Off-the-Person Electrocardiography

Hugo Pl

acido da Silva

, Carlos Carreiras

, Andr

e Lourenc¸o

1,2

and Ana Fred

IT - Instituto de Telecomunicac¸

oes, Av. Rovisco Pais, 1, 1049-001 Lisboa, Portugal

ISEL - Instituto Superior de Engenharia de Lisboa, R. Cons. Em

ıdio Navarro, 1, 1959-007 Lisboa, Portugal

Keywords:

Electrocardiography, Off-the-Person, Taxonomy, Instrumentation, Experimental Evaluation.

Abstract:

Electrocardiography (ECG) methods are still mostly bound to hospital and short-time data acquisition settings.

Still, a paradigm shift is emerging, in which everyday technology is increasingly capable of measuring ECG

signals in a more pervasive manner. This is paving the way for systems that can better analyze and adapt to

perceived changes in the health status or behavior of the user. In this paper we present a taxonomy for the

intrusiveness of ECG data acquisition systems, describe a sensor design for what we call the ”off-the-person”

approach, and provide a discussion of the main challenges posed by these new methodologies. Our work is

targeted at pervasive electrocardiography through signal acquisition at the hand palms or ﬁngers, by providing

a simpliﬁed sensor setup that can be integrated into virtually any object with which the person interacts with.

Experimental results show that data acquired using our proposed approach is highly correlated with data

obtained through conventional methods.

1 INTRODUCTION

The ﬁrst practical implementation of what we known

today as Electrocardiography (ECG) appeared around

1887 and is credited to Augustus Waller, a British

physiologist that was able to record the electrical ac-

tivity of the human heart using non-invasive meth-

ods, and provided the ﬁrst known systematic ap-

proach to the study of the electrical properties of

the heart (Besterman and Creese, 1979). Questions

surrounding the clinical applicability of the signals

and limitations of the measurement instrumentation

used at the time made the ECG have a slow start.

It was not until 1906 that a more widespread accep-

tance and use occurred. The groundbreaking work by

Willem Einthoven provided signiﬁcant advances both

in terms of the measurement methods and signal char-

acterization, enabling the ECG to be more clearly un-

derstood. His work was recognized in 1924 through

the Nobel Prize in Medicine, several years after the

initial experimentations in the ﬁeld (Barold, 2003).

Nowadays, the ECG is a perfectly established and

mainstream technique, and it provides vital informa-

tion for the diagnosis and observation of a wide array

of complex cardiovascular problems that include ar-

rhythmias, myocardial ischemia, prolonged QT inter-

val, among many others (Drew et al., 2004)(Chung,

2000). Most importantly, early detection of changes

in the cardiac patterns is crucial to anticipate severe

and long lasting problems, and to develop preven-

tive clinical interventions. Standard clinical prac-

tices are still based on short-term ECG data; the most

widespread approach is the 12-lead ECG for momen-

tary assessment in a clinical setting (∼ 1 minute of

data), and in selected cases, Holter monitors are used

for ∼ 24 hour assessment in an ambulatory setting.

In this paper, we present a taxonomy for the intru-

siveness of ECG data acquisition methods, together

with an off-the-person sensor design targeted at data

acquisition in a pervasive framework. Our work was

not devised with the purpose of replacing existing

data acquisition procedures. Instead, our goal is to

complement current practices with a simpliﬁed sen-

sor setup that can be introduced in multiple aspects of

the everyday life of patients or even of healthy sub-

jects, as a way of enabling a more comprehensive as-

sessment of cardiovascular parameters, and potenti-

ating preventive interventions. The rest of the paper

is organized as follows: Section 2 describes a taxon-

omy for the intrusiveness of ECG data acquisition ap-

proaches; Section 3 describes our off-the-person sen-

sor approach, highlighting the main technical options;

Section 4 provides an experimental comparison be-

tween the off-the-person approach and medical-grade

equipment; and ﬁnally, Section 5 outlines the main

conclusions.

Plácido da Silva H., Carreiras C., Lourenço A. and Fred A..

Off-the-Person Electrocardiography.

