ArticlePDF Available

Abstract and Figures

Electrocardiography (ECG) is widely used in clinical practice, for example to diagnose coronary artery disease or the cause of chest pain during a stress test, while the patient is running on a treadmill. Ambulatory ECG monitoring is used for long term recording of ECG signals, while the patient carries out his/her daily activities. Artefacts in ECG are caused by the patient's movement, moving cables, interference from outside sources, electromyography (EMG) interference and electrical contact from elsewhere on the body. Most of these artefacts can be minimized by using proper electrode design and ECG circuitry. However, artefacts due to subject's movement are hard to identify and eliminate and can be easily mistaken for symptoms of arrhythmia and the physiological effects of exercise, leading to misdiagnosis and false alarms. This study provides an overview of various ambulatory ECG measurement systems and electrode topologies used for ECG sensing. The challenges associated with ambulatory ECG monitoring and the techniques used to overcome those challenges have also been discussed in this paper.
Content may be subject to copyright.
See discussions, stats, and author profiles for this publication at:
Critical review of electrocardiography measurement systems and technology
ArticleinMeasurement Science and Technology · November 2018
DOI: 10.1088/1361-6501/aaf2b7.
3 authors:
Some of the authors of this publication are also working on these related projects:
Multifrequency bioimpedance analysis of human tissues View project
Non contact biopotential sensing View project
Anubha Kalra
Auckland University of Technology
Andrew Lowe
Auckland University of Technology
Ahmed M. Al-Jumaily
Auckland University of Technology
All content following this page was uploaded by Anubha Kalra on 22 November 2018.
The user has requested enhancement of the downloaded file.
Anubha Kalra*1, Andrew Lowe1, and Ahmed Al-Jumaily1
1 Institute of Biomedical Technologies, Auckland University of Technology
Auckland, New Zealand
*Corresponding author Email:
AbstractElectrocardiography (ECG) is widely used in clinical practice, for example to diagnose
coronary artery disease or the cause of chest pain during a stress test, while the patient is running on a
treadmill. Ambulatory ECG monitoring is used for long term recording of ECG signals, while the
patient carries out his/her daily activities. Artefacts in ECG are caused by the patient’s movement,
moving cables, interference from outside sources, electromyography (EMG) interference and
electrical contact from elsewhere on the body. Most of these artefacts can be minimized by using
proper electrode design and ECG circuitry. However, artefacts due to subject’s movement are hard to
identify and eliminate and can be easily mistaken for symptoms of arrhythmia and the physiological
effects of exercise, leading to misdiagnosis and false alarms. This study provides an overview of
various ambulatory ECG measurement systems and electrode topologies used for ECG sensing. The
challenges associated with ambulatory ECG monitoring and the techniques used to overcome those
challenges have also been discussed in this paper.
Index Terms—Ambulatory, electrocardiography, diagnostic yield, injury current, motion artefact.
Biopotentials are the electrical signals that are produced inside the body due to various biochemical
processes that form a part of living metabolism. The monitoring/recording of these signals are essential
in the analysis of various physiological parameters and diagnosis. Electrocardiogram (ECG),
electromyogram (EMG) and electroencephalogram (EEG) are biopotential signals from the heart,
Critical Review of Electrocardiography
Measurement Systems and Technology
muscles and brain, respectively. The ECG gives a measurement of the electrical activity of the heart
over time. It is an important clinical diagnostic measure which is widely known and practiced. It can
be used to diagnose conditions such as cardiac arrhythmias, murmurs, pulmonary embolism and left
ventricular hypertrophy [1]. Other than medical applications, ECG monitoring finds use in sports, for
example optimizing an athlete’s performance during training [2]. Studies of the ECG date back to
1838 when Carlo Mateucci demonstrated the electrical signal accompanying each heartbeat with his
‘rheoscopic frog’ experiment [3]. The first human ECG was obtained by Waller (1887) [4], using
saline sensors connected to crude galvanometers. Einthoven (1901) [5] developed a string
galvanometer and devised an improvement over Waller’s method. He used buckets filled with saline
as the sensors, one for each hand and one for the left leg. Through his work, in 1912, the ‘Einthoven’s
Triangle’ became an important basis for interpretation of ECG measurements [6]. The lead system of
electrode configuration has been a gold standard for signal acquisition for various events such as
depolarization and polarization occurring in an ECG. The shape of Einthoven’s triangle represents an
inverted equilateral triangle with its center at the heart as seen in Fig. 1. The branches of the triangle
produce their vector sum as zero voltage.
A 12-lead ECG configuration gives a spatial information about the heart’s electrical activity [7], [8].
Electrode placement for a 12-lead ECG is shown in Fig. 2. Ambulatory ECG devices such as Holter
monitors employ 12-lead configuration, while event monitors employ 3-lead configuration with
electrodes on right arm, left arm and right leg. The electrical conduction in heart is carried out by
special cells responsible for the flow of current from one cell to another. A normal heart beat starts in
in the sino-atrial node (S-A node), located in the right atrium. The S-A node is connected to the left
atrium through an electrical path. This makes both the atria contract at the same time when the S-A
node fires. There is a slight delay in the conduction while the electrical signal travels down to the
ventricles through the atrio-ventricular node (A-V node). The action potentials generated by different
cells in the myocardium leads to the formation of an ECG signal. This action potential causes the
muscle tissue to contract by the flow of ions from cell to cell. This ionic flow causes a change in
potential which is known as depolarization and restoring back the original potential is called
An ECG test is generally used to analyze the heart rate, and to identify abnormal electrical behavior
of the cardiac muscle. Some common examples of abnormal ECGs can be seen in Fig. 3. A cardiac
rhythm with a correctly orientated P wave, where the depolarization begins at the sinus node is called
as a sinus rhythm [9]. Ventricle fibrillation is caused when the ventricles contract in a random and
unsynchronized manner; atrioventricular block occurs due to an impairment of conduction between
atria and ventricles; and premature ventricular contraction is due to the early contraction of the
ventricles during depolarization [10]. Chaotic and irregular fluctuations in the ECG baseline can be
due to atrial fibrillation, while the presence of saw-toothed flutter waves instead of P waves can
indicate the condition of atrial flutter.
A multi-lead ECG can be used by the cardiologists to approximate the electrical axis of the heart, by
observing the deflection in the QRS complex [11, Ch. 19]. The electrical axis refers to the mean
position of the action potentials during the formation of the QRS complex. Inclination of the electrical
axis to the left can be due to hypertension, aortic stenosis, ischemic heart disease etc., while the
deviation to the right is a consequence of chronic obstructive lung disease, pulmonary emboli,
congenital heart disease etc. [11]. An ECG test can be used to find the heart beat frequency, to
distinguish between normal sinus rhythms (60-100 beats per minute), sinus bradycardia (less than 60
beats per minute) and sinus tachycardia (higher than 100 beats per minute).
The ECG is also important in protecting and guiding the health of the athletes [12]. In a recent survey
by Harmon and Asif (2011) [13], all cases of sudden deaths were identified using the National
Collegiate Athletic Association (NCCA) database. During the 5-year period, there were 273 sudden
deaths in a total of 1 969 663 patients. Cardiovascular-related sudden death was the leading cause of
death in medical cases (56%), accounting for 75% of sudden deaths during exertion. The chances of
sudden death in male athletes were more than twice as compared to female athletes. The inclusion of
a 12-lead ECG in screening tests of all the athletes was recommended by the European Society of
Cardiology (ESC) [14]. However, it is difficult to distinguish the abnormal patterns from the
physiological effects of training. Many clinical findings that may cause a concern in the general
population are normal for athletes.
Having discussed the importance of the ECG as a one-off measure, it is also recognised that long term
monitoring of the ECG is clinically important. There is worldwide demand for a continuous health
monitoring system that can detect heart rate variability through which cardiovascular diseases
(accounting for 48% of non-communicable disease deaths, as per 2012 WHO Statistics [15]) can be
diagnosed and managed at an early stage. Although standard clinical devices are employed with
techniques for sensing blood perfusion, cardiac sounds and vascular blood velocity; ambulatory ECG
and blood pressure monitoring are considered the most mature techniques [16]. Cardiac arrhythmia
events can be life threating, therefore regular monitoring and recording of ECG is valuable and can be
used by a physician to achieve timely and accurate diagnosis or to determine the cause of the symptoms
in patients [17].
The ambulatory ECG device provides a continuous monitoring of the heart for longer periods of time,
ideally without interrupting the patient’s daily activities. Ambulatory ECG systems aim to achieve
patient comfort and ease of use as well as efficient signal acquisition. Ambulatory systems are used to
identify infrequent and highly variable cardiac arrhythmias which normally pass undetected in clinical
situations [18] [19]. Ambulatory ECG monitoring, in conjunction with clinical findings can be useful
in investigating heart conditions such as palpitations, light-headedness or syncope [20].
The clinical importance of arrhythmia detected using ambulatory ECG monitoring can be determined
by finding its correlation with the simultaneous occurrence of suggestive symptoms, also known as
diagnostic yield. The symptoms of cardiac arrhythmia occur more frequently in outpatients than in
hospitalized patients. Surawicz et al. [21] through his experiments found that the symptoms of cardiac
arrhythmia in outpatients and inpatients were 55% and 6% respectively. However, the correlation of
symptoms with detected arrhythmia in inpatients (95%) was higher than in outpatients (44%). The
performance of an ECG monitoring device is evaluated by determining its diagnostic yield. The
primary features and diagnostic yield of the most commonly used ambulatory ECG devices such as
Holter Monitors, Event Monitors, Implantable Loop Recorders and Patch Sensors have been
elaborated in this section. The benefits and downsides of latest commercially available long-term ECG
measuring wearable sensors have also been discussed.
a. Holter Monitors
The first ambulatory cardiac monitoring device, now known as a Holter monitor was built by a famous
American biophysicist Norman J. Holter (1914-1983). Modern Holter monitors are battery-operated,
portable devices which can measure the ECG continuously for 24-48 hours [22], and are worn by the
patients with suspected, frequent palpitations having slow, fast or an uneven heartbeat. They may also
be used if a person has a pacemaker to ensure its proper functioning. 12 channel Holter monitors obtain
ECG signals in the same representation as during common rest ECG and/or stress test measurement
using the Mason-Likar lead system. However, the resolution of recordings using 12 channel Holter
monitors is significantly lower than standard 12-lead ECG. Modern Holter monitors employ two or
three channel ECG [22].
