contact pressure and flexibility of multipin dry eeg ... · and also with the biomagnetic center,...

750 IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 26, NO. 4, APRIL 2018

Contact Pressure and Flexibility of MultipinDry EEG Electrodes

Patrique Fiedler , Richard Mühle, Stefan Griebel, Paulo Pedrosa, Carlos Fonseca,Filipe Vaz, Frank Zanow, and Jens Haueisen

Abstract— In state-of-the-art electroencephalogra-phy (EEG) Silver/Silver-Chloride electrodes are appliedtogether with electrolyte gels or pastes. Their applicationrequires extensive preparation, trained medical staffand limits measurement time and mobility. We recentlyproposed a novel multichannel cap system for dry EEGelectrodes for mobile and out-of-the-lab EEG acquisition.During the tests with these novel polymer-based multipindry electrodes, we observed that the quality of therecording depends on the applied normal force andresulting contact pressure. Consequently, in this paper wesystematically investigate the influence of electrode-skincontact pressure and electrode substrate flexibility oninterfacial impedance and perceived wearing comfort in astudy on 12 volunteers. The normal force applied to theelectrode was varied between the minimum required forceto achieve impedances <1.3 M� and a maximum of 4 N,using a new force measurement applicator. We found thatfor a polymer shore hardness A98, with increasing normalforce, the impedance decreases from 348 ± 236 k� and257 ± 207 k� to 29 ± 14 k� and 23 ± 11 k� at frontal

Manuscript received December 22, 2015; revised February 23,2017 and October 22, 2017; accepted January 8, 2018. Date ofpublication March 6, 2018; date of current version April 6, 2018. Thiswork was supported in part by the German Federal Ministry of Edu-cation and Research under Grant 03IPT605A, in part by the FreeState of Thuringia through funds of the European Social Fund underGrant 2015FGR0085, in part by the German Academic ExchangeService under Grant D/57212996, in part by the European Union(FP7 Marie Curie IAPP) under Project 610950. (Corresponding author:Patrique Fiedler.)

P. Fiedler and R. Mühle are with the Institute of Biomedical Engineer-ing and Informatics, Technische Universität Ilmenau, 98693 Ilmenau,Germany (e-mail: [email protected]).

S. Griebel is with the Department of Mechanism Technology,Technische Universität Ilmenau, 98693 Ilmenau, Germany (e-mail:[email protected]).

P. Pedrosa was with CEMUC–Department of Mechanical Engineer-ing, University of Coimbra, 3030-788 Coimbra, Portugal, also with theFaculdade de Engenharia, Universidade do Porto, 4200-465 Porto,Portugal, and also with the Institut FEMTO-st, 25030 Besançon Cedex,France. He is now with the Centro de Física das Universidades doMinho e Porto, 4710-057 Braga, Portugal (e-mail: [email protected]).

C. Fonseca is with CEMMPRE–Centre for Mechanical Engineering,Meterials and Processes, University of Coimbra, 3030-788 Coimbra,Portugal, and also with the Faculdade de Engenharia, Universidade doPorto, 4200-465 Porto, Portugal (e-mail: [email protected]).

F. Vaz is with the Centro de Física, Universidade do Minho, 4710-057Braga, Portugal (e-mail: [email protected]).

F. Zanow is with eemagine Medical Imaging Solutions GmbH, 10243Berlin, Germany (e-mail: [email protected]).

J. Haueisen is with the Institute of Biomedical Engineering andInformatics, Technische Universität Ilmenau, 98693 Ilmenau, Germany,and also with the Biomagnetic Center, Department of Neurology, Uni-versity Hospital Jena, 07747 Jena, Germany (e-mail: [email protected]).

Digital Object Identifier 10.1109/TNSRE.2018.2811752

hairless and temporal hairy positions, respectively. Similarresults were obtained for shore A90, A80, and A70. Thebest compromise of low and stable impedances as well asa good wearing comfort was determined for applied normalforces between 2 and 3 N using electrodes with shoreA98 or A90. Our results provide the basis for improved EEGcap designs with optimal wearing comfort and recordingquality for dry multipin electrodes, which will enable newfields of application for EEG.