DOI: 10.5220/0004647700990106

In Proceedings of the International Congress on Cardiovascular Technologies (IWoPE-2013), pages 99-106

ISBN: 978-989-8565-78-5

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

2 A TAXONOMY FOR ECG DATA

ACQ. INTRUSIVENESS

Given the bioelectrical nature of the cardiac activ-

ity, the voltage potential differential is the most com-

monly used ECG measurement principle. Still, other

approaches can be used to sense the cardiac activity,

such as capacitive and mechanical methods. Despite

the underlying operating principle, several devices

have emerged over the years, that enable ECG signal

acquisition, and which we have classiﬁed according

to the intrusiveness level of the hardware setup with

respect to its placement on the body of the user. Fig-

ure 1 shows an overview of our taxonomy.

2.1 In-the-Person

Devices is this category are placed inside the body

of the person, and are generally used only in extreme

clinical scenarios to monitor or address medical con-

ditions. The devices are surgically placed inside the

body with measurement leads attached directly to the

heart, enabling the continuous monitoring of its be-

havior and the delivery of electrical impulses when-

ever a deviation from a normal cardiac pattern is de-

tected. Nowadays, most devices enable the exter-

nal access and remote monitoring of basic parameters

both by clinicians and patients.

One class of devices are the Implantable systems,

of which artiﬁcial cardiac pacemakers are the most

widely known example; these are used to compensate

for shortcomings of the electrical conduction system

of the heart, due to degenerative or pathological con-

ditions (Timperley et al., 2008). Modern implantable

devices can have extremely compact and lightweight

form factors, an example of which is the HD-X11 sys-

tem from Data Sciences International

, with 2.2 g and

a volume of just 1.4 cm

Another class of devices can be deﬁned as Min-

imally Invasive, which includes implantable loop

recorders (ILR); albeit being placed inside the body

of the person, these are only applied subcutaneously,

through a simple medical procedure that typically

only requires local anesthesia, and enable the contin-

uous recording of the cardiac activity.

2.2 On-the-Person

The most common approaches to ECG measurement

used nowadays work by attaching a device, or some

of its components, externally to the body surface.

http://www.datasci.com/products/implantable-telemetry

Currently, devices designed to be used in an on-

the-subject approach are perfectly commoditized, and

range from medical-grade equipment to personal use

and self-monitoring devices for heart rate assessment

in sports and wellbeing activities (e.g. Polar Wear-

Link+

Standard ECG devices can be classiﬁed as Sta-

tionary systems. Devices in this category are typi-

cally characterized by workbench and bedside mon-

itors for medical use, and require the placement of

12 or more leads mounted on the chest and limbs, al-

though conﬁgurations with a lower number of leads

can also be found. Other properties of these devices

include the fact that they generally need conductive

paste or gel to lower the skin impedance, and that the

patient is bound to a limited physical space.

New developments in signal acquisition technolo-

gies greatly improved the usability, and enable more

practical approaches that fall into the class of Am-

bulatory systems. In the clinical domain, Holter

monitors are used for ambulatory cardiac assessment.

These are devices in which less measurement leads

are used, and a partial recording of the activity of

the heart is made in an internal memory, typically

over the course of 24 hours. Recently, there has also

been an increasing interest in integrating ambulatory

ECG sensing into portable devices, as shown by the

EPI Life ”Doctor in Your Pocket”

, claimed to be

the ﬁrst ECG-enabled mobile phone, and also by the

AliveCor

monitor, proposed as a clinical diagnosis

tool, and which enables real-time ECG measurement

on the iPhone/iPad when the subject places the acces-

sory on the chest.

Within ambulatory systems, a vast amount of

work has been done around t-shirt and other wearable

form factors, such as the VitalJacket from BioDevices

. Another smart t-shirt is evaluated in (de Isla et al.,

2011), in which the measurement leads are embedded

into the fabric as a way of achieving a more practical

acquisition setup. In (Chi et al., 2010), the authors

present a comprehensive review of capacitive sensor

technologies that can be applied to a chest strap or

t-shirt to monitor the cardiac activity. A wearable

device designed as a necklace is described in (Silva

et al., 2011b), which uses dry Ag/AgCl electrodes and

enables heart rate measurement on the user’s neck.

http://www.polar.com/en

http://epimhealth.com.sg

http://www.alivecor.com

http://www.biodevices.pt

CARDIOTECHNIX2013-InternationalCongressonCardiovascularTechnologies

100

Figure 1: Overview of the intrusiveness and potential outreach of ECG data acquisition systems.