In a study performed by Zeldis et al. in 1980 [23], the concurrence of the symptoms with an associated
arrhythmia was found only in 50 of 371 patients (13%) using Holter monitors. Cardiac diagnosis
obtained using Holter monitor by Drake (in 1984) [24], showed no significant relevance with
associated arrhythmias. Thus, it can be inferred that Holter monitors employing 24-hour ECG
monitoring exhibit a low diagnostic efficacy. Continuous long term monitoring is necessary for the
correct diagnosis of the arrhythmia detected by an ambulatory device. When the monitoring time of
the Holter monitor was increased from 24 to 72 hours in 95 patients with syncope [25], the occurrence
of symptomatic events increased from 15% to 27%. Therefore, 24 hour ambulatory monitoring might
not be enough to identify potentially important arrhythmias [26].
b. Event Monitors
In cases where patients experience intermittent or rare symptoms of palpitations, event monitors are
used for ambulatory monitoring of the ECG. Event monitors, also known as loop monitors, are small
and light-weight devices which can be triggered by the patient when they feel symptoms arise. The
latest event recorders employ real-time continuous cardiac monitoring, where the arrhythmic event
data is automatically transferred to the monitoring station [27] [28]. In comparison to Holter monitors,
event monitors are smaller in size and can record cardiac activities for longer periods of time, although
they are triggered by the patients during symptoms [17].
c. Implantable Loop Recorders
Implantable Loop Recorders (ILRs) or Insertable Cardiac Monitors (ICM) are used for the detection
of infrequent arrhythmias or in the cases where other ambulatory devices are indeterminate. They are
used for cardiac monitoring for prolonged periods of months to up to 3 years [29] [30]. ILR devices
are placed under the skin and can automatically record continuous long-term signals.
d. Patch Sensors
In recent years, innovative engineering has led to the development of non-invasive thin patch
electrodes for long-term ECG monitoring. Two examples of adhesive patch electrodes available in the
market are Zio Patch (iRhythm, CA, USA) and SEEQ Mobile Cardiac Telemetry (Medtronic,
Minneapolis, USA). The Zio patch device can measure one-lead ECG using a small adhesive patch. It
can record cardiac activity for up to 14 days continuously without requiring any leads, wires or
batteries [31] [32]. SEEQ Mobile Cardiac Telemetry (MCT) sensors can be used for 30 day ECG
monitoring, are water resistant and can be worn during showering [33]. Devices such as this are a
convenient type of ECG monitor and can be used to obtain the average, maximum and minimum heart
rate, number of premature beats, longest R-R intervals and ECG recordings for patient triggered events
Another patch type ECG device is the Netguard, developed by Mindray [34]. The device is composed
of two custom electrodes worn on the chest. A drawback of the product was that it was limited to
operating within range of its base station.
A similar device, V patch was developed by Intelesens [35]. The device performed in a similar way to
Netguard, offering advantages of portability and a battery life of a week after a full charge. The
shortcoming was the addition of a bulky base station, to be worn along with the device.
The integration of the sensing electrodes in a patch to form a Band-Aid adds to the list of existing
smart devices to sense bio signals. It provides very convenient way ambulatory cardiac monitoring
along with features like real time analysis and wireless ECG telemetry. Also, the monitoring can be
complemented with functions like auto-trigger event handling and recurring event handling.
A glass bottle cap has also been implemented as a reusable, compact ECG patch electrode. Engineering
World Health’s ECG pads do not have to be thrown away after use; they can be boiled to sterilize
them [36]. The conductive gel required to fix the pad on the chest can be made of water, flour or salt.
There are several commercial research groups and companies that have developed variants of wearable
patches as biomedical sensors [37].
The main advantage of using patch sensors is that they are easy to use, can be used for long-term ECG
monitoring, require less maintenance, are less intrusive to daily activities and water-resistant [38]. In
2014 , Barret et al. [32] conducted a study in which both Zio patch and SEEQ MCT systems were well
accepted by the subjects, where 93.7 % of them found the former more comfortable and 81% preferred
them over Holter monitor. However, their disadvantages include high cumulative consumer costs and
dependence on the device’s company for accurate generation of a summary report. In one study by
Shinbane et al. (in 2013) [39], the average time to diagnose a clinically relevant arrhythmia was found
to be 5.8 ± 6.1 days, therefore patch electrodes have a higher diagnostic efficacy than Holter and
Event monitors as they can be worn for a longer duration. In another study conducted by Rosenberg
et al. (2013) in 75 subjects [40], the use of Zio patch for ~10.8 days resulted in the determination of
81% more arrhythmias compared with 24-hour Holter monitoring. The diagnostic yield efficiency
using Zio patch increased from 43.9 to 62.2% when the duration of ECG monitoring was increased
from 48 hours to 7.6 ± 3.6 days [31] [41]. A comparison of their features and diagnostic yield has
been illustrated in Table 1.
ECG Monitors Features [26] Diagnostic Yield
Are used in patients with daily or nearly daily
symptoms. A Holter monitor report includes:
1) Total heart beats
2) Average heart rate
3) Maximum and minimum heart rates
4) Number of premature beats
5) Episodes of tachyarrhythmia
6) Longest R-R interval
7) ST segment changes
8) Patient-triggered symptoms and any associated
ECG findings
9) Hourly samples of the ECG tracing
5 % - 13 %
Event Monitors
Are used in patients with weekly to monthly
symptoms. An Event monitor report includes:
1) Patient triggered ECG recordings
2) Technician’s interpretation of the tracings
3) Reported symptoms and their duration
6 % - 25 %
39 % - 68 %
Implantable Loop
Are used in patients with infrequent symptoms (less
than monthly). An implantable loop recorder report
1) ECG tracings for each patient-triggered or
auto-triggered event
2) Technician's interpretation of the tracing
3) Reported symptoms and their duration
45 % - 88 %
Zio Patch Are used for long-term non-invasive ECG recording.
Zio patch report includes:
1) Average heart rate
2) Maximum and minimum heart rates
3) Number of premature beats.
4) Episodes of tachyarrhythmia
5) Longest R-R interval.
88 %
e. Other Sensor Systems
Other types of commercially available ambulatory ECG devices include EPIC sensors [42] [43], chest
harnesses [44] and multi-purpose vest shirts [45].
An unconventional approach of using EPIC (Electric Potential Integrated Circuit) sensors was
designed to track the heart and respiration rates for vehicle drivers. These sensors were mounted on
the back side of the chair and operate by capacitively sensing the ECG. To ensure proper safety
conditions, the person touches both the sensor and some metal at ground potential. A steady ECG
signal can be obtained after a settling time of tens of seconds due to large time constants pertaining to
large impedance parameters.
An ambulatory ECG chest harness was developed by Cleveland based Orbital Research, which can
measure ECG up to 48 hours. No skin preparation or use of any conductive gel is required as dry
electrodes are embedded in the chest harness. A direct contact with skin is made and the electrodes
are held in place so as to reduce the effect of motion artefacts and improve the signal-to-noise ratio of
Orbital’s dry electrodes. A vest shirt allows medical professionals to perform frequent and less costly
fitness and ECG monitoring. The body temperature, motion and ECG can be measured through this
shirt. Motion detection is evaluated using an accelerometer. Another proposition by IMEC (Belgium)
for a wearable ECG device comprises of three leads along with a 3-axis accelerometer and Bluetooth
radio for wireless transmission. It has a battery life of up to a month. However, the limitation of the
device is that it doesn’t transmit raw ECG data, but derived waveform information like heart rate,
offset of P, QRS and T waves, etc. [46].
ECG is generally measured using wet, dry contact and non-contact capacitive electrodes. Electrical
models of ECG measurement electrodes can be seen in Fig. 4, where
, ; , ;  ,  are the resistances and capacitances of the conductive
gel, skin and clothing respectively.
a. Wet Electrodes
In Fig. 4, In ECG monitoring, conventional silver-silver chloride (Ag/AgCl) electrodes are widely
used. These electrodes use a conductive gel to maintain good electrical contact between the electrode
and the skin and typically incorporate an electrolyte gel or solution that buffers the electrolytic
composition through the outer and inner layers of the skin. This poses problems for long term ECG
monitoring, mainly because the gel might dry out over time [47]. Also, the use of electrolyte benefits
from the region of application being as stationary as possible so that the electrode-skin impedance
doesn’t change if the electrolyte egresses due to movement of the patient. Moreover, the gel and
adhesive can cause allergic reactions for some patients’ skin. Despite decades of research in non-
contact electrodes, conventional wet Ag/AgCl electrodes are still used universally for clinical and
research applications [48].
b. Dry Contact Electrodes
Dry electrodes make electrical contact with the skin but do not employ paste or gel media. Some
flexible dry electrodes made of rubber, fabric or foam are quite appealing in terms of comfort of the
patient and reducing motion artefacts by conforming to the body during motion. Although the use of
dry electrodes is advantageous in terms of comfort for the patient, its use is limited as dry electrodes
tend to fail to maintain contact with skin for long times. The electrode-skin interface structure for dry
and non-contact electrodes is much more variable than for wet electrodes. The skin-electrode interface
can be described as a layered capacitive and conductive structure, with a series combination of parallel
RC elements [49]. The performance of dry electrodes depends on the presence of sweat on the skin.
The conventional notion that low resistance (high conductance) is essential for electrode performance
could be misleading. There is a trade-off between the performance of dry and wet electrodes in the
transient and stability periods. The wet electrodes perform well allowing for a short time to stabilize
the electrochemical interface, whereas dry electrodes take a comparatively longer time to achieve a
stable trace, as their performance depends on skin’s perspiration [49].
c. Capacitive Electrodes
Conventionally, dry and wet electrodes are operated through direct physical contact with the skin.
Capacitive sensing provides a non-electrical contact mode of operation. The capacitive electrodes
sense the signals with a significant gap between the sensor and the skin. The signal is essentially
coupled through an insulation media such as hair, clothing or air. Electrostatic frictional effects also
contribute to the input voltage noise. A coupling capacitance is formed between the skin and the
electrode. The thickness of the dielectric between the skin and the electrode, and the surface area of
the plates decides the value of coupling capacitance [50]. Conductive threads when integrated into
garments can act as a capacitive sensor and are also classified as textile-based sensors or textile
electrodes. They have become a desirable form of ambulatory ECG monitoring. Although they provide
comfort to the patient, their use is limited as they provide high skin contact impedance due to their
asymmetrical surfaces [47], [51]–[55].
The rate of electrochemical processes occurring between the electrode and the skin surface is directly
proportional to the area of their interface. For this reason, porous polymer wet electrodes that provide
an immensely high electrode/electrolyte interaction area are favorable to implement in an ambulatory
ECG system [56]. Conformal polymer electrodes stuck onto the chest of the patient using an adhesive
can reduce inaccuracies due to change in subject-sensor gap in capacitive electrodes. Also, they
provide a better electrical conduction than dry and capacitive electrodes. Lee (2014) [57] developed
thin flexible polymer electrodes using carbon nano tubes (CNTs) and polydimethylsiloxane (PDMS)
with similar mechanical properties to the skin. Carbon nano tubes (CNTs) are an allotrope of carbon
which are cylindrical in shape and are potentially useful in a variety of applications like optics,
nanotechnology etc. There are two main kinds of CNTs: single-walled CNTs (SWCNTs) and multi-
walled CNTs (MWCNTs). They are cost effective and good conductors of electricity [58]. The CNTs
are tangled and assembled randomly, therefore, they exhibit good electrical contact if embedded in a
polymer electrode when the polymer is stretched or bent [57]. The advantage of using CNT based
electrodes is that they penetrate the wrinkles of the skin and maintain a steady contact. This leads to
an increase in the contact area, thereby reducing the contact impedance. CNT polymer electrodes are
discussed further in the next section.
The choice of electrodes for biomedical applications, especially ECG monitoring, depends not only
on the comfort that they offer to the patient, but also on the quality of the signal obtained. The skin-
electrode interface decides the operational characteristics of any electrode system in conjunction with
the properties of the electrode material.
The presence of noise sources having different frequency ranges deteriorate the ECG signal quality.
The low frequency noise is called the baseline wander, medium frequency noise includes power
frequency interference and high frequency noise can be substantially due to electromyography (EMG)
signals. The non-physiological sources of artefact such as external electromagnetic signals and power
line frequency can be successfully removed with the use of a driven right leg (DRL) circuit, which is
discussed in more detail in section V. In modern ECG devices, digital filters are typically used to
eliminate baseline wander from ECG waveforms. Artefacts in ECG are also generated due to the
contraction of the muscles in the vicinity of the electrodes [59], which can be reduced to some extent
by proper electrode design and placement [60].
Holter monitors have electrodes attached with tape or adhesives, which might cause skin irritation and
discomfort to the patient. Moreover, the moist inner pad of the Ag/AgCl electrodes used with Holter
monitors can dry up over time, leading to a poor connection. A performance comparison of currently
used ECG sensors has been presented in table 2.