Index Terms— Biomedical electrodes, Biopotential elec-trode, Brain-computer interfaces, Dry-contact sensors,Electrode-skin impedance, Electroencephalography, Wear-able sensors

I. INTRODUCTION

ELECTROENCEPHALOGRAPHY (EEG) is one of themajor modalities for non-invasive brain measurement.

EEG has recently become widely used not only for stan-dard neurological applications but also long-term monitor-ing [1], [2], mobile monitoring [3]–[7], sports [8]–[10] andrehabilitation applications [11], [12] as well as brain computerinterfaces [13],[14].

Conventional electrode technologies are most-often basedon Silver/Silver-Chloride (Ag/AgCl) electrodes contacting thescalp via application of liquid gel or paste-like electrolytematerials [15]–[17]. Application of this kind of electrodesis time-consuming and costly as it requires specially trainedmedical staff in order to clean and abrade the skin separatelyat each electrode site, applying electrolyte gel while avoidingelectrical shortcutting adjacent electrodes and subsequentlycleaning the volunteer’s head as well as the equipment.Moreover, electrolyte gels limit measurement times due todrying effects and corresponding changes in their electrochem-ical characteristics [15]. In addition, gel-running can causeunnoticed electrode shortcuts and falsify the measurementresults [18], [19]. These major disadvantages limit conven-tional electrodes to be applied only in well-controlled lab-environments.

During the last years multiple new sensor concepts forgel-free and ubiquitous EEG acquisition have been pro-posed including non-contact optical [20], [21] or capaci-tive [22]–[25] sensors as well as novel dry-contact electrodesin various shapes and contact materials [26]–[29].

Our research is focused on flexible multipin-shaped elec-trodes as a cost-effective, comfortable and easy to applyalternative to standard wet electrodes, while maintaining com-patibility with state-of-the-art biosignal amplification elec-tronics [27]–[30]. Moreover, our recent studies proved that

1534-4320 © 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

https://orcid.org/0000-0001-9196-0717

FIEDLER et al.: CONTACT PRESSURE AND FLEXIBILITY OF MULTIPIN DRY EEG ELECTRODES 751

Fig. 1. Multipin electrode: a), b) side and top views of the design scheme,c) perspective view of the 3D CAD model, and d) photograph of the finalAg/AgCl coated PU electrode.

these electrodes provide signal quality comparable to con-ventional electrodes not only when manually positioned insingle-electrode setups [27], [30] but also in high-densitymultichannel cap-setups [28], [29]. Our results also indicatedthat the signal quality of these dry electrodes relies on a stableand reliable electrode-skin contact at each electrode position.

In this paper, we investigate the influence of differentsubstrate flexibilities, applied electrode-skin normal force andresulting contact pressure on contact impedance. Appliedforce, resulting contact pressure and flexibility are the majorinfluencing factors for the electrodes’ ability to pass the hairlayer, to adapt to the head shape, and to establish a stable butcomfortable contact. Contact impedances below the amplifier’scut-off value are essential for signal recording.

II. MATERIALS AND METHODS

A. Electrodes

A dry non-invasive electrode must comprise a designenabling passing through the hair layer, while ensuring areliable, stable electrode-skin contact and sufficient patientcomfort. Previously, we proposed multipin EEG electrodesand investigated different designs, pin diameters and coat-ings [27], [29], [30] proving Ag/AgCl coated flexible, dryelectrodes to provide qualitatively similar signal quality likeconventional wet electrodes. In this paper we use an incremen-tally improved version of the earlier designs with an increasednumber of 30 single pins with spherical tops of a diameter of1 mm, a height of 6 mm, and pin distances of 2.4 mm (centerto center). All pins are integrated onto a common baseplate.The design of the multipin electrode is shown in Fig. 1.