2.3 Off-the-Person

One of the challenges that has recently started to be

more prominently addressed in the state-of-the-art, is

the improvement of the sensor acceptability into what

can be deﬁned as an off-the-person approach, given

that the sensors are integrated in objects with which

the subject regularly interacts with. The rationale be-

hind this approach is that, unlike the on-the-person

methods, in which the user needs to wear the sensor

or perform a voluntary action to have the sensor in

contact with his body, in this case the sensor is inte-

grated in a pervasive manner, so that the user does not

need to change his/her normal interaction patterns.

Within this trend, one of the categories comprises

Contact Based systems. An example can be found

in previous work by our team (Silva et al., 2011a),

where a bipolar sensor with virtual ground and dry

electrodes was proposed for ECG data acquisition at

the hand palms or ﬁngers. The main advantage is that

these methods can be easily integrated into everyday

items without impacting on the user’s routines. Ap-

plications for such devices include ECG monitoring

while the person is working at a computer keyboard,

holding a game station controller, the steering wheel

of a car, and many others items, enabling its use in

a pervasive electrocardiography framework, and pro-

moting novel long-term monitoring paradigms.

The off-the-person approaches also comprise

Contactless systems; this class includes capacitive

and mechanical methods, which albeit not being ca-

pable of measuring the tradicional ECG signal, mea-

sure an ECG-like activity. Capacitive sensors mea-

sure the small time-varying electric ﬁelds associated

with the bioelectric activity of the heart (Chi et al.,

2010)(Martins et al., 2011). These sensors do not re-

quire direct contact with the body of the user, and can

be designed to measure the ECG at distances of ∼ 1

cm or more, even with clothing in-between the body

and the sensor, enabling its integration in the back

of a chair, in a car seat, or other analogous items.

Mechanical sensors measure heart-related events by

sensing the small vibrations propagated to the body

surface after the contraction of the cardiac muscula-

ture. This technique is known as ballistocardiogra-

phy or seismocardiography, and current approaches

are based either on accelerometers or electromechan-

ical ﬁlms (Postolache et al., 2010).

These novel approaches are aligned with the lat-

est trends in medical applications of technology. As

Eric Topol states in his book (Topol, 2012), the fu-

ture of healthcare will inevitably lie on the inﬁltra-

tion of medical devices into our daily lives, collecting

more data about the human being over longer periods

of time, and reasoning about these large volumes of

data. This is the very essence of pervasive health.

Off-the-PersonElectrocardiography

101

Figure 2: Contact based off-the-person sensor integrated in

a standard computer keyboard.

3 A CONTACT BASED SENSOR

FOR OFF-THE-PERSON ECG

In our work, we have been focusing on minimizing

the number of electrical contact points with the sub-

ject’s body, eliminating the need of any gel or con-

ductive paste in the interface with the skin, and devis-

ing a non-intrusive sensor system. Figure 2 shows an

example of our sensor integrated in a computer key-

board. Our tests have shown that even when com-

pared with conventional approaches in which pre-

gelled electrodes are used, this conﬁguration provides

an output signal with adequate quality both with dry

Ag/AgCl or conductive textile electrodes, and no skin

preparation in either case.

3.1 Sensor Design

We developed analog signal conditioning circuitry

adapted to the ECG in terms of gain and bandwidth

(Malmivuo and Plonsey, 1995). Our design is single-

ended and is based on the classical voltage poten-

tial differential principle, one singularity being the

fact that the typical ground electrode is replaced by

a reference voltage produced by the circuit (generally

referred to as the ”virtual ground”). Table 1 shows

the typical physiological speciﬁcations for ECG sig-

nals (Webster, 2009); to measure the low potential

differences associated with these signals (in the mV

range), our sensor design includes a precision instru-

mentation ampliﬁer (In-Amp) offering high common-

mode rejection (110 dB at gains greater or equal than

10). Furthermore, we use low-noise high speed op-

erational ampliﬁers (Op-Amp) to perform bandpass

ﬁltering and ampliﬁcation.

Table 1: Speciﬁcations of the ECG.

Range Frequency

0.5 − 4 mV 0.01 − 250 Hz

Figure 3: Block diagram of the ECG sensor design.

Figure 4: Frequency response of the ECG sensor block.