Cable movements during exercise can introduce noise, which can be reduced by using a unity gain
buffer amplifier (voltage follower) at each electrode [61]. The capacitive mismatch in active non-
contact electrodes can be significantly reduced by bootstrapping [62]. Other sources of noise include
contact noises, which are introduced in ECG signals due to disturbances in electrode-skin impedance
caused by poor adhesion and conductance of the electrodes [63].
Motion artefacts refer to the noise generated in the ECG due to movement of the electrodes.
Ambulatory devices can mistake motion artefact for fatal arrhythmias such as ventricular fibrillation
(VF) or ventricular tachycardia (VT) and may trigger a false alarm [64]. Motion artefacts strongly
affect the ECG morphology and remain one of the major problems in short-term and long-term ECG
monitoring. An example showing the effect of motion artefacts on ECG is illustrated in Fig. 5.
Commercial ECG
Epic Sensors
1. Can monitor patient’s
alertness while driving.
1. Patient needs to sit still
on the seat for a
considerable amount of
time to allow the circuit
to settle.
2. Very user specific,
therefore can’t be used
when person is
performing his/her daily
Chest Harness
1. Can monitor ECG of the
patient for up to 48
hours without using
any adhesive or
conductive paste.
1. Provide low electrical
2. Their performance
depends on the presence
of sweat on the chest of
the patient.
3. Not comfortable for
long time ECG
Multipurpose Vest Shirts
1. Very comfortable to
2. Can monitor
acceleration and ECG
of the patient.
1. Prone to signal
distortions due to the
varying amount of
skin-sensor gap.
ECG motion artefacts are generated due to such things as tremor or shivering, exercising and heavy
breathing. In addition to measurement artefacts such as electrode-skin interface changes, physiological
artefacts resulting from this motion also occur and are generated through the skin; therefore different
measures such as skin abrasion and mechanical stabilization are adopted to minimize motion artefacts
The movement of the patient results in skin stretch; which in turn generates potentials in the epidermis.
Skin stretch is considered a major physiological source of motion artefact in the ECG [65]. Odman
[66] established a non-linear relationship between motion artefact and the mechanical stress applied
on the skin. The skin stretch potentials obtained at different equally spaced time series were found to
vary between individuals. In 1981, Odman measured the magnitude of deformation-induced potentials
in the skin area beneath the electrodes [67]. Two metal plates were adhered to the skin and the
rectangular area of the skin between the plates was stretched. The magnitude of potentials decreased
with increasing distance from the rectangular zone. The movement-induced potentials were studied in
different electrode configurations by Odman and Oberg in 1982 [68]. The study concluded that only
small potential variations occur due to change in conduction caused by electrode electrolyte
displacement during motion, while skin deformation is the major source of motion artefact. The origin
of skin stretch induced motion artefacts was explained by Talhouet and Webster in 1986 [69]. From
their experiments, it was inferred that both motion artefact and impedance increase logarithmically
with skin stretch. In 1989 [70], Odman found that the potential changes due to skin stretch were higher
in curved skin surfaces than in flatter skin surfaces.
Skin is composed of three layers: epidermis, dermis, and hypodermis [71] as illustrated in Fig. 6. The
capillary loops in the corneum nourish the skin. The new cells form in the stratum germinativum and
move outward towards the stratum granulosum and the barrier layer [59]. The cells die in the barrier
layer, stay on the surface of stratum corneum and fall off after some time. Bioelectric currents are
generated due to various biological activities occurring in the body. The bioelectricity in the skin is
caused due to the flow of ions between the dead cells on the epidermis and the new cells on the inner
skin layers [65].
The epidermis layer is a storage of negative ions (anions) and is permeable to positive ions (cations)
[72]. On the other hand, the inner layers of the skin have a positive charge on them due to the
accumulation of positive ions (cations). Therefore, the skin behaves like a dc battery where the current
is generated due to the flow of positive and negative ions across the barrier layer [72]. The skin’s
bioelectricity may depend on various factors such as hydration, emotions, stress and disease.
The impedance of the barrier layer is 50 kΩ/cm2 and the skin potential between the inside and outside
of the barrier layer is 30mV. When the skin stretches, the skin potential can drop to 25 mV, and this
5mV change in the potential appears as motion artefact in biopotential measurements. Thakor and
Webster hypothesized that the difference in metabolic activities between stratum corneum and stratum
germinativum lead to the flow of ‘injury current’ through the extracellular resistance [73] [65]. The
5mV difference in the skin potential can be reduced by scratching the skin with about 20 strokes of
fine sandpaper [74].
Although many ambulatory ECG monitoring biosensors have been commercialized to date, a major
problem is still faced due to patients performing motion related activities that introduce unwanted
signal noise and makes monitoring less effective [75]. The frequency spectrum of the motion artefact
overlaps the ECG, therefore, it is the most difficult form of noise to be removed from the ECG signals
[76]. Beyond skin abrasion, various motion biosensors used to date to remove motion related noise
don’t respond well in cases where the patient is performing vigorous exercises [77].
Approaches to motion artefact reduction include modification of materials involved in the skin
electrode interface and by implementing models and algorithms for reducing motion artefact
contribution [78], as discussed in the following section. While much research has been conducted to
remove time invariant noise, the removal of motion induced artifacts remains an unsolved problem
[79]. The latest motion artefact rejection techniques employ motion tracking devices to identify motion
and incorporate their use into adaptive algorithms like neural networks [80, p.] and fuzzy-rule-based
adaptive nonlinear filters [81] to adjust digital filter coefficients. Various motion sensing and signal
processing techniques employed to eliminate motion artefacts from ECG signals have been discussed
in section V.
The techniques which are generally employed to overcome the challenges faced in ambulatory ECG
monitoring are: A. ECG circuit improvisation; B. signal processing and C. motion tracking and
padding. They have been discussed in more detail in this section.
a. ECG Circuit Improvement
The earliest machines used to record ECG were large, cumbersome devices that required patients to
immerse their limbs into bucket electrodes filled with saline solution. Improvements in electrodes and
instrumentation electronics, as well as the development of analogue to digital converters and digital
computers have revolutionized the ECG. Many modern ECG’s are small enough for a single person
to easily carry and often include digital filtering techniques and computerized interpretation methods.
In a typical modern device using skin contact or non-contact electrodes, the differential voltages
caused by the depolarization and polarization of the heart muscle can detected. These signals are then
amplified using an instrumentation amplifier. At this stage analogue filtering and further amplification
takes place before the signal is digitized by an analogue to digital Converters (ADC). This digital
signal can then undergo digital signal processing (DSP) such as filtering. Lastly the signal can be
displayed, stored, and transmitted as required. A patient protection circuit protects the patient from
potential electrical shocks or burns [82] [83]. A generalized ECG block diagram is shown in Fig. 7.
The first stage of ECG circuit consists of an instrumentation amplifier that amplifies the weak ECG
signal, which has a typical amplitude of 0.5 mV and eliminates the high frequency noise received by
the antenna (the leads connecting the electrodes to the amplifier) [84].
The instrumentation amplifier is essentially a combination of two buffer stages, which eliminates the
need for input impedance matching. Chi et al. (2009) [62] integrated an instrumentation amplifier with
an additional bootstrapping amplifier. Bootstrapping is a method through which an operational
amplifier restores its losses by increasing the input impedance and a part of the output of the
instrumentation amplifier is used to drive the input. Integrated analogue front end solutions (IAFEs)
available today have made it possible to produce high quality ECG recordings with very small,
portable, low power devices. These IAFEs include a range of different features and, compared to
discrete components, have excellent electrical characteristics and very high resolution ADCs at a
relatively low cost in extremely small packages.
Another elementary aspect of the ECG circuit is the introduction of a band pass filter network with a
lower cut-off frequency of 0.5 Hz and a higher cut-off frequency of 100 Hz, which corresponds to the
typical frequency bandwidth of the ECG signal. A notch filter with a cut-off frequency of 50 Hz (or
60 Hz) can be used to remove interference due to mains power [84]. An inverter can correct negative
QRS in the ECG. The introduction of a DC-offset stage with the band pass filter adjusts the offset of
the recorded ECG waveform from the reference voltage, thereby making Analogue to Digital
Conversion easier [84]. The main purpose of adding a driven right leg circuit to an ECG circuit is to
reduce the common mode voltage in isolated and non-isolated measurements [85]. For a three lead
configuration, the voltage (
) between the right leg and amplifier common (due to right leg
impedance ) is countered by the output of the driven right leg amplifier as shown in Fig. 8.
A sufficient galvanic isolation is necessary to ensure the safety of the patient if he or she contacts the
mains line voltage, which is typically 220-240 V, 50-60 Hz [86],[87]–[89]. An efficient biopotential
amplifier configuration must account for the electrical interferences from the power line along with
effective guarding/shielding and driven right leg circuits ([90]–[92]). Huhta and Webster (1973) [87]
clarified the interpretation of all sources of electrical interferences during an ECG measurement. They
established that the unsymmetrical electrode placement and unbalanced electrode impedances
contribute majorly to the interferences and can be reduced using a right-leg driver amplifier. A basic
model illustrating the purpose of an isolation mode amplifier can be seen in Fig. 9.  is the
capacitance between body and ground &  is the capacitance between body and mains power
causing the interference current  to flow through .
In Fig. 9,  flows through  to the isolated common of the amplifier and via the isolated
capacitance  to the ground. The isolation amplifier gives an output voltage V2 which serves the
purpose of suppressing the isolation mode voltage  . With the switch closed, the value of  is
reduced to a great extent, which in-turn reduces .
b. Signal Processing
In addition to measurement hardware, software techniques have also been employed for noise rejection
in several studies and are discussed here.
The wavelet transform is a proven tool for efficient filtering of signals in bio-signal processing [93]
[98]. In this procedure, the signal is first decomposed, followed by its thresholding and then proper
reconstruction [98]. Raghuram et al. [99] used different wavelets to reduce motion artefacts from
corrupted photo-plethysmography (PPG) signals. The wavelets used included the daubechies, bi-
orthogonal, symlet, coiflet etc. out of which the daubechies produced the best performance. In a study
performed by Foo [100], the adaptive filtering technique was found to be more efficient in removing
noise from PPG signals as compared to discrete wavelet transformation.
Adaptive filtering has been applied in several works as a common filtering technique for the treatment
of bio signals [101]–[106]. Thakur and Zhu, 1991, applied it in foetal ECG recording, cancelling the
cardiogenic interference signal from that obtained from impedance plethysmography (IPG), noise
reduction from muscles, cancelling the 60 Hz power supply interference and ECG motion artefact
reduction [76]. Generally, adaptive filtering is realized by the subtraction of the noise from the
received signal in an adaptive manner. The technique employs two inputs, one being the overall ECG
signal, and the other being the noise source. As seen in Fig. 10, the adaptive filter estimates the noise
from its source sensor which is then subtracted from the first input [65]. Principal Component Analysis
(PCA) and Independent Component Analysis (ICA) are widely used for noise cancellation in ECG
signals [107]–[114]. In PCA, the data matrix is decomposed into a set of orthogonal components
arranged in the order of their importance.
In other words, if the first component of PCA is the best representation of the data set, then the second
component will be the second-best representation and will be orthogonal to the first component. In
ICA, uncorrelated components of the data are generated. ICA aims at producing such non-Gaussian
transformations which assure that the output signals are statistically independent [115]. The main
difference between PCA and ICA is that the former decomposes the data into a set of uncorrelated
components, whereas the later provides a set of statistically independent components.
Ramaswamy [116], implemented PCA by adaptively segmenting uni-channel ECG signals. A higher
increase in SNR was observed when the signal was corrupted with more noise. The three most
significant principal components showing the highest correlation with the clean ECG were chosen.