Polyurethane (PU) is applied as the substrate material. It isa flexible material, which is moldable and hence allows forfree shape specification and easy, cost-efficient production.Furthermore, several PU types are available, which providebiocompatibility and different grades of shore hardness. Theamount of flexibility is represented by the shore hardness ofthe material must be carefully selected in order to fulfill twoopposing demands: a stiff material enables easy and rapidpassing through the hair layer and provides mechanical sta-bility. In contrast, a soft material increases the patient comfortand enables shape adaptability to the local head geometry.In this study we compared four different types of thermoset

Fig. 2. Exploded assembly drawing of the main components of the forcemeasurement applicator (FMA): a) Multipin electrode inside electrodefixation support; b) contact plate with silicone skin protection; c) forcesensor; d) runner; e) counter-acting cylinder; f) base plate with fixationholes, and g) precision screw. Minor components and fixation screwshave been excluded for the purpose of clarity. The dotted line runninga) through g) indicates the normal direction of the force.

PU exhibiting shore values of A98 (Biresin U1419, SikaChemie GmbH, Bad Urach, Germany), A90 (Vacuum CastingResin 7190, MCP HEK Tooling GmbH, Lübeck, Germany),A80 (Vacuum Casting Resin 7180, MCP HEK Tooling GmbH)and A70 (Vacuum Casting Resin 9070, MCP HEK ToolingGmbH).

All substrates were cleaned by means of an ultrasonicbath in isopropanol followed by rinsing with distilled water.Subsequently, the PU multipin electrodes were silver coatedduring a multi-phase chemical coating process. Finally, allelectrodes were chlorinated. The electroless chemical coatingprocedure produces well adherent, dense thin-film coatings.Ag/AgCl is well known for its electrochemical characteristicsand signal quality [15] and has already been successfullyapplied in dry electrodes [29].

B. Force application and sensor assembly

In order to ensure a quasi-constant force application andsimultaneously measure electrode-skin impedance as well asapplied normal force, a force application and measurementdevice was developed, which we call force measurement appli-cator (FMA). Fig. 2 shows an exploded assembly drawing ofthe developed FMA comprising seven main components. Thecontact plate (Fig. 2b) is positioned directly on the skin andinhibits extensive tilting of the overall assembly. It is equippedwith a silicone ring in order to increase comfort of thevolunteer and avoid skin damage during the measurement. Thecontact plate is fixated via a counter-acting cylinder (Fig. 2e)to the cover plate (Fig. 2f). These three components establishthe static part of the FMA, which is fixated on the head of eachvolunteer using non-elastic headbands. The runner (Fig. 2d) isplaced inside the counter-acting cylinder. Its vertical positioncan be adjusted using the precision screw (Fig. 2g) insertedinto the cover plate. The precision class 0.1 force sensorKD24s (ME-Messsysteme GmbH, Henningsdorf, Germany) isinserted into and fixated on the runner. On top of the forcesensor a multipin electrode is mounted using a specific fixationsupport (Fig. 2a). Electrode, force sensor, runner, and precisionscrew establish the dynamic part of the FMA.

The runner can perform a guided translational movementinside the FMA without the friction between runner andcounter-acting cylinder influencing the force measurement.


Fig. 3. Overall measurement setup for simultaneous impedance andforce measurement using a 2-port impedance measurement setupbetween two test positions Fpz and T4 and a conventional wet Ag/AgClelectrode at Cz position. The photographs on the right show the applica-tion of the FMA for the two positions in a top and side view.

The force sensor allows for measurement of normal forcesof ±10 N and was calibrated at 0 and 5 N prior to themeasurements. A calibration validation resulted in an accuracyof −5.5×10−3 N ± 0.5×10−3 N. The application procedureof the FMA ensures minimization of forces directed awayfrom the normal direction (see Fig. 2) on the sensor to reducemeasurement errors.

C. In-vivo tests

Electrode-skin impedances were measured for a frequencyrange of 5 Hz to 10 kHz using a commercial impedanceanalyzer (4192A LF, Hewlett Packard Company, Palo Alto,USA) in a standard two-port setup. The applied test signalhas a sinusoidal shape and selectable frequency. The maxi-mum measureable impedance for all frequencies is 1.3 M�.Therefore, we define the normal force Fmin as the minimumforce which is required to measure impedances below 1.3 M�at the lowest test frequencies of 5 Hz and 10 Hz.

