COMPARING DRY ELECTRODE MATERIALS FOR LONG-TERM

ECG MONITORING

Evaluation of a New Method

Silvester Fuhrhop, Stefan Lamparth, Tobias Seemann, Wilhelm Stork

KIT - Institute for Information Processing Technology, Karlsruhe, Germany

Stephan Heuer, Malte Kirst

FZI Research Center for Information Technology, Haid-und-Neu-Str. 10-14, 76131 Karlsruhe, Germany

Keywords:

Dry electrode materials, Signal quality, ECG, Comparison, Long-Term-ECG

Abstract:

Studies prove, that ECG monitoring over longer periods of time can signiﬁcantly improve the diagnosis of

heart failure. Dry electrodes are a valuable alternative to conventional wet electrodes, if the monitoring pe-

riod is longer than one week, because they are more stable and cause less skin irritations. One drawback

however is the minor signal quality compared to wet electrodes. As there are different electrode materials for

dry electrodes available, a qualitative comparison concerning signal quality is carried out in this work. The

signal quality of well known dry-electrode-materials used in ﬁtness-heart belts are compared with metal dry

electrodes. All electrodes are assembled in an identical mechanical electrode-setup. A parcours-based ﬁeld

trial with 13 subjects is carried out in two stages. In order to estimate the reproducibility of the parcours, the

ﬁrst stage is used to benchmark the differences in signal quality of consecutive measurements not originating

from the electrode material. In the second stage however, the actual measurement takes place. The accuracy

of the method is presented and the comparison results are discussed.

1 INTRODUCTION

Longterm ECG monitoring provides a promising ap-

proach to face the rising numbers of heart diseases.

Established Holter recorders equipped with adhesive

wet electrodes feature good signal quality, but show

bad wearing comfort and bad usability when looked

at from the daily regular use perspective. Recording

approaches with textile integrated dry electrodes in-

deed prevail over these drawbacks (Fuhrhop et al.,

2009) (Yoo et al., 2009) (M. Di Rienzo, 2005).

Although Gruetzmann et al. state that dry elec-

trodes are more resistant to movement artifacts than

adhesive electrodes compared by short-term analy-

sis (A. Gr

utzmann and M

uller, 2007) it is shown

in (Fuhrhop et al., 2009), that there is a big dif-

ference between laboratory measurements and real

world measurements where Ag/AgCl electrodes still

outperform the dry electrodes in a real 24/7 life sce-

nario. It is for this reason that dry electrode ECG sys-

tem have failed to establish themselves as alternative

for Holter systems with conventional Ag/AgCl elec-

trodes in the medical environment. Multiple sources

e. g. (Connolly and Buckley, 2004) (David Ruhle-

mann and Frosch, 2010) document that conventional

Ag/AgCl electrode gels also cause contact dermati-

tis in multiple cases. As to the knowledge of the au-

thors there are no systems with capacitively coupled

electrodes available that can cope with dry electrode

systems concerning robustness against movement ar-

tifacts. In terms of artifact robustness and long-term

usability dry electrodes are the most promising acqui-

sition approach for good signal coverage in long-term

ECG monitoring to date.

2 RELATED WORK

The investigations on dry electrodes for ECG mon-

itoring are numerous. (PJ. Xu, 2008) gives a solid

overview over this ﬁeld. Most groups that use dry

electrodes have benchmarked their system against a

Holter recorder (M. Di Rienzo, 2005) (G. Gargiulo,

2010). This approach is feasible, to compare com-

plete dry electrode systems with each other. However,

485

Fuhrhop S., Lamparth S., Seemann T., Stork W., Heuer S. and Kirst M..

COMPARING DRY ELECTRODE MATERIALS FOR LONG-TERM ECG MONITORING - Evaluation of a New Method.