The performance of PCA was better when more principal components were retained in cases of highly
corrupted signals. A reduction in signal to noise ratio (SNR) was observed on reducing the ECG
channels from 8 to 2. Therefore, the performance of PCA was found to be dependent on the number
of ECG measurements in the input data set.
In 2011, Romero [117], investigated the performance of PCA and ICA in the context of noise rejection
from ECG signals acquired for 10 seconds. It was observed that both PCA and ICA showed similar
performances when the SNR of the noisy ECG was up to 2dB, while ICA outperformed PCA for SNRs
below that value.
A comparative analysis of ICA and adaptive filtering was performed by Rehman and Khan in 2016
[118]. Due to its non-iterative nature, ICA performed better than the three iterative gradient based
algorithms employing LMS, normalized LMS (NLMS) and recursive LMS (RLMS). NLMS is
generally used to normalize the LMS filter’s input to attain better filter stability, whereas RLMS (also
known as RLS) recursively finds the coefficients to minimize the linear least squares cost function
related to the input signals.
The wavelet transform technique combined with ICA was implemented by Abbaspour (2015) [119]
for motion artefact rejection. Motion artefact was added to 30 minutes of simulated ECG signal which
was then filtered using wavelet transform. After that, ICA was used on the wavelet transformed signal
and a higher SNR of 14.2 dB was achieved compared to 13.85 dB using only wavelet transform.
For bio-signals such as ECG, ICA has been found to be increasingly used since it does not require any
prior knowledge of the system [120] [121].
c. Motion Tracking and Padding
Motion artefact due to skin stretch can be eliminated by abrading the skin with a sand paper. The sand
paper scratches through the barrier layer and short circuits the skin potential [59]. The use of sandpaper
can cause bleeding, which might lead to skin infection. A skin puncturing technique developed by
Burbank and Webster in 1978 showed significant reduction in motion artefacts [122]. The barrier layer
provides a protection to the underlying layers of the skin from irritating substances like electrode gels.
Therefore, only mild electrode gels should be used after the skin is abraded with sand paper [59].
However, the motion artefact returns as the skin regrows in 24 hours [74].
The electrode contact pressure with the skin and the surface moisture have been considered as two
main factors influencing the skin-electrode impedance [123]–[125]. According to experiments
performed by Kim et al. (2008) [126], an improvement in motion artefact rejection was observed when
higher contact pressures were applied on the electrodes. In 2013, Comert et al. [127], recorded a two
channel ECG from the chest of a subject as shown in Fig. 11.
A reference electrode was placed on the lateral upper arm location, where motion artefact was
introduced by random arm movements. This was done to simultaneously record ECG signal with
motion artefacts (between reference and Ch1) and without motion artefacts (between Ch1 and Ch2).
The noise due to motion was estimated by calculating the difference in detected R peaks between both
signals. A reduction in noise was observed when a pressure of between 5 and 25 mmHg was applied
on the reference electrode using a foam pad. Pressure exerted on the foam pad greatly influenced the
motion artefact depending on the foam material and the intensity of pressure exerted. The ECG signal
deteriorated on application of pressure above 20mmHg.
Several studies identify motion artefacts in ECG by employing external devices for motion tracking
using accelerometers, linear variable differential transformers (LVDTs), gradiometers and optical
sensors. The motion is then quantified and adaptively used in the filters to reduce motion artefacts
from ECG signals. Several devices employed for motion tracking or quantification are discussed in
detail here.
a) Accelerometers
The accelerometers are used predominantly to measure motion parameters in a mechanical model of
a system. For the accurate detection of motion, 3-D spatial measurements are taken to consider the
effects in each orthogonal plane. This can be achieved by using 3-axis accelerometers and have been
proposed to realize adaptive motion artefact reduction. In 2008, Yoon and Min [128] implemented an
adaptive filtering technique to estimate the subject’s movement using a 3D accelerometer. The motion
information was then subtracted from the ECG signal to derive a refined ECG output. In Fig. 12, the
3 orthogonal axes- U, V and W constitute the acceleration coordinates, while U’, V’ and W’ account
for global Cartesian coordinates. The chest of the subject is aligned in the V’-W’ plane, and it was
presumed that ECG was generated in the U’ axis. The motion artefact will be aligned along V’- axis
when the subject is walking or jumping. A tri-axial accelerometer offering the measurement of both
dynamic and static acceleration was used by Kishimoto and Kutsana [129] to sense the motion artefact
in ECG during sleep. In 2011, Liu [130] used the signals from an accelerometer to cancel the motion
artefacts through adaptive filtering in a portable ECG recorder with Bluetooth technology.. From
experiments conducted by Raya and Sison [102], it was inferred that the use of a single axis
accelerometer (particularly vertical axis) was sufficient for motion artefact cancellation. This was
supported by the fact that the major kinematic acceleration in humans was found in vertical direction.
b) Linear Variable Differential Transformers
Kang (2007) [47] made use of linear variable differential transformers (LVDTs) for the estimation of
changing position and deformations due to stretching while measuring ECG. A double substrate sensor
structure with stretchable and non-stretchable textiles was implemented, as seen in Fig. 13.
Fine magnetic wires were stitched on the stretchable area. The convention has been to employ
piezoelectric films due to their flexibility; however, their use is limited only to detection of micro
As the sensor stretches, two rectangular spirals slide over each other changing the mutual inductance
between them. Non-washable fabric active electrodes were employed for long term monitoring of
respiratory and ECG signals; therefore, electrode sensors may malfunction due to sweat and other
water sources. Screen printing technology was implemented to incorporate the sensors into textiles,
which may dramatically reduce the inductance due to geometric limitations of screen printed lines on
rough textile substrates.
c) Gradiometers
Non-contact electrodes with a polymer over-layer of thickness 0.18 mm were fabricated from a copper
printed circuit board for measurement of ECG signals by Peng (in 2014) [131]. Dual and quad
gradiometer electrode configurations were devised to account for a subject’s activity or ambient
modulation. The output of dual electrode gradiometers was taken as the difference between the
positive and the negative voltages. Whereas for quad electrodes, the output was measured by
calculating the difference between the diagonal summations. The obtained signal depended upon
certain parameters like the distance between the subject and the sensor ‘h’, and the angles ϴx and ϴy,
as observed in Fig. 14. Change in these parameters modulated the source capacitance by changing the
subject-sensor gap. The source capacitance was a series combination of the capacitance from the
polymer over-layer, air gap and the subject’s clothing.
The ECG signals were obtained from two gradiometer electrodes and conventional Ag/AgCl
electrodes. Using the measured R-R intervals, the average beats per minute (bpm) were determined
along with its standard deviation. The performance of quad configuration was better than dual configuration
d) Optical sensor
Image-based motion tracking techniques involve feature selection for tracking objects in consecutive
frames using color based comparison, edge detection, optical flow methods or texture intensity
methods [132]. Several point detection techniques in MATLAB® employ feature tracking algorithms
like Kanade Lucas-Tomasi (KLT) and surface feature detection. Another popular technique ‘estimate
geometric transform’ returns a 2D geometric transformed image and employ M-estimator Sample
Consensus (MSAC) algorithm to map the initial image to the final image [133]. A sensor to estimate
skin stretch simultaneously with ECG was developed by Liu in 2007 [65],.
The surface underneath the sensor was illuminated using a light emitting diode (LED), and the
displacement of the skin from the reference was measured using a CMOS image sensor. Assuming L
as the distance between the fixed edge and the imaging area, the optical sensor output was calculated
as shown in eq (1).
=( − 0)2+ ( − 0)2  (1)
where and are the optical sensor outputs and 0 and 0 are initial optical sensor outputs.
These uniaxial displacements were used to adjust the filter coefficients of an adaptive filter employing
a least mean squares algorithm (LMS), first devised by Widrow and Hoff in 1960 [101].
The clinical approaches used for cardiac arrhythmia diagnosis only provide a periodic assessment of
the disease. The majority of cardiac deaths are sudden, therefore continuous monitoring of the heart is
necessary to enable timely detection of any cardiac instability. Many arrhythmias occur paroxysmally
and might be hard to record in the doctor’s office. Thus, ambulatory ECG monitoring can be useful
for short term or long-term evaluation of cardiac arrhythmias. Many symptoms arise only while
performing certain activities like eating, exercising or sleeping. A continuous ECG recording can help
in detection of such events, the type of pattern they produce, how long they last and whether they are
related in time. Ambulatory ECG monitors can be externally worn as Holter monitors, event monitors
and patch sensors or implanted in the skin’s subcutaneous layer as ILRs. As discussed, the clinical
importance of ECG recordings increases if the monitoring duration is increased. Adhesive patch
sensors have been demonstrated to be superior to Holter and event monitors, mainly because of their
longer study period.
The ECG signals are mainly acquired using wet, dry and non-contact capacitive electrodes. The choice
of electrodes for biomedical applications, especially ECG monitoring, depends not only on the comfort
that they offer to the patient, but also on the quality of the signal obtained. The skin-electrode interface
decides the operational characteristics of any electrode system in conjunction with the properties of
the electrode material.
Various sources of noise in ambulatory ECG monitoring such as main power frequency and common
mode voltage interference can be minimized by the implementation of notch filter and driven right leg
in the ECG circuitry respectively. Noise rejection in ECG signals using ICA has proven to be more
effective than adaptive filtering, wavelet transform and PCA. However, the outstanding issue is the
sensitivity to motion artefacts. This may hinder the cardiac diagnosis and lead to inappropriate
treatment decisions. Skin stretch is a major physiological source of motion artefact in ECG due to the
flow of injury current across the barrier layer of the skin. The amplitude and frequency range of motion
artefacts due to skin stretch is comparable to ECG, therefore it is difficult to identify and eliminate
them. Applying pressure on the electrodes can aid in minimizing these artefacts, however, the ECG
signals can deteriorate if more pressure is applied. Despite the numerous efforts made in reducing
motion artefacts from ECG signals using motion tracking, the underlying issue due to the flow of
injury current hasn’t been explicitly addressed in literature elsewhere.
Fig. 1 Formation of Einthoven’s triangle with heart at the centre of the three leads-Lead I, Lead II
and Lead III.
Fig. 2 Electrode placement for a 12-lead ECG configuration, with electrodes on right arm (RA), left
arm (LA), left leg (LL), Right leg (RL) and chest electrodes V1 to V6 [reproduced from 134].
Fig. 3 Common examples of abnormal ECGs [11]; (a) Normal Sinus Rhythm; (b) Ventricular
Fibrillation; (c) Atrioventricular Block; (d) Premature Ventricular Contraction; (e) Atrial Flutter; (f)
Atrial Fibrillation. [reproduced from 11].
Fig. 4 (a) Wet silver-silver chloride electrodes; (b) dry contact electrodes; (c) Non-contact capacitive
Fig. 5 (a) ECG with motion artefacts; (b) ECG at rest.
Fig. 6 Schematic diagram of the skin [65]
Fig. 7 Schematic diagram of ECG detection circuit
Fig. 8 Use of a driven right leg circuit to reduce common mode voltage [reproduced from 85].
Fig. 9 Schematic diagram of an isolated biomedical circuit [86].
Fig. 10 Signal denoising using adaptive filtering
Fig. 11 Depiction of electrode locations [127]
Fig. 12 Diagrammatic representation of the coordinate system in a 3-axis accelerometer.
Fig. 13 Single Inductive Linear Variable Differential Transformers.
Fig. 14 Parameters affecting the source capacitance of the gradiometer (quad configuration).
[1] C. Bowden and J. Masters, Veterinary nursing medical textbook. Oxford: Butterworth-Heinemann,
[2] S. Korsakas et al., “Electrocardiosignals and motion signals telemonitoring and analysis system for
sportsmen,” in Computers in Cardiology, 2005, 2005, pp. 363366.
