary study”, IEEE Trans. Biomed. Eng., vol. 54, no. 4, pp.
759-766, 2007.
[8] S. Devot, A. Bianchi; E. Naujokat, M. Mendez, A. Brauers and S.
Cerutti, “Sleep monitoring through a textile recording system”, Proc.
29th Ann. Int. Conf. IEEE EMBS, Lyon, France, 2007.
[9] P. S. Luna-Lozano, R. Pallas-Areny, “Heart Rate Detection from
impedance plethymography based on concealed capacitive electrodes”
presented at the XIX IMEKO World Congress Fundamental and
Applied Metrology, September 6-11, 2009, Lisbon.
[10] R. González-Landaeta, O. Casas and R. Pallas-Areny, “Heart rate
detection from an electronic weighing scale”, Physiol. Meas., vol. 29,
pp. 979-988, July. 2008.
[11] O. Inan, M. Etemadi, A. Paloma, L. Giovangrandy and G. Kovacs,
“Non-invasive cardiac output tending during exercise recovery on a
bathroom-scale-based- ballistocardiograph”, Physiol. Meas., vol. 30,
pp. 261-274, Feb. 2009.
[12] J. Shin, K. Lee and K. Park “Non-constrained monitoring of systolic
blood pressure on a weighing scale”, Physiol. Meas., vol. 30, pp. 679-
693, Jun. 2009.
[13] K. Oguma, "Body weight scales equipped with body fat meter",
TANITA CORP, WO9920175-A, Apr. 29, 1999.
[14] R. González-Landaeta, O. Casas and R. Pallas-Areny, “Heart rate
detection from plantar bioimpedance measurements”, IEEE Trans.
Biomed. Eng., vol. 55, No. 3, pp. 1163-1167, 2008.
[15] E. M. Spinelli, R. Pallas-Areny and M. A: Mayosky, “AC-Coupled
Front-End for Biopotential Measurements,” IEEE Trans. Biomed.
Eng., vol. 50, no. 3, pp. 391-395, March 2003.
6492