A block diagram of the circuit is shown in Fig-

ure 3. The use of dry electrodes introduces a higher

impedance and is more prone to high frequency noise;

as such, our design uses a gain of 2000 and a band-

width between 0.5-40 Hz. We have a ﬁrst ampliﬁca-

tion stage that uses an instrumentation ampliﬁer with

gain 20, to measure the weak voltage potential differ-

ences produced by the cardiac activity and increase

their amplitude. Afterwards, an active 1st order high-

pass ﬁlter (Gain = 1; Cutoff = 0.5 Hz) is applied in or-

der to minimize the impact of low frequencies; in par-

ticular to remove modulation introduced by the res-

piratory activity. Finally, we use a 4th order Butter-

worth low-pass ﬁlter (Gain = 100; Cutoff = 40 Hz), to

limit the bandwidth of the signal to a range that dis-

cards the powerline noise, and to further increase the

amplitude of the signal in order to obtain higher def-

inition in the digital domain. The frequency response

of the sensor is presented in Figure 4.

Although our sensor was speciﬁcally designed

for 1-lead differential measurement at the ﬁngers or

hands (left / right) with virtual ground, it can also be

used in standard chest or limb locations, with the op-

tion to use a ground lead as well. Equation 1 shows

the transfer function for this sensor (V

denotes the

reference voltage).

CARDIOTECHNIX2013-InternationalCongressonCardiovascularTechnologies

102

Figure 5: Time and frequency response of the ECG sensor

to the chirp signal.

Out

= (V

In+

−V

In−

) × 2000 +V

(1)

3.2 Experimental Characterization

Tests were performed to characterize the quality of

the analog front-end in terms of Signal-to-Noise

Ratio (SNR), Signal-to-Noise Ratio plus Distortion

(SINAD), and Total Harmonic Distortion (THD). In

all experimental tests, the signals were generated us-

ing an Agilent 33220A function generator, and ac-

quired using an analog-to-digital converter (ADC)

with 10-bit resolution, a sampling rate of 1000 Hz,

and a 3.3 V peak-to-peak dynamic range (V

) (Guer-

reiro et al., 2013). The function generator used in our

experiments is not able to accurately generate wave-

forms in the near-millivolt range, and as such, to char-

acterize the real response of the analog circuit, we re-

duced its gain to 100 (Gain = 1 at the In-Amp), ensur-

ing an output signal between 0 −3.3 V , that is, within

the dynamic range of the ADC.

Table 2: Dynamic speciﬁcations.

SNR [dB] SINAD [dBc] THD [dBc]

44.54 42.49 -46.74

For this test we injected a synthesized chirp wave

spanning the 0 −100 Hz frequency range, with 1 sec-

ond duration, 28 mV peak-to-peak, and offset of V

being the supply voltage). Figure 5 shows the fre-

quency response of our ECG circuit; as shown in the

top ﬁgure, the output signal of the circuit is a chirp

wave with ≈ 2.8 V peak-to-peak (V

) and with at-

tenuation in the low and high frequencies, a natural

response of the bandpass ﬁlter that we have used. To

characterize the dynamic speciﬁcations of the circuits,

we used a synthesized sine wave with a frequency of

24 Hz, 28 mV peak-to-peak (V

), and offset of V

/2;

in Table 2 we summarize the results.

4 MEDICAL-GRADE VS.

OFF-THE-PERSON LEADS

One of the open questions in our off-the-person ap-

proach, with acquisition at the hands, is the relation

between the signals obtained with this type of sensor,

and those obtained with medical-grade stationary on-

the-person equipments, which are the gold standard

for ECG measurement. In this section we provide a

comparison of both approaches.

4.1 Methodology

We conducted experimental tests involving 8 vol-

unteers (4 males and 4 females), in which simul-

taneous recordings were performed using a Philips

PageWriter Trim III ECG device, and our off-the-

person sensor design with data acquisition performed

using the Biosignal Igniter Toolkit (BIT). The Philips

equipment was used in the standard 12-lead setting

(I-III, V1-V6, aVF, aVL, aVR), with conductive paste

applied to each of the electrodes. Our sensor was used

in the virtual ground setting with dry Ag/AgCl elec-

trodes, and the subject was asked to hold one of the

electrode leads in the right hand and the other on the

left hand.