[3] C. Matteucci, “Sur le courant électrique ou propre de la grénouille,” Bibl. Univ. Genève, vol. 15, pp.
157168, 1838.
[4] A. D. Waller, “A demonstration on man of electromotive changes accompanying the heart’s beat,”
The Journal of physiology, vol. 8, no. 5, pp. 229–234, 1887.
[5] W. Einthoven, Un nouveau galvanometre. Soc. Holl. des Sciences, 1901.
THEIR SIGNIFICATION.,” The Lancet, vol. 179, no. 4622, pp. 853861, 1912.
[7] C. L. Stanfield, W. J. Germann, M. J. Niles, and J. G. Cannon, Principles of human physiology. San
Francisco: Pearson/Benjamin Cummings, 2009.
[8] W. A. N. Dorland, Dorland’s illustrated medical dictionary. Philadelphia, PA: Saunders, 2007.
[9] “Definition of Sinus rhythm,” MedicineNet. [Online]. Available: [Accessed: 28-Dec-2017].
[10] “Categories of Arrhythmias - Texas Heart Institute Heart Information Center.” [Online]. Available: [Accessed: 28-Dec-2017].
[11] “Academic paper (PDF): Bioelectromagnetism - Principles and Applications of Bioelectric and
Biomagnetic Fields,” ResearchGate. [Online]. Available:
_Principles_and_Applications_of_Bioelectric_and_Biomagnetic_Fields. [Accessed: 12-Apr-2017].
[12] H. J. Wellens and A. P. M. Gorgels, “How Important Is the Electrocardiogram in Protecting and
Guiding the Athlete?,” Circulation, vol. 124, no. 6, pp. 669671, Aug. 2011.
[13] K. G. Harmon, I. M. Asif, D. Klossner, and J. A. Drezner, “Incidence of Sudden Cardiac Death in
National Collegiate Athletic Association AthletesClinical Perspective,” Circulation, vol. 123, no. 15, pp.
1594–1600, 2011.
[14] A. Uberoi et al., “Interpretation of the electrocardiogram of young athletes,” Circulation, vol. 124, no.
6, pp. 746–757, 2011.
[15] E.-M. Fong and W.-Y. Chung, “Mobile Cloud-Computing-Based Healthcare Service by Noncontact
ECG Monitoring,” Sensors, vol. 13, no. 12, pp. 1645116473, 2013.
[16] Y. Liu, V. Eveloy, and M. Pecht, “Non-invasive electrocardiogram and blood pressure monitors for
cardiovascular disease,” in 38th International Symposium on Microelectronics, Sensor and MEMS Packaging,
2005, pp. 66–74.
[17] “Cardiovascular Medical Device Series: Introduction and Technical Trends in the Cardiac Monitoring
Industry |” [Online]. Available: [Accessed: 29-Apr-2017].
[18] H. Boudoulas, S. F. Schaal, R. P. Lewis, J. L. Robinson, and R. S. Goodwin, “Superiority of 24-hour
outpatient monitoring over multi-stage exercise testing for the evaluation of syncope,” Journal of
electrocardiology, vol. 12, no. 1, pp. 103108, 1979.
[19] P. F. Poblete, H. L. Kennedy, and D. G. Caralis, “Detection of ventricular ectopy in patients with
coronary heart disease and normal subjects by exercise testing and ambulatory electrocardiography,” Chest,
vol. 74, no. 4, pp. 402–407, 1978.
[20] “Ambulatory ECG Monitoring and Related Investigations AECG. Patient.” [Online]. Available: [Accessed: 29-Mar-2017].
[21] B. Surawicz and R. P. Pinto, “Symptoms in hospital patients and outpatients with ventricular
arrhythmias during ambulatory ECG monitoring,” J Ambul Monit, vol. 4, p. 83, 1991.
[22] “Holter Monitor.” [Online]. Available:
Monitor_UCM_446437_Article.jsp#.WNuT5mW9qpo. [Accessed: 29-Mar-2017].
[23] S. M. Zeldis, B. J. Levine, E. L. Michelscn, and J. Morganroth, “Cardiovascular complaints:
correlation with cardiac arrhythmias on 24-hour electrocardiographic monitoring,” Chest, vol. 78, no. 3, pp.
456462, 1980.
[24] T. Drake, “Diagnostic efficacy of 24-hour electrocardiographic monitoring for syncope: Gibson TC,
Heitzman MR. Am J Cardiol 1984; 53: 1013–1017,” The Journal of Emergency Medicine, vol. 2, no. 1, p. 65,
[25] E. B. Bass et al., “The duration of Holter monitoring in patients with syncope: is 24 hours enough?,”
Archives of Internal Medicine, vol. 150, no. 5, pp. 1073–1078, 1990.
[26] “Ambulatory ECG monitoring - UpToDate.” [Online]. Available: [Accessed: 30-Mar-2017].
[27] A. K. Joshi et al., “First experience with a Mobile Cardiac Outpatient Telemetry (MCOT) system for
the diagnosis and management of cardiac arrhythmia,” The American journal of cardiology, vol. 95, no. 7, pp.
878881, 2005.
[28] J. A. Olson, A. M. Fouts, B. J. Padanilam, and E. N. Prystowsky, “Utility of mobile cardiac outpatient
telemetry for the diagnosis of palpitations, presyncope, syncope, and the assessment of therapy efficacy,”
Journal of cardiovascular electrophysiology, vol. 18, no. 5, pp. 473477, 2007.
[29] F. Giada et al., “Recurrent unexplained palpitations (RUP) study: comparison of implantable loop
recorder versus conventional diagnostic strategy,” Journal of the American College of Cardiology, vol. 49, no.
19, pp. 1951–1956, 2007.
[30] T. Furukawa, R. Maggi, C. Bertolone, D. Fontana, and M. Brignole, “Additional diagnostic value of
very prolonged observation by implantable loop recorder in patients with unexplained syncope,” Journal of
cardiovascular electrophysiology, vol. 23, no. 1, pp. 6771, 2012.
[31] M. P. Turakhia et al., “Diagnostic utility of a novel leadless arrhythmia monitoring device,” The
American journal of cardiology, vol. 112, no. 4, pp. 520524, 2013.
[32] P. M. Barrett et al., “Comparison of 24-hour Holter monitoring with 14-day novel adhesive patch
electrocardiographic monitoring,” The American journal of medicine, vol. 127, no. 1, pp. 95–e11, 2014.
[33] “SEEQ Mobile Cardiac Telemetry System | Diagnostics & Monitoring.” [Online]. Available: [Accessed: 29-Apr-
[34] C. V. Pollack Jr, “Wireless cardiac event alert monitoring is feasible and effective in the emergency
department and adjacent waiting areas,” Critical pathways in cardiology, vol. 8, no. 1, pp. 711, 2009.
[35] “Intelesens | Responsive Healthcare - V-Patch.” [Online]. Available: [Accessed: 30-Jul-2014].
[36] “Solutions Library | Engineering for Change.” [Online]. Available: [Accessed: 15-Jul-2014].
[37] I. Romero et al., “Motion artifact reduction in ambulatory ECG monitoring: an integrated system
approach,” in Proceedings of the 2nd Conference on Wireless Health, 2011, p. 11.
[38] E. Fung et al., “Electrocardiographic patch devices and contemporary wireless cardiac monitoring,”
Front Physiol, vol. 6, May 2015.
[39] J. S. Shinbane, M. Merkert, R. Fogoros, V. Mehta, M. Cao, and L. A. Saxon, “Wearable wireless
arrhythmia detection patches: diagnostic arrhythmia yield, time to first arrhythmia, and patient compliance,”
Heart Rhythm, vol. 10, pp. 5S–S305, 2013.
[40] M. A. Rosenberg, M. Samuel, A. Thosani, and P. J. Zimetbaum, “Use of a noninvasive continuous
monitoring device in the management of atrial fibrillation: a pilot study,” Pacing and Clinical
Electrophysiology, vol. 36, no. 3, pp. 328333, 2013.
[41] “Atrial Fibrillation and Cryptogenic Stroke: Is There a Link? Rhea C. Pimentel, MD Associate
Professor of Medicine Fellowship Program Director, Clinical. - Documents,” [Online].
pimentel.html. [Accessed: 28-Apr-2017].
[42] Plessey Semiconductors, “Application Note # 291566 Non-contact ECG measurement using EPIC.”
[43] C. J. Harland, T. D. Clark, N. S. Peters, M. J. Everitt, and P. B. Stiffell, “A compact electric potential
sensor array for the acquisition and reconstruction of the 7-lead electrocardiogram without electrical charge
contact with the skin,” Physiological Measurement, vol. 26, no. 6, pp. 939950, Dec. 2005.
[44] “The bumps have it when it comes to CardioWare | Cardiovascular content from Medical Design.”
[Online]. Available:
[Accessed: 15-Jul-2014].
[45] “Maxim ‘Fit’ shirt,” EDN. [Online]. Available: [Accessed: 15-Jul-2014].
[46] M. K. Delano, “A long term wearable electrocardiogram (ecg) measurement system,” Massachusetts
Institute of Technology, 2012.
[47] T.-H. Kang, “Textile-embedded sensors for wearable physiological monitoring systems,” 2007.
[48] Z. Ping, L. Zhoucheng, W. Feng, and J. Hongyu, “Non-contact ECG Monitoring Based on Capacitive
Electrodes,” in World Congress on Medical Physics and Biomedical Engineering May 26-31, 2012, Beijing,
China, 2013, pp. 15061509.
[49] Y. M. Chi, Tzyy-Ping Jung, and G. Cauwenberghs, “Dry-Contact and Noncontact Biopotential
Electrodes: Methodological Review,” IEEE Reviews in Biomedical Engineering, vol. 3, pp. 106119, 2010.
[50] L. A. Geddes, L. E. Baker, and A. G. Moore, “Optimum electrolytic chloriding of silver electrodes,”
Med. & biol. Engng., vol. 7, no. 1, pp. 4956, Jan. 1969.
[51] M. Catrysse, R. Puers, C. Hertleer, L. Van Langenhove, H. van Egmond, and D. Matthys, “Towards
the integration of textile sensors in a wireless monitoring suit,” Sensors and Actuators A: Physical, vol. 114,
no. 23, pp. 302–311, Sep. 2004.
[52] F. Carpi and D. De-Rossi, “Electroactive polymer-based devices for e-textiles in biomedicine,” IEEE
Transactions on Information Technology in Biomedicine, vol. 9, no. 3, pp. 295318, Sep. 2005.
[53] M. Coyle, “Ambulatory cardiopulmonary data capture,” in Microtechnologies in Medicine amp;
Biology 2nd Annual International IEEE-EMB Special Topic Conference on, 2002, pp. 297300.
[54] D. Marculescu et al., “Electronic textiles: A platform for pervasive computing,” Proceedings of the
IEEE, vol. 91, no. 12, pp. 1995–2018, Dec. 2003.
[55] E. Jovanov, A. O’Donnell Lords, D. Raskovic, P. G. Cox, R. Adhami, and F. Andrasik, “Stress
monitoring using a distributed wireless intelligent sensor system,” IEEE Engineering in Medicine and Biology
Magazine, vol. 22, no. 3, pp. 4955, May 2003.
[56] S. Ghosh and O. Inganäs, “Conducting polymer hydrogels as 3D electrodes: applications for
supercapacitors,” Advanced Materials, vol. 11, no. 14, pp. 12141218, 1999.
[57] S. M. Lee et al., “Self-adhesive epidermal carbon nanotube electronics for tether-free long-term
continuous recording of biosignals,” Scientific Reports, vol. 4, p. 6074, Aug. 2014.