DOI: 10.5220/0003165104850488

In Proceedings of the International Conference on Bio-inspired Systems and Signal Processing (BIOSIGNALS-2011), pages 485-488

ISBN: 978-989-8425-35-5

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

a direct conclusion about the inﬂuence of the used

dry-electrode material on the signal quality cannot

be drawn, because the mechanical setup of the elec-

trode mounting differs. Lacking the adhesive ﬁxation

to the body, the ﬁxture of dry electrodes has great

inﬂuence on the signal quality. There are multiple

methods to compare dry electroconductive electrodes

for ECG measurements: Subjective comparison of

different electrode types with a reference electrode

(Ag/AgCl), the skin-electrode impedance (Hoffmann

and Ruff, 2007) (Baek JY, 2008), detectability of

ECG waves (Pola and Vanhala, 2007) (Fuhrhop et al.,

2009), power spectral density (Scilingo et al., 2005)

(Puurtinen et al., 2006). As dry electrode systems

have their advantages in long-term applicability, it is

very likely that the signal analysis of the recorded data

is performed automatically by ECG-waves detection

algorithms. That is why the comparison of differ-

ent electrode materials concerning their signal quality

will be accomplished by automated ECG waves de-

tection in this work.

The quality of an ECG signal decreases with ris-

ing activity of the subject, because the relative move-

ment between body and electrode induce potentials at

the electrode-skin interface that add to the body sur-

face potentials of the heart and reside in the same fre-

quency band. Hence electrodes for long-term applica-

bility need to be robust against potentials at the skin-

electrode interface that are induced by movement. For

the in vivo comparison of ECG-electrodes it has to

be taken into account, that ECG morphology is de-

pendent on the electrode position on the body. Po-

sition changes of the electrode can heavily inﬂuence

the signal morphology. By performing a simultaneous

measurement with two electrode-types attached to the

subject, the evaluation measurement is not compara-

ble due to the changes in morphology caused by slight

differences in electrode position. Therefore the mea-

surements have to be performed consecutively and

special care has to be taken, that the measurement is

conducted in an identical way.

3 MATERIALS AND METHOD

3.1 Tested Electrode Materials

The ﬁve materials compared in this test are presented

in the list below. Two dry electrode materials, that

are widely used in ﬁtness chest belts and three metal

based materials are evaluated:

• Conductive Rubber (www.wacker.com)

• Silver coated Jersey-Textile

• Stainless-steel (14301 alloy)

• Silver (925 sterling silver)

• Stainless-Steel mesh (www.beissermetall.de)

3.2 Measurement Setup

The electrode material is integrated into a rigid carrier

that is ﬁxed to the body by a belt. The deployed elec-

trode design ensures that the ﬁxture of each electrode-

material-setup is identical hence a relative compari-

son between the selected electrode materials is possi-

ble.

Electrode carrier

Material

under Test

Body

22mm

50mm

Figure 1: Electrode design.

Two electrodes like the one depicted in ﬁgure 1 are

mounted in a stretchable belt. The length of the belt is

adjusted for each subject individually. A differential

measurement setup similar to Einthoven lead I is ac-

quired. Figure 2 shows the conﬁguration of the leads

on the left and a subject of the trial with the attached

unit (right).

ectro

Recorded

Body

+

‐

Recorded

Signal

Reference

electrode

Figure 2: Electrode conﬁguration.

3.3 Field Trial Setup

In order to acquire comparable subsequent measure-

ments a parcours-based ﬁeld trial with well deﬁned

stages was designed. To identify the robustness

against movement artifacts, several activities where

included in the trial stages. The selected actions oc-

cur in the daily routine and induce heavy movements

of the thorax and the upper extremities. The list below

shows the duration of each activity:

• Sitting in a chair (120s)

• Stirring rice in a jar (60s, 20s break)

• Taking a jar out of a shelf, put it on the ﬂoor and

back up on a shelf (60s, 20s break)

• Cleaning teeth (60s, 20s break)

BIOSIGNALS 2011 - International Conference on Bio-inspired Systems and Signal Processing

486

• Walking stairs (60 s, 20s break)

• Using a wheelchair (60s, 20s break)

• Walking with cruchtes (60 s, 40s break)

• Fast walking at 6 km/h in a treadmill (180s)

The overall duration of the parcours is 13:20 min-

utes. No detergent or soap was used for skin prepa-

ration but before each run the chest area of each sub-

ject was cleaned with tap water and dried with a fresh

towel before applying the chest belt. The belt was at-

tached with a tension of 3.5N. Before each measure-

ment, the subject rested with the belt attached for 15

minutes in order to let the dry electrodes reach the

equilibrium. The duration of each station in the par-

coursis controlled by a timer, so that every activity is

done with similar duration for every material. Start

and end of each activity is signaled acoustically and

by a vibration-motor in the belt. The ﬁeld trial was

divided in two stages. Stage 1: The parcours was con-

ducted four times in a row by each subject with stain-

less steel electrodes. Between every run, the electrode

belt was removed and reapplied in the way described

above. Stage 2: In the second stage every subject con-

ducted the parcours once for each electrode material.