The raw data from each device was bandpass ﬁl-

tered using the same procedure, and all the individual

heartbeat waveforms were segmented. For a detailed

description of the adopted pre-processing methodol-

ogy, we refer the reader to (Canento et al., 2013) and

references therein. We focused on determining the

morphological similarity between individual heart-

beat waveforms collected using each of the sensor de-

vices. Given that dry electrodes are used in our sensor,

and also due to the fact that the on-the-person equip-

ment uses leads scattered through different anatomi-

cal locations, the amplitude of the signals collected by

each device is affected by a variable scale factor. To

account for such differences, in this study we adopted

the cosine distance as similarity metric (Equation 2).

cos

) = 1 −

∑

k=1

[k]x

[k]

∑

k=1

[k]

∑

k=1

[k]

. (2)

Off-the-PersonElectrocardiography

103

Figure 6: Boxplot of the distance between the off-the-person and each of the medical-grade leads.

Figure 7: Example of the segmented heartbeat waveforms obtained with the off-the-person and each of the medical-grade

leads for one of the tested subjects.

Figure 8: Histograms of the waveform distance between the off-the-person and each of the medical-grade leads.

CARDIOTECHNIX2013-InternationalCongressonCardiovascularTechnologies

104

4.2 Results

As shown in Figure 6, the signal obtained with our

off-the-person approach is clearly a lead I derivation.

The average cosine distance of nearly zero between

the off-the-person waveform and the lead I waveform

of the medical-grade device, together with the low

standard deviation across the overall set of subjects,

allow us to conclude that both signals are fully corre-

lated. This is further reinforced by the visual observa-

tion of the data; in Figure 7 we depict an overlay with

the segmented individual heartbeat waveforms for one

of the tested subjects. The off-the-person data is rep-

resented in blue, while the on-the-person data for each

of the leads is represented in green; in this case, the

off-the-person lead and the lead I data present an al-

most exact match.

Analyzing the case-by-case statistics, we are able

to observe that for some of the subjects, leads II, aVF,

aVR, and V3-V6 also exhibit a low average distance

to the waveform obtained using the off-the-person ap-

proach, and hence a high morphological similarity.

For example in Figure 7, V3 and V4 are quite simi-

lar to the off-the-person lead. The distributions of the

individual distances between our off-the-person lead

and each of the medical-grade leads for the overall

population can be found in Figure 8.

5 DISCUSSION AND FUTURE

WORK

Electrocardiography (ECG) has progressed a long

way since it was ﬁrst introduced in the clinical prac-

tice. In the recent years, an increasingly growing

community has focused on improving the usability of

ECG equipment, and while most of the work has been

targeting wearable form factors (e.g. t-shirts), our

work has been pivoting towards what can be classiﬁed

as an off-the-person approach. In this paper we have

proposed a taxonomy of ECG data acquisition meth-

ods with respect to their intrusiveness level, described

an off-the-person sensor designed for ECG data ac-

quisition at the hands and ﬁngers using dry electrodes,

and provided experimental results regarding the com-

parison between the off-the-person approach and con-

ventional medical-grade equipment.

Comparative tests have shown that the signals

obtained through our off-the-person approach are

matched to the conventional lead I derivation, and

that even without skin preparation or the use of con-

ductive paste to lower the impedance with the skin,

the morphology of the heartbeat waveform can be re-

trieved. Our work is targeted at ECG data acquisition

in a pervasive framework, by providing a simpliﬁed

sensor setup that can be used for everyday monitor-

ing. The applicability of our pervasive ECG approach

is not bound to the healthcare and clinical domains,

given that the ECG and derived measurements are

also appealing in a wide range of emerging applica-

tions, which include self-management, affective com-

puting (Medina, 2009) or even security (Lourenc¸o

et al., 2011)(Silva et al., 2013).

Future work will focus on further validating our

approach by increasing the number of tested sub-

jects, and also in the evaluation of contactless off-the-

person approaches targeting the evaluation of the rela-

tion between the signals obtained using such methods,

and the signals obtained using conventional methods.

ACKNOWLEDGEMENTS

This work was partially funded by Fundac¸

para a Ci

encia e Tecnologia (FCT) under grants

PTDC/EEI-SII/2312/2012, SFRH/BD/65248/2009

and SFRH/PROTEC/49512/ 2009, whose support the

authors gratefully acknowledge. We would also like

to thank to Hospital de Santa Marta, Dr. Rui Cruz

Ferreira, Rui C

esar das Neves, and Jos

e Guerreiro for

their technical support during the development of this

work.