[58] “Carbon nanotube,” ScienceDaily. [Online]. Available: [Accessed: 16-Feb-2017].
[59] J. G. Webster, “Reducing Motion Artifacts and Interference in Biopotential Recording,” IEEE
Transactions on Biomedical Engineering, vol. BME-31, no. 12, pp. 823826, Dec. 1984.
[60] N. V. Thakor and J. G. Webster, “Electrode studies for the long-term ambulatory ECG,” Med. Biol.
Eng. Comput., vol. 23, no. 2, pp. 116121, Mar. 1985.
[61] J. a. J. Klijn and M. J. G. M. Kloprogge, “Movement artefact suppressor during ECG monitoring,”
Cardiovasc Res, vol. 8, no. 1, pp. 149152, Jan. 1974.
[62] Y. M. Chi, S. R. Deiss, and G. Cauwenberghs, “Non-contact Low Power EEG/ECG Electrode for
High Density Wearable Biopotential Sensor Networks,” 2009, pp. 246250.
[63] M. H. Limaye and M. V. Deshmukh, “ECG Noise Sources and Various Noise Removal Techniques:
A Survey.”
[64] “Artefact - Types and Causes of Interferancen - Cardiology Teaching Package - Practice Learning -
Division of Nursing - The University of Nottingham.” [Online]. Available: [Accessed: 29-
[65] Y. Liu, “Reduction of skin stretch induced motion artifacts in electrocardiogram monitoring using
adaptive filtering,” 2007.
[66] S. Ödman, “Potential and impedance variations following skin deformation,” Med. Biol. Eng.
Comput., vol. 19, no. 3, pp. 271278, May 1981.
[67] S. Odman, “On the spread of deformation potentials in the skin,” Med Biol Eng Comput, vol. 20, no.
4, pp. 451–456, Jul. 1982.
[68] S. Ödman and P. \AAke Öberg, “Movement-induced potentials in surface electrodes,” Medical and
Biological Engineering and Computing, vol. 20, no. 2, pp. 159166, 1982.
[69] H. de Talhouet and J. G. Webster, “The origin of skin-stretch-caused motion artifacts under
electrodes,” Physiological Measurement, vol. 17, no. 2, p. 81, 1996.
[70] S. Ödman, “Changes in skin potentials induced by skin compression,Medical and Biological
Engineering and Computing, vol. 27, no. 4, p. 390, 1989.
[71] R. Groves, “Quantifying the mechanical properties of skin in vivo and ex vivo to optimise
microneedle device design,” Cardiff University, 2012.
[72] “E-QURE - The Science of Bioelectricity.” [Online]. Available:
of-bioelectricity/the-science-of-bioelectricity. [Accessed: 02-Apr-2017].
[73] N. Thakor and J. Webster, “The origin of skin potential and its variations,” in Proc. Ann. Conf. Eng.
Biol. Med, 1978, vol. 20, p. 212.
[74] H. Tam and J. G. Webster, “Minimizing Electrode Motion Artifact by Skin Abrasion,IEEE
Transactions on Biomedical Engineering, vol. BME-24, no. 2, pp. 134–139, Mar. 1977.
[75] D. Vala, T. Pawar, and V. K. Thakar, “Motion Artifact removal in Ambulatory ECG Signal using
[76] N. V. Thakor and Y.-S. Zhu, “Applications of adaptive filtering to ECG analysis: noise cancellation
and arrhythmia detection,” IEEE Transactions on Biomedical Engineering, vol. 38, no. 8, pp. 785794, Aug.
[77] K. A, L. A, and A.-J. Am, “Quantifying Skin Stretch induced Motion Artifact from an
Electrocardiogram signal-A Pilot Study,” Journal of Biosensors & Bioelectronics, vol. 7, no. 2, pp. 17, Apr.
[78] A. Lanata, A. Guidi, P. Baragli, G. Valenza, and E. P. Scilingo, “A Novel Algorithm for Movement
Artifact Removal in ECG Signals Acquired from Wearable Systems Applied to Horses,” PLOS ONE, vol. 10,
no. 10, p. e0140783, Oct. 2015.
[79] F. Strasser, M. Muma, and A. M. Zoubir, “Motion artifact removal in ECG signals using multi-
resolution thresholding,” in Signal Processing Conference (EUSIPCO), 2012 Proceedings of the 20th
European, 2012, pp. 899–903.
[80] Q. Xue, Y. H. Hu, and W. J. Tompkins, “Neural-network-based adaptive matched filtering for QRS
detection,” IEEE Transactions on Biomedical Engineering, vol. 39, no. 4, pp. 317329, Apr. 1992.
[81] T. Kiryu, H. Kaneko, and Y. Saitoh, “Motion Artifact Elimination Using Fuzzy Rule Based Adaptive
Nonlinear Filter,” IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer
Sciences, vol. E77-A, no. 5, pp. 833838, May 1994.
[82] M. R. Neuman, “Biopotential amplifiers,” Medical instrumentation: application and design, pp. 292
296, 1998.
[83] J. Webster, Medical instrumentation: application and design. John Wiley & Sons, 2009.
[84] “ECG Circuit Analysis and Design « Engineers Labs.” .
[85] A. C. M. van Rijn, A. Peper, and C. A. Grimbergen, “High-quality recording of bioelectric events,”
Med. Biol. Eng. Comput., vol. 29, no. 4, pp. 433440, Jul. 1991.
[86] A. Metting van Rijn, A. Peper, and C. . Grimbergen, “The isolation mode rejection ratio in bioelectric
amplifiers,” IEEE Transactions on Biomedical Engineering, vol. 38, no. 11, pp. 11541157, Nov. 1991.
[87] J. C. Huhta and J. G. Webster, “60-Hz Interference in Electrocardiography,” IEEE Transactions on
Biomedical Engineering, vol. BME-20, no. 2, pp. 91101, Mar. 1973.
[88] N. V. Thakor and J. G. Webster, “Ground-Free ECG Recording with Two Electrodes,” IEEE
Transactions on Biomedical Engineering, vol. BME-27, no. 12, pp. 699704, Dec. 1980.
[89] I. C. Forster, “Measurement of patient body capacitance and a method of patient isolation in mains
environments,” Medical and Biological Engineering and Computing, vol. 12, no. 5, pp. 730732, 1974.
[90] B. B. Winter and J. G. Webster, “Driven-right-leg circuit design,” IEEE Transactions on Biomedical
Engineering, no. 1, pp. 6266, 1983.
[91] B. B. Winter and J. G. Webster, “Reductionl of interference due to common mode voltage in
biopotential amplifiers,” IEEE Transactions on Biomedical Engineering, no. 1, pp. 5862, 1983.
[92] C. Levkov, “Amplification of biosignals by body potential driving,” Medical and Biological
Engineering and Computing, vol. 20, no. 2, pp. 248–250, 1982.
[93] C. K. Chui, An introduction to wavelets. Boston: Academic Press, 1992.
[94] A. N. Akansu, W. A. Serdijn, and I. W. Selesnick, “Emerging applications of wavelets: A review,”
Physical Communication, vol. 3, no. 1, pp. 118, Mar. 2010.
[95] Ho Tatt Wei and V. Jeoti, “A wavelet footprints-based compression scheme for ECG signals,” 2004,
vol. A, pp. 283286.
[96] I. Provazník, J. Kozumplík, J. Bardonova, M. Nováková, and Z. Nováková, “Wavelet transform in
ECG signal processing,” in Proceeding of the 15th Biennial Eurasip Conference BIOSIGNAL, 2000, pp. 21
[97] J. S. Sahambi, S. N. Tandon, and R. K. P. Bhatt, “Using wavelet transforms for ECG characterization.
An on-line digital signal processing system,” IEEE Engineering in Medicine and Biology Magazine, vol. 16,
no. 1, pp. 77–83, 1997.
[98] M. Alfaouri and K. Daqrouq, “ECG Signal Denoising By Wavelet Transform Thresholding,”
American Journal of Applied Sciences, vol. 5, no. 3, pp. 276–281, Mar. 2008.
[99] M. Raghuram, K. V. Madhav, E. H. Krishna, and K. . Reddy, “On the Performance of Wavelets in
Reducing Motion Artifacts from Photoplethysmographic Signals,” in 2010 4th International Conference on
Bioinformatics and Biomedical Engineering (iCBBE), 2010, pp. 14.
[100] J. Y. A. Foo, “Comparison of wavelet transformation and adaptive filtering in restoring artefact-
induced time-related measurement,” Biomedical Signal Processing and Control, vol. 1, no. 1, pp. 9398, Jan.
[101] B. Widrow and M. E. Hoff, “Adaptive switching circuits,” STANFORD UNIV CA STANFORD
[102] M. A. Raya and L. G. Sison, “Adaptive noise cancelling of motion artifact in stress ECG signals using
accelerometer,” in Engineering in Medicine and Biology, 2002. 24th Annual Conference and the Annual Fall
Meeting of the Biomedical Engineering Society EMBS/BMES Conference, 2002. Proceedings of the Second
Joint, 2002, vol. 2, pp. 17561757 vol.2.
[103] S. Luo and W. J. Tompkins, “Experimental study: brachial motion artifact reduction in the ECG,” in
Computers in Cardiology 1995, 1995, pp. 3336.
[104] D. . Tong, K. . Bartels, and K. S. Honeyager, “Adaptive reduction of motion artifact in the
electrocardiogram,” in Engineering in Medicine and Biology, 2002. 24th Annual Conference and the Annual
Fall Meeting of the Biomedical Engineering Society EMBS/BMES Conference, 2002. Proceedings of the
Second Joint, 2002, vol. 2, pp. 1403–1404 vol.2.
[105] E. R. Ferrara and B. Widraw, “Fetal electrocardiogram enhancement by time-sequenced adaptive
filtering,” IEEE Transactions on Biomedical Engineering, no. 6, pp. 458–460, 1982.
[106] M. Z. U. Rahman, R. A. Shaik, and D. Reddy, “Noise cancellation in ECG signals using
computationally simplified adaptive filtering techniques: Application to biotelemetry,” Signal Processing: An
International Journal (SPIJ), vol. 3, no. 5, pp. 112, 2009.
[107] F. Castells, P. Laguna, L. Sörnmo, A. Bollmann, and J. M. Roig, “Principal Component Analysis in
ECG Signal Processing,” EURASIP Journal on Advances in Signal Processing, vol. 2007, no. 1, p. 074580,
Feb. 2007.
[108] M. P. S. Chawla, H. K. Verma, and V. Kumar, “RETRACTED: A new statistical PCAICA
algorithm for location of R-peaks in ECG,” International journal of cardiology, vol. 129, no. 1, pp. 146148,
[109] M. P. S. Chawla, “PCA and ICA processing methods for removal of artifacts and noise in
electrocardiograms: A survey and comparison,” Applied Soft Computing, vol. 11, no. 2, pp. 2216–2226, 2011.
[110] T. Stamkopoulos, K. Diamantaras, N. Maglaveras, and M. Strintzis, “ECG analysis using nonlinear
PCA neural networks for ischemia detection,” IEEE Transactions on Signal Processing, vol. 46, no. 11, pp.
3058–3067, 1998.
[111] T.-P. Jung et al., “Independent component analysis of biomedical signals,” in Proc. Int. Workshop on
Independent Component Analysis and Signal Separation, 2000, pp. 633644.
[112] P. Gao, E.-C. Chang, and L. Wyse, “Blind separation of fetal ECG from single mixture using SVD
and ICA,” in Information, Communications and Signal Processing, 2003 and Fourth Pacific Rim Conference
on Multimedia. Proceedings of the 2003 Joint Conference of the Fourth International Conference on, 2003,
vol. 3, pp. 14181422.