For each volunteer and shore hardness two different elec-trode positions were tested: the hairless frontal Fpz positionand the hairy temporal T4 position (position names accordingto the international nomenclature for electrode positions inEEG [31]). For both positions, the impedances were mea-sured independently between the dry multipin electrode anda conventional Ag/AgCl ring electrode (ANT B.V., Enschede,Netherlands) at Cz position. The positions were selected dueto similar distances to Cz. Furthermore, the head geometryat Fpz and T4 is relatively flat, minimizing non-normal forcedirection influences on the sensor assembly. The overall mea-surement setup is shown in Fig. 3.

The in-vivo tests were performed in two phases: In phase I,the influence of electrode flexibility (shore hardness) onelectrode-skin impedance and wearing comfort was investi-gated within an applied range of normal forces ranging fromFmin to 4 N. All electrodes in phase I had a constant numberof 30 pins. In phase II, the influence of the electrode pin

Fig. 4. Top views of the tested electrodes with pin numbers involvinga) 30, b) 19, c) 14, and d) 10 pins for phase II of the study.

number on electrode-skin impedance was investigated for areduced range of applied normal forces ranging from 1 to 3 N.In phase II, the flexibility was constant using shore hardnessA98 only. The pin arrangement and inter-pin distances on thebaseplate were constant, while the different pin numbers wereimplemented by stepwise reducing the outer rings of pins ofthe electrode. Thus, the investigated pin numbers are 30, 19,14 and 10 pins as shown in Fig 4.

All in-vivo tests were performed sequentially and inde-pendently for each shore hardness and pin number. Phase Iinvolved 12 volunteers (6 male, 6 female), while phase IIinvolved 10 volunteers (8 male, 2 female). The volunteers hadan average age of 26 ± 2 years, a healthy skin condition,normal nutrition, and a hair length ranging from bald patchesto 50 cm with an average of 22 ± 19 cm. Ethics committeeapproval was received prior to the study (ref. 2841-06/10)from the responsible committee at the university hospital Jena,Jena, Germany.

Prior to application, the skin at the electrode positions wascleaned using ethanol and a soft cloth. Next, the conventionalAg/AgCl ring electrode at Cz position was applied usingcommercial EEG gel (ECI Electro-Gel, Electro-Cap Inter-national Inc., Ohio, USA). The FMA was applied ensuringa stable contact of FMA and head at normal forces up to4 N. Subsequently, the comfort was rated by the volunteersand the impedance was measured. A scale ranging from1 (absolute comfort, no pain) to 10 (maximum imaginablepain) [32] was used. After each measurement, which lastedabout 30 min, the comfort was rated again. Both ratings didnot show statistically significant differences and consequentlywere averaged. All tests were performed at room temperatureof 22 ± 2 °C and average air humidity of 45 ± 5 %.

D. Data recording and conditioning

Data recording and processing were performed usingcustom MATLAB algorithms (The Mathworks, Natick,USA). For each measurement frequency, 5 sec of impedanceand force data were recorded using a sampling frequencyof 4.2 kHz (analogue force sensor and impedance analyzeroutputs). The first 2 sec of each recording were discardedin order to reduce settling influences on the results. Theremaining samples were averaged resulting in single meanforce and mean impedance values per parameter set in eachof the three repetitions.

The contact pressures pElectrode were calculate based onthe measured normal forces FElectrode according to (1):

PElectrode = FElectrode

n · APin= FElectrode

0.236.cm2 (1)


Fig. 5. Impedance at normal forces of 2 ± 0.1 N and different testfrequencies: a) absolute value, and b) phase shift of the impedance.Solid lines represent mean over all volunteers and three test repetitionsper volunteer; dotted lines represent standard deviation.

Herein, APin corresponds to shadow surface area of a singlepin with 1 mm top diameter, while n corresponds to the overallnumber of 30 pins per electrode. Please note that all forcemeasurements were performed in the direction of the gravitiyforce, therefore enabling subsequent conversion from N/cm2 tokg/cm2 dividing pElectrode by the standard gravity 9.81 m/s2.