REFERENCES

Barold, S. (2003). Willem Einthoven and the birth of clin-

ical electrocardiography a hundred years ago. Card.

Electrophysiol. Rev., 7:99–104.

Besterman, E. and Creese, R. (1979). Waller - pioneer of

electrocardiography. British Heart Journal, 42:61–64.

Canento, F., Lourenc¸o, A., Silva, H., Fred, A., and Raposo,

N. (2013). On real time ECG algorithms for biometric

applications. In Proc. of the 6th BIOSIGNALS Conf.

Chi, Y., Jung, T.-P., and Cauwenberghs, G. (2010).

Dry-contact and noncontact biopotential electrodes:

Methodological review. IEEE Rev. Biomed. Eng.,

3:106 –119.

Chung, E. K. (2000). Pocketguide to ECG Diagnosis.

Blackwell Publishing.

de Isla, L. P., Lennie, V., Quezada, M., Guinea, J., Arce,

C., Abad, P., Saltijeral, A., Carolina, N., Crespo, J.,

Gonz

alvez, B., Macia, A., and Zamorano, J. (2011).

New generation dynamic, wireless and remote car-

diac monitorization platform. Int’l Journ. of Card.,

153(1):83–85.

Drew, B. J., Califf, R. M., Funk, M., Kaufman, E. S., Kru-

coff, M. W., Laks, M. M., Macfarlane, P. W., Sommar-

gren, C., Swiryn, S., and Van Hare, G. F. (2004). Prac-

Off-the-PersonElectrocardiography

105

tice standards for electrocardiographic monitoring in

hospital settings. Circ., 110(17):2721–2746.

Guerreiro, J., Martins, R., Silva, H., Lourenc¸o, A., and Fred,

A. (2013). BITalino: A multimodal platform for phys-

iological computing. In Proc. of the 10th ICINCO

Conf.

Lourenc¸o, A., Silva, H., and Fred, A. (2011). Unveiling

the biometric potential of Finger-Based ECG signals.

Comp. Intel. and Neuroscience, 2011.

Malmivuo, J. and Plonsey, R. (1995). Bioelectromagnetism:

Principles and Applications of Bioelectric and Bio-

magnetic Fields. Oxford Univ. Press.

Martins, R., Primor, D., and Paiva, T. (2011). High-

Performance groundless EEG/ECG capacitive elec-

trodes. In Proc. of the 6th IEEE MeMeA, pages 503–

506.

Medina, L. (2009). Identiﬁcation of stress states from

ECG signals using unsupervised learning methods.

Master’s thesis, Univ. T

ecnica de Lisboa, Inst. Sup.

ecnico.

Postolache, O. A., Girao, P. M. B. S., Mendes, J., Pinheiro,

E. C., and Postolache, G. (2010). Physiological pa-

rameters measurement based on wheelchair embed-

ded sensors and advanced signal processing. IEEE

Trans. Instrum. Meas., 59(10):2564–2574.

Silva, H., Lourenc¸o, A., Canento, F., Fred, A., and Raposo,

N. (2013). ECG biometrics: Principles and applica-

tions. In Proc. of the 6th BIOSIGNALS Conf.

Silva, H., Lourenc¸o, A., Lourenc¸o, R., Leite, P., Coutinho,

D., and Fred, A. (2011a). Study and evaluation of

a single differential sensor design based on electro-

textile electrodes for ECG biometrics applications. In

Proc. of the 10th IEEE Sensors Conference, pages

1764 – 1767.

Silva, H., Palma, S., and Gamboa, H. (2011b). AAL+: Con-

tinuous institutional and home care through wireless

biosignal monitoring systems. In Handbook of Digi-

tal Homecare, volume 3, pages 115–142. Springer.

Timperley, J., Leeson, P., Mitchell, A., and Betts, T.

(2008). Cardiac Pacemakers and ICDs (Oxford Spe-

cialist Handbooks in Cardiology). Oxford Univ. Press.

Topol, E. (2012). The Creative Destruction of Medicine.

Basic Books.

Webster, J. G. (2009). Medical Instrumentation Application

and Design. Wiley.

CARDIOTECHNIX2013-InternationalCongressonCardiovascularTechnologies

106