[113] R. Vigário, J. Sarela, V. Jousmiki, M. Hamalainen, and E. Oja, “Independent component approach to
the analysis of EEG and MEG recordings,” IEEE transactions on biomedical engineering, vol. 47, no. 5, pp.
589593, 2000.
[114] M. E. Davies and C. J. James, “Source separation using single channel ICA,” Signal Processing, vol.
87, no. 8, pp. 18191832, 2007.
[115] A. Cichocki and S. Amari, Adaptive blind signal and image processing: learning algorithms and
applications, vol. 1. John Wiley & Sons, 2002.
[116] P. Ramaswamy, “Uni-channel PCA for noise reduction from ECG signals.”
[117] I. Romero, “PCA and ICA applied to noise reduction in multi-lead ECG,” in 2011 Computing in
Cardiology, 2011, pp. 613–616.
[118] A. u Rehman, F. Khan, and B. K. Jadoon, “Analysis of adaptive filter and ICA for noise cancellation
from a video frame,” in 2016 International Conference on Intelligent Systems Engineering (ICISE), 2016, pp.
[119] S. Abbaspour, H. Gholamhosseini, and M. Linden, “Evaluation of Wavelet Based Methods in
Removing Motion Artifact from ECG Signal,” SpringerLink, pp. 14, 2015.
[120] B. S. Kim and S. K. Yoo, “Motion artifact reduction in photoplethysmography using independent
component analysis,” IEEE Transactions on Biomedical Engineering, vol. 53, no. 3, pp. 566568, Mar. 2006.
[121] J. Y. . Foo, “Use of Independent Component Analysis to Reduce Motion Artifact in Pulse Transit
Time Measurement,” IEEE Signal Processing Letters, vol. 15, pp. 124126, 2008.
[122] D. P. Burbank and J. G. Webster, “Reducing skin potential motion artefact by skin abrasion,Med.
Biol. Eng. Comput., vol. 16, no. 1, pp. 3138, Jan. 1978.
[123] E. P. Scilingo, A. Gemignani, R. Paradiso, N. Taccini, B. Ghelarducci, and D. De Rossi,
“Performance evaluation of sensing fabrics for monitoring physiological and biomechanical variables,” IEEE
Transactions on information technology in biomedicine, vol. 9, no. 3, pp. 345–352, 2005.
[124] A. T. Vehkaoja, J. A. Verho, and J. O. Lekkala, “Miniature wireless measurement node for ECG
signal transmission in home area network,” in Engineering in Medicine and Biology Society, 2006. EMBS’06.
28th Annual International Conference of the IEEE, 2006, pp. 20492052.
[125] L. Beckmann et al., “Characterization of textile electrodes for bioimpedance spectroscopy,” in
Proceedings of Ambience 2008: International Scientific Conference Smart TextileTechnology and Design,
2008, pp. 79–83.
[126] S. Kim et al., “Influence of contact pressure and moisture on the signal quality of a newly developed
textile ECG sensor shirt,” in Medical Devices and Biosensors, 2008. ISSS-MDBS 2008. 5th International
Summer School and Symposium on, 2008, pp. 256259.
[127] A. Cömert, M. Honkala, and J. Hyttinen, “Effect of pressure and padding on motion artifact of textile
electrodes,” Biomedical engineering online, vol. 12, no. 1, p. 26, 2013.
[128] S. W. Yoon, S. D. Min, Y. H. Yun, S. Lee, and M. Lee, “Adaptive Motion Artifacts Reduction Using
3-axis Accelerometer in E-textile ECG Measurement System,” Journal of Medical Systems, vol. 32, no. 2, pp.
101106, Apr. 2008.
[129] Y. Kishimoto, Y. Kutsuna, and K. Oguri, “Detecting Motion Artifact ECG Noise During Sleeping by
Means of a Tri-axis Accelerometer,” in 2007 29th Annual International Conference of the IEEE Engineering
in Medicine and Biology Society, 2007, pp. 26692672.
[130] S.-H. Liu, “Motion artifact reduction in electrocardiogram using adaptive filter,” Journal of Medical
and Biological Engineering, vol. 31, no. 1, pp. 6772, 2011.
[131] G. Peng, “A non-contact, low-noise electrocardiogram sensor ASIC employing motion artifact
reduction,” 2014.
[132] A. Yilmaz, O. Javed, and M. Shah, “Object Tracking: A Survey,” ACM Comput. Surv., vol. 38, no. 4,
Dec. 2006.
[133] H. Matsumura, N. Yoshizawa, K. Watanabe, and N. B. Vedder, “Preconditioning of the distal portion
of a rat random-pattern skin flap,” British journal of plastic surgery, vol. 54, no. 1, pp. 58–61, 2001.
[134] X. Gao, “Non-invasive Detection and Compression of Fetal Electrocardiogram,” 2017.
View publication statsView publication stats
... The 12-lead ECG configuration gives spatial information about the cardiac electric activity (Kalra, Lowe, and Al-Jumaily, 2019). Since ECG leads have both positive and negative poles, they may be viewed from two spatial directions. ...
... Ambulatory 24 h ECG monitoring-Holter monitor-is employed in diagnostics of paroxysmal cardiac events (such as paroxysmal arrhythmias). The Holter monitor may use a 12-lead system; however, modern devices record two or three modified leads only (Kalra, Lowe, and Al-Jumaily, 2019). In case of rare symptoms, an implantable loop recorder is valuable (Giada et al., 2007). ...
... In case of rare symptoms, an implantable loop recorder is valuable (Giada et al., 2007). Such long-term monitor is placed under the skin on the chest and can automatically record long continuous signals (up to 3 years) (Kalra, Lowe, and Al-Jumaily, 2019). Recently, several innovative methods for ECG monitoring were introduced, including patch sensors, EPIC (electric potential integrated circuit) sensors, chest harnesses, and vest shirts (Kalra, Lowe, and Al-Jumaily, 2019;Soroudi et al., 2019). ...
Full-text available
Cardiovascular system and its functions under both physiological and pathophysiological conditions have been studied for centuries. One of the most important steps in the cardiovascular research was the possibility to record cardiac electrical activity. Since then, numerous modifications and improvements have been introduced; however, an electrocardiogram still represents a golden standard in this field. This paper overviews possibilities of ECG recordings in research and clinical practice, deals with advantages and disadvantages of various approaches, and summarizes possibilities of advanced data analysis. Special emphasis is given to state-of-the-art deep learning techniques intensely expanded in a wide range of clinical applications and offering promising prospects in experimental branches. Since, according to the World Health Organization, cardiovascular diseases are the main cause of death worldwide, studying electrical activity of the heart is still of high importance for both experimental and clinical cardiology.
... Statistical reports indicated that the leading cause of death in the world comes from cardiovascular diseases [9,20]. The World Health Organization (WHO) reported that the total number of deaths from cardiovascular diseases in 2012 was approximately 17.5 million, compared with 17.7 million in 2015, and this number has been increasing every year [1,3,9]. ...
... Hence, monitoring ECG and performing automatic diagnosis become particularly important. In cardiology, the electrical actions of a human's heart are simply and painlessly recorded by electrocardiogram (ECG) via single or multiple-lead detections [8,20]. The real-time ECG sequence of a patient represents one of the most useful clinical diagnostic features on cardiovascular diseases, reflecting the electrophysiological activity of cardiac excitement, and indicating great importance on the aspects of basic heart functions and related pathological research [12]. ...
... Previously reported medications and medical procedures such as pacemaker insertion and surgery offer well-established treatments for most arrhythmias; meanwhile, a large quantity of signal and image processing algorithms as well as sensor devices provided useful tools on electrocardiogram-assisted diagnosis [8,18,20,26,31,32]. Recently, many researchers have been devoting themselves on computer-aided ECG analysis, where the technical developments are enriched from the booming growth on machine learning and deep learning algorithms [6, 9, 11, 13-17, 21, 24, 25, 27, 30, 33, 35-37]. ...
... An electrocardiogram (ECG) records the abundant physiological information of the heart, which has been widely applied in the diagnosis for a variety of cardiovascular diseases. Te research of the analytical method of ECG is always a hot spot in the feld of biomedical signal processing [1][2][3][4][5]. Te nonlinear analyses are usually employed to analyze ECG since it refects a nonlinear characteristic of the biomedical signal. ...
... Te values of CCTM(r) in four quadrants are displayed in Figure 8. Figure 8(a) shows the values of CCTM1(r) in the frst quadrant, which indicate a substantial diference when the r ∈ [2,5]. Figures 8(b)∼8(d) are CCTM2(r), CCTM3(r), and CCTM4(r), respectively. ...
Full-text available
In this study, an improved second-order difference plot is proposed to analyze the variability of heart rate variability. Although the variation of the physiological status of the cardiovascular system can be shown graphically by the second-order difference plot, the descriptive ability of existing indicators for this plot is insufficient. As a result, the physiological information contained in the second-order difference plot cannot be extracted adequately. Addressing the problem, the temporal variation measure analysis is presented to describe distribution patterns of scatter points in the second-order difference plot quantitatively and extract the acceleration information for a variation of heart rate variability. Experiment results demonstrate the effectiveness of the temporal variation measure analysis. As a quantitative indicator, the temporal variation entropy is properly designed and successfully applied in the recognition and classification of the physiological statuses of the heart.
... Cardiovascular diseases and strokes remain the two biggest causes of death worldwide [1,2]. Thus, monitoring and recording electrocardiogram (ECG) and electroencephalogram (EEG), bio-potential signals from the heart and brain of patients, are essential in the analysis of various physiological parameters and diagnosis [3,4]. An effective method for monitoring these signals while patients are moving around to do their daily activities is to utilize non-invasive, wearable, and implantable systems [1,[3][4][5][6][7][8]. ...
... Thus, monitoring and recording electrocardiogram (ECG) and electroencephalogram (EEG), bio-potential signals from the heart and brain of patients, are essential in the analysis of various physiological parameters and diagnosis [3,4]. An effective method for monitoring these signals while patients are moving around to do their daily activities is to utilize non-invasive, wearable, and implantable systems [1,[3][4][5][6][7][8]. ...
Full-text available
Recently, due to their abundant benefits, current-mode instrumentation amplifiers have received considerable attention in medical instrumentation and read-out circuit for biosensors. This paper is focused on the design of current-mode instrumentation amplifiers for portable, implantable, and wearable electrocardiography and electroencephalography applications. To this end, a CMOS differential voltage second-generation current conveyor (DVCCII) based on a linear transconductor is presented. A new band-pass instrumentation amplifier, based on the designed DVCCII, is also implemented in this paper. The concept of the proposed differential voltage current conveyor and instrumentation amplifier is validated numerically and their predicted performance is presented. The simulation results of the presented circuits were tested for 0.18 μm TSMC CMOS technology in a post layout simulation level using the Cadence Virtuoso tool with a ±0.9 V power supply, and demonstrated that the designed DVCCII has a wide dynamic range of ±400 mV and ±0.85 mA and a power consumption of 148 μW. The layout of the DVCCII circuit occupies a total area of 0.378 μm2. It is shown that the designed DVCCII benefits from good linearity over a wide range of input signals and provides a low input impedance at terminal X. Two versions of the proposed band-pass instrumentation amplifier using pseudo resistances were designed with different specifications for two different applications, namely for EEG and ECG signals. Numerical analyses of both designs show proper outputs and frequency responses by eliminating the undesired artifact and DC component of the EEG and ECG input signals.
... Non-contact ECG measurement is an advanced sensing technique for detection of cardiac signals that uses capacitive electrodes to sense ECG signals via dielectric materials, including hair, cloth, etc [1][2][3]. Comparing to dry -contact electrods, non-contact capacitive sensing provides a convenient method without the advance preparation. The capacitive electrode was first presented by Richardson in 1967 and utilized a thin aluminum oxide layer as a dielectric to detect the bioelectric signals [4]. ...