We tested the statistical significance of the results obtainedfor the different parameters, applying the non-parametricFriedman’s test [33] using a significance level p = 0.05.

III. RESULTS

A. Electrode-skin impedances

Fig. 5 shows the impedance measurement results over theinvestigated frequency range at an applied force of 2 N. Themean impedances are decreasing from 41±20, 35±14, 39±23,and 49 ± 32 k� at 5 Hz test frequency to 5 ± 2, 5 ± 2, 5 ± 3,and 7±3 k� at 10 kHz for the shore hardness values of A98,A90, A80 and A70, respectively. In the same frequency rangethe phase shift changes from −51° (shore A98, A90), −49°(A80), and −46° (A70) at 5 Hz to −3° (shore A98, A90, A70)and −4° (shore A80) at 10 kHz. Consequently, the electrode-skin impedance shows the typical semi-capacitive character-istics of a dry electrode-skin contact [17], [27]. In the rangeof 1−40 Hz, the phase shift is below 10° for all shore values.A qualitatively similar behavior is observed for the forcesof 1 N, 3 N, and 4 N.

Fig. 6 shows the results of the simultaneous impedance andnormal force measurements at a test frequency of 10 Hz for

the 12 volunteers, three repetitions and both electrode testpositions (Fig. 6a & 6c and 6b & 6d for temporal and frontalpositions, respectively). For shore hardness A98, the orderof magnitude as well as the variation of the electrode-skincontact impedance decreases with increasing normal forcefrom 348±236 k� and 257±207 k� for Fmin to 29±14 k�and 23 ± 11 k� for 4 N at the frontal and temporal positions,respectively. A similar trend is visible for the other shorehardness values. With exception of Fmin , the mean and theSTD of the electrode with shore hardness A70 are the highestamong all tested electrodes, especially when tested at temporalposition, with values from 517 ± 298 k� and 491 ± 305 k�at Fmin down to 28 ± 16 k� and 40 ± 29 k� at 4 N. Theimpedances of shore A70 significantly differ from electrodeswith higher shore at all applied normal forces with p = 0.017.For normal forces above 2 N all impedances are reliably andreproducibly below 250 k� at both test positions. For normalforces above 2 N the differences in measured impedancevalues between the shore hardness values A90 and A98 arenegligible (p = 0.21). They are the lowest values amongall tested shore hardness values. A significant difference inimpedance between frontal and temporal electrode positionsis visible for normal forces of 1 and 2 N (p = 9.7 × 10−4)when comparing frontal and temporal positions. In contrast, nosignificant differences between both positions exist for normalforces of 3 and 4 N (p = 0.71).

B. Minimum required force

The minimum force Fmin that was necessary to achieveimpedances below 1.3 M� is depicted in Fig. 7 for bothpositions and all four shore hardness values. The necessaryforces at temporal position are 0.6 ± 0.1 N and significantlydiffer (p = 1.0 × 10−5) to Fmin at the frontal positionbeing 0.2 ± 0.1 N. An influence of the shore hardness on theminimum required normal force cannot be observed. Duringthe three repetitions of the force and impedance measurementsin each volunteer, no significant set-reset influences wereobserved (p = 0.43).

C. Electrode pin number

The electrode-skin impedance at varied normal forcesachieved with different electrode pin numbers are shown inFig. 8. Both the scatter plot (Fig. 8a) and the averagedresults (Fig. 8b) show that the investigated pin numbers haveno considerable influence on the mean impedance or theimpedance variation. However, due to the different contactsurface, the applied normal forces cause increasing contactpressure when decreasing the pin number. Due to the differentcontact surface and resulting adduction pressure, we did notperform a statistical test for these data.