The trial took place in an air conditioned room, the

temperature was between 24 and 25

◦

C, humidity was

50 to 54% rel.

3.4 Subject Population

The population of this study is composed of 13 adults,

both males and females, with a range of ages from 21

to 26. All of the subjects were healthy and none re-

vealed a past history of any cardiac problems. All

subjects were surveyed prior to study inclusion to en-

sure each met study qualiﬁcations.

3.5 Estimation of the Signal Quality by

ECG Waves Detection

The applied method, described in (Fuhrhop et al.,

2009), uses the detectability of QRS waves. A ref-

erence annotation of the ECG signal is created by an

expert and a trigger list is computed by a QRS detect-

ing algorithm. This trigger list is compared beat by

beat with the reference annotation hence sensitivity

(Se) and positive predictive value (pP) of each mea-

surement can be calculated. Those parameters repre-

sent the signal quality of the recorded ECG from the

QRS detecting algorithms point of view.

3.6 Results

The signal quality of each run was estimated by the

beat-by-beat method described above. At ﬁrst the re-

producibility of well deﬁned activities was evaluated:

The comparison of successively recorded parcours

runs implies that the difference in signal quality be-

tween subsequent recordings only originates from the

used electrode material. To evaluate if other factors

inﬂuence the comparability each subject conducted

stage 1 (described above). Table 1 shows mean value

and standard deviation for Se and pP of 13 subjects.

It is clear, that despite of a parcours based trial and

a reproducible application of the electrodes, the sig-

nal quality of each subject differs between each run

signiﬁcantly. The mean value of the standard devi-

ation for Se is 0.4% and for pP is 1.7%. Hence it

is only possible to reliably compare subsequent mea-

surements, if the mean values of the quality indica-

tors for different electrode materials exhibit signiﬁ-

cant higher differences than the above listed standard

deviations.

Table 1: Se and pP of the four subsequent parcours-runs

with stainless steel material (stage 1).

Subject Se [%] pP [%]

mean std mean std

1 98.7 0.4 95.6 0.6

2 98.1 1.6 97.2 1.3

3 99.5 0.5 99.3 0.4

4 99.3 0.3 99.0 0.8

5 99.2 0.2 97.4 1.5

6 99.3 0.5 97.5 0.2

7 99.6 0.1 97.1 2.6

8 99.7 0.1 99.1 0.9

9 99.8 0.1 99.4 0.2

10 99.5 0.5 97.1 3.6

11 99.6 0.5 99.0 1.4

12 98.6 0.5 93.3 2.4

13 99.3 0.7 96.7 6.2

mean 99.3 0.4 97.5 1.7

3.7 Estimation of the ECG Signal

Quality for the Electrode Materials

In the succeeding trial stage each subject performed

the parcours for each electrode material once. There-

fore 13 measurements for each material are available

for comparison. Table 2 shows mean values and stan-

dard deviation for both pP and Se for all tested mate-

rials. Se ranges between 98,91 and 99.29%, pP ranges

between 96.94 and 98.25% respectively.

COMPARING DRY ELECTRODE MATERIALS FOR LONG-TERM ECG MONITORING - Evaluation of a New

Method

487

Table 2: Sp an pP for different electrode materials (stage 2).

Material Se [%] pP [%]

mean std mean std

Silver 99.29 1.2 97.95 1.8

Mesh 98.91 1.4 98.25 1.0

Jersey 99.23 0.6 97.22 2.5

Conductive Rubber 99.06 1.2 96.94 2.6

Stainless Steel 99.27 0.7 97.55 2.8

mean 99.15 1.0 97.57 2.1

4 DISCUSSION

A larger measurement database (both subjects and

recording time) could have averaged out outliers, but

for the available number of measurements the accu-

racy of the adopted method ranges at 0.4% for Se and

1.7% for pP. The calculated performance difference of

the dry electrode materials of Se and pP reside in the

same scale as the accuracy of the method. Due to this

fact it is evident that on the given database the applied

comparison method is not accurate enough to allow a

conclusion, what material is more robust against arti-

facts.