... Our system generates the recognizable heartbeat at the small limb-electrode capacitance above 8.5 pF and the maximum through-thickness of 0.4 mm by peak detection algorithm. According to (1) and (6), high equivalent capacitance tends to provide an excellent capacitive coupling interface between the limb and electrode and signals resolution by theoretical calculation. 2 3.83 0.2 mm 17.0 pF 1cm 2 3.83 0.3 mm 11.3 pF 1cm 2 3.83 0.4 mm 8.5 pF 1cm 2 3.83 0.5 mm 6.8 pF ...
Full-text available
Rat electrocardiography (ECG) is frequently used in biomedical research as models for exploring heart function in a wide variety of experimental conditions. The subcutaneous ECG is a common approach to record rat heart rhythm using implanted needle electrodes to sense the rat ECG signals with the animal under deep anesthesia. However, such an invasive measurement could cause inconvenience due to cumbersome animal preparation, and the anesthetics are likely to interfere with the autonomic regulation of cardiac rhythm. Most studies used the galvanic contact between animal limb and electrode sensing surface to record cardiac signals from small animals. However, the non-contact capacitive ECG sensing approach for small laboratory animals has not been extensively investigated. This study aims to develop a non-contact ECG system to promote the laboratory animal ECG measurement for biomedical research. The method utilizes the capacitive coupling technique to detect cardiac signals in the awake rats (R-wave amplitude of only 0.2 mV) through a non-conductive layer. The proposed system generates non-contact ECG signals with distinguishable R-peaks at the limb-electrode capacitance above 8.5 pF and maximum through-thickness of the non-conductive layer of 0.4 mm for heart rate assessment. In conclusion, this study provides the non-contact ECG monitoring based on capacitive electrodes to improve the throughput of ECG measurement procedures for biomedical research and establish a lower bound of coupling capacitance for non-contact heart rate application. The new method is ideally suited for the rapid evaluation of autonomic regulation of heart rhythm in awake laboratory small animals.
... It is difficult to identify comparable review papers in the literature on EMG noise filtration from ECG signal. Kalra et al. [27] and Luo and Johnston [28] consider the problem to minimize the EMG noise from ECG signal with different methods: Wavelets, Principal Component Analysis, Independent Component Analysis, Low-pass filters. Both materials give common and basic knowledge about existing filters and the topic of suppression of muscle noise is not deeply treated. ...
... Common current clinical applications include: (i) detecting bio-electric events, such as electrocardiography (ECG), electroencephalography (EEG) and electromyography (EMG); (ii) therapeutic modality, for example, transcutaneous electrical nerve stimulation (TENS); (iii) iontophoresis and (iv) bioelectrical impedance analyses (BIA) for the characterization of soft tissue [1,2]. For (i)-(iii), the systems are designed to send a charge at low frequencies, for example: (i) ECG-0.5 to 100 Hz, EEG-0.5 to 42 Hz, EMG-20 to 500 Hz; (ii) TENS-1 to 150 Hz and (iii) iontophoresis-5 to 1000 Hz [3][4][5][6][7]. ...
Measuring electrical activity in the human body is vital in the diagnosis and monitoring of patients; thus, attention to the design of biopotential electrodes is essential. It is important that electrodes are designed accordingly adapting to a specific device and application. By embroidering electrodes, we can tailor the electrode parameters to suit the application and integrate them into textile outfits. However, embroidered electrodes possess unwanted polarizing impedance (Zp) relative to the frequency of the current applied by the system. Dry embroidered electrodes are preferred to wet electrodes for Bioelectrical Impedance Analysis (BIA) recordings providing stable measurements. BIA is a relatively simple and non-invasive technique that measures the resistivity of biological tissue. This research analyzes the impact of embroidery characteristics (i.e., electrode surface area, stitch type, stitch density and stitch length) on embroidered electrodes by identifying the parameters reducing Zp for bioelectrical impedance analysis (BIA) in a dry and wet state. In addition, the influence of the amount of conductive thread utilised for the fabrication of the sample electrodes was studied. For dry electrodes, we identified that a larger electrode surface area, increased stitch length and stitch density reduce Zp. Moreover, it was observed that potentially there is a threshold on the amount of conductive yarn used for the embroidered electrode. In contrary, wet electrodes displayed irregular Zp and thus, it can be concluded that they could potentially impair BIA measurements. In essence, our findings show that embroidered electrodes can be incorporated into wearable BIA systems for health monitoring in home settings.
This paper provides values of the electrical elements of equivalent models for several textile-based electrodes. In ECG recording the skin-electrode-amplifier interface is modelled as an equivalent electrical circuit having one or two parallel C-R networks and series resistance. The electrical properties have been established using a switched dc current source, modulated with a sinusoidal source. For the single C-R electrode model values of resistance ranged from 10 kΩ to 28 MΩ and of capacitance from 0.03 nF to 15 μF. In the double C-R model values of resistance ranged from 1 kΩ to 25 MΩ and of capacitance from 9 pF to 872 μF. Recording of the ECG using textile-based electrodes has recently been explored in the context of wearable technology for telemedicine and remote monitoring. Knowledge of the electrical properties of such electrodes is essential for the design of amplifiers in ECG monitors to meet the IEC 60601 requirements.
Full-text available
The research subject of the article is the methods of locally adaptive filtering of non-stationary (from the point of view of its variance) noise in long-term electrocardiogram (ECG) signals. The goal is to develop locally adaptive algorithms for filtering noise with different a priori unknown levels of variance in real-time for ECG signals recorded with a standard sampling rate of 500 Hz. The tasks to be solved are: to investigate the effectiveness of the developed adaptive ECG filtering algorithms using numerical statistical estimates of processing quality in a wide range of additive Gaussian noise variance variation, to investigate the suppression of real non-stationary electromyographic (EMG) noise, and to analyze the application for normal and pathological ECG signals. The methods are integral and local indicators of the filter quality according to the criteria of the mean square error and the signal-to-noise ratio was obtained using numerical simulation (via Monte Carlo analysis). The following results were obtained: an adaptive method for real-time suppression of non-stationary noise in the ECG is proposed, the one-pass and the two-pass algorithms, and the algorithm with selective depending on the preliminary estimates of noise levels re-filtering have been developed on the method basis. Statistical estimates of the filters' efficiency and analysis of their outputs show a high degree of suppression of the noise with different levels of variance in the ECGs. The distortions absence while processing QRS-complex and high efficiency of suppression of Gaussian and real EMG noise with varying variance are demonstrated. The analysis of the output signals and plots of the local adaptation parameters and the adaptable parameters of the proposed algorithms confirms the high efficiency of filtering. The developed algorithms have been successfully tested for normal and pathological ECG signals. Conclusions. The scientific novelty of the results is the development of a locally adaptive method with noise and signal-dependent filter parameters switching and of the adaptive algorithms based on this method for non-stationary noise reduction in the ECG in real-time. This method does not require time for filter parameters adaptation and a priori information about the noise variance, and it has a high-speed performance in real-time mode.
Objective: The influence of the skin-electrode-amplifier interface and the input impedance of ECG recording amplifiers on transient response performance is investigated using accurate modeling of surface contact electrodes. The paper aims to establish the input impedance requirements of ECG recording amplifiers based on the electrical properties of the electrodes. Approach: The IEC 60601 standard stipulates the performance requirements for electrocardiographs. Analyses and simulation of both dc and ac modes of coupling the electrodes to the input of the amplifier have been undertaken using an accurate double time-constant model of the electrodes in order to establish design constraints for amplifier input impedance to meet this performance specification. These have been backed up with bench tests. Main results: Investigations carried out indicate that the minimum 10 MΩ resistance at the amplifier input recommended in the specification is insufficient when using some modern adhesive electrodes and increasingly popular un-gelled or dry electrodes. Design constraints are established based on the electrical properties of the electrodes. These constraints suggest that the input impedance of the amplifier should be as high as 10 GΩ and the amplifier ac coupled cut-off frequency should not to be higher than 0.05 Hz for all the electrode models considered. Significance: Signal distortions in the form of false S-wave creation and depression of the S-T segment have been observed when the previously recommended 10 MΩ input impedance is used in the amplifier. This distortion can lead to clinical misdiagnosis but can be avoided if the design constraints established in the paper are adopted.
Full-text available
Noninvasive detection of fetal electrocardiogram (FECG) from abdominal ECG recordings is highly dependent on typical statistical signal processing techniques such as independent component analysis (ICA), adaptive noise filtering, and multichannel blind deconvolution. In contrast to the previous multichannel FECG extraction methods, several recent schemes for single-channel FECG extraction such as the extended Kalman filter (EKF), extended Kalman smoother (EKS), template subtraction (TS), and support vector regression (SVR) for detecting R waves on ECG, are evaluated via the quantitative metrics such as sensitivity (SE), positive predictive value (PPV), F-score, detection error rate (DER), and range of accuracy. A correlation predictor that combines with multivariable gray model (GM) is also proposed for sequential ECG data compression, which displays better percent root mean-square difference (PRD) than those of Sabah’s scheme for fixed and predicted compression ratio (CR). Automatic calculation on fetal heart rate (FHR) on the reconstructed FECG from mixed signals of abdominal ECG recordings is also experimented with sample synthetic ECG data. Sample data on FHR and T/QRS for both physiological case and pathological case are simulated in a 10-min time sequence.
Full-text available
This work presents a 2D quantification of strain field caused due to the motion artifact in an Electrocardiogram (ECG) measurement. The objective of this work is to estimate the skin stretch induced motion artifact in an ECG signal. An ECG measurement was obtained from a subject for 10 seconds using standard Ag/AgCl electrodes by continuously moving the arm back and forth during the measurement. A Poly dimethyl siloxane (PDMS) patch of dimensions 40 mm × 45 mm × 0.254 mm was adhered to the arm during motion. The movement of the PDMS patch during the ECG measurement was recorded in a video and motion artifact was quantified in terms of normal and shear strain components εx, εy and εxy. These values were derived using feature detection and Euclidean distance feature mapping. The motion artifact was eliminated from the ECG signal using Extended Kalman Filtering (EKF).
• To determine the incremental yield of ambulatory monitoring in the evaluation of syncope, three serial 24-hour Holter recordings were obtained in a consecutive series of 95 patients with syncope, the cause of which was not explained by history, physical examination, or 12-lead electrocardiogram. The mean age of patients was 61 years and 41% were men. Major electrocardiographic abnormalities were found in 26 patients (27%), including unsustained ventricular tachycardia (19 patients), pauses of at least 2 seconds (8 patients), profound bradycardia (1 patient), and complete heart block (1 patient). The first 24-hour Holter recording had at least one major abnormality in 14 patients (15%) (95% confidence interval, 8.3% to 23.4%). Of the 81 patients without a major abnormality on the first Holter recording, the second Holter recording had major abnormalities in 9 (11%) (95% confidence interval, 5.1% to 20.0%). Of the 72 patients without a major abnormality on the first two Holter recordings, only 3 patients (4.2%) had a major abnormality on the third Holter recording (95% confidence interval, 0.8% to 11.7%). Four factors were significantly associated with an increased likelihood of a major abnormality on 72 hours of monitoring: age above 65 years (relative risk, 2.2), male gender (relative risk, 2.0), history of heart disease (relative risk, 2.2), and an initial nonsinus rhythm (relative risk, 3.5). These results suggest that 24 hours of Holter monitoring is not enough to identify all potentially important arrhythmias in patients with syncope. Monitoring may need to be extended to 48 hours if the first 24-hour Holter recording is normal. (Arch Intern Med. 1990;150:1073-1078)