5 CONCLUSIONS

AND OUTLOOK

A comparison between dry electrode-materials used

in ﬁtness belts that feature a low prize and easy tex-

tile integrability and metal dry electrode-materials has

been carried out. The accuracy of the method used

for comparison was evaluated. It could be proven

that even during a well deﬁned reproducible parcours

based study the signal quality of each run signiﬁcantly

differs due to other factors than the electrode mate-

rial. The fact that the number of subjects was lim-

ited and the recording duration was short reduced the

accuracy of the applied method. The comparison be-

tween the selected materials revealed, that the iden-

tiﬁed performance differences range in the measur-

ing inaccuracy, hence no conclusion about the inﬂu-

ence of the material can be drawn. The method must

be adapted in such a way, that either the amount of

recorded data is enlarged or the measurements of the

different materials are taken simultaneously. In this

case the effect of different electrode positions must

be taken into account. Because all electrode materi-

als are attached mechanical identically in this evalu-

ation, it is also possible that the electrode mounting

of dry electrodes is more relevant for movement arti-

facts than the electrode material itself. In the ongoing

work the presented method will be adapted to allow

a more accurate conclusion about the inﬂuence of dry

electrode material on signal quality and the effects of

the mechanical ﬁxture on the signal quality will be

analyzed.

REFERENCES

A. Gr

utzmann, S. H. and M

uller, J. (2007). Novel dry elec-

trodes for ecg monitoring. Physiological Measure-

ment, 28.

Baek JY, An JH, C. J. e. a. (2008). Flexible polymeric dry

electrodes for the long-term monitoring of ecg. In Sen-

sors and Actuators A: Physical. Elsevier.

Connolly, M. and Buckley, D. A. (2004). Contact dermati-

tis from propylene glycol in ECG electrodes, compli-

cated by medicament allergy, volume 50 of Contact

Dermatitis. Wiley-Blackwell.

David Ruhlemann, Katrin Kuegler, B. M. and Frosch, P. J.

(2010). Contact dermatitis to self-adhesive ECG elec-

trodes, volume 62 of Contact Dermatitis. Wiley-

Blackwell.

Fuhrhop, S., Lamparth, S., and et al, K. (2009). Ambu-

lant ecg recording with wet and dry electrodes. In

World Congress on Medical Physics and Biomedi-

cal Engineering, volume 25/5 of IFMBE Proceedings.

Springer.

G. Gargiulo, P. Bifulco, M. C. e. a. (2010). An ultra-high

input impedance ecg ampliﬁer for long-term monitor-

ing of athletes. In Medical Devices: Evidence and

Research. Dove Press.

Hoffmann, K.-P. and Ruff, R. (2007). Flexible dry surface-

electrodes for ecg long-term monitoring. In Engi-

neering in Medicine and Biology Society, 2007. EMBS

2007.

M. Di Rienzo, F. Rizzo, G. P. e. a. (2005). Magic system: a

new textile-based wearable device for biological sig-

nal monitoring. In Engineering in Medicine and Biol-

ogy Society, 2005. IEEE-EMBS 2005.

PJ. Xu, H. Zhang, X. T. (2008). Textile-structured elec-

trodes for electrocardiogram. Textile Progress.

Pola, T. and Vanhala, J. (2007). Textile electrodes in ecg

measurement.

Puurtinen, M., Komulainen, S., Kauppinen, P., Malmivuo,

J., and Hyttinen, J. (2006). Measurement of noise

and impedance of dry and wet textile electrodes, and

textile electrodes with hydrogel. In Engineering in

Medicine and Biology Society, 2006. EMBS ’06.

Scilingo, E. P., Gemignani, A., Paradiso, R., Taccini, N.,

Ghelarducci, B., and De Rossi, D. (2005). Perfor-

mance evaluation of sensing fabrics for monitoring

physiological and biomechanical variables. Informa-

tion Technology in Biomedicine, IEEE Transactions

on, 9(3).

Yoo, J., Yan, L., Lee, S., Kim, H., and Yoo, H.-J. (2009). A

wearable ecg acquisition system with compact planar-

fashionable circuit board-based shirt. Information

Technology in Biomedicine, IEEE Transactions on,

13(6).

BIOSIGNALS 2011 - International Conference on Bio-inspired Systems and Signal Processing

488