D. Comfort rating

In Fig. 9 the results of the comfort rating according tothe 1–10 scale are shown for the frontal (a) and temporal(b) positions. The results show decreasing comfort whenincreasing the force with values from 1 (maximum comfort)


Fig. 6. Simultaneous impedance and force measurements for electrodes of all four shore hardness values, all 12 volunteers and all measurementrepetitions at a test frequency of 10 Hz for phase I: a), c) between Cz and frontal position Fpz; b), d) between Cz and temporal position T4;a), b) scatter plot of all measurements; c), d) box plots of the results averaged over the desired force value and shown without outliers (99%confidence interval). All measurements were done with electrodes comprising 30 pins. The resulting contact pressure values are given on thesecondary abscissa.

observed at Fmin for all shore hardness values up to 3 ± 1.2for A98 and 2.3 ± 1.0 for A70 at 4 N for the frontal positionas well as 2.3 ± 1.2 and 2.2 ± 0.9 at temporal position,respectively. In all tests, the observed comfort at temporalpositions is significantly higher (lower rating values) than forthe frontal position (p = 0.006). Furthermore, electrodes withhigher shore hardness are perceived less comfortable at frontalposition for normal forces above 2 N (p = 0.02). Moreover,the volunteers reported the electrodes with shore A98 to bethe least comfortable among all electrodes. No considerabledifferences were observed between ratings prior and postimpedance measurements ( p ≥ 0.29) with an approx. timeof 30 min between both ratings.

IV. DISCUSSION

We investigated the influence of electrode flexibility, nor-mal force and resulting contact pressure on electrode-skininterfacial impedances and perceived wearing comfort for twoelectrode positions on the head. Our findings show normalforces above 2 N to enable multipin-shaped electrodes of shorehardness A98 and A90 to pass through the hair layer andreproducibly and reliably establish sufficiently low electrode-skin impedances. Simultaneously, normal forces below 2 N

Fig. 7. Boxplot of the minimum required normal force for impedancesbelow 1.3 MΩ plotted for the different shore hardness values and bothelectrode positions.

are perceived comfortable by the majority of volunteers withrating values below 3 on a 1–10 scale.

Our test setup and the developed FMA showed to beadequate for simultaneous measurement of force and


Fig. 8. Impedance and contact pressure for electrodes of all investigatedpin numbers, all 10 volunteers, both positions, and all measurementrepetitions investigated in phase II at a test frequency of 10 Hz: a) scatterplot of all results; b) mean (dots) and standard deviation calculated overall volunteers and both positions. All measurements were done withelectrodes of shore hardness A98.

impedance with low errors. The applied force sensor isof precision class 0.1. The overall setup was tested beforethe study on volunteers was carried out. In the ideal casewithout lateral forces, the prior test results indicated a mea-surement error of −5.5 × 10−3 N ± 0.5 × 10−3 N at5 N applied force. Furthermore, we tested a worst casescenario, where 5 N were applied at a working angle of 10°.This led to maximum measurement errors between +3.1%and −1.9%. The impedance analyzer was used according tospecifications (30 min warmup; 23 ± 5 °C room tempera-ture; auto zero; spot measurement) and provides an accuracyof 0.15% (absolute value of the impedance) between 1 � and1 M� as well as a frequency accuracy of ± 50 ppm.

The observed impedance graphs comply very well with theexpected semi-capacitive electrode characteristics of the pro-posed dry Ag/AgCl coatings [17], [27]. In the frequency rangeof standard EEG (1-40 Hz) no considerable dependency of theimpedance on signal frequency was determined. The low phaseshift of <10° implies that the dry electrode-skin interface isessentially resistive in this frequency band.

Applying normal forces of 2 N or more leads to repro-ducibly low electrode-skin impedances. In fact, for 2 N ofapplied force all measurements of the electrodes with A98 andA90 shore hardness are below 250 k�, which is sufficient forstate of the art biosignal amplifiers. However, for A80 andA70 outliers are visible even for higher forces (cf. Fig. 6b).Moreover, an increased impedance variation is visible forthe temporal position in comparison to the frontal position,which can be attributed to the hair layer impeding electrode-skin contact on each electrode pin, thus increasing contactimpedance.

In a related study, Cuadrado et al. investigated the pressurepain threshold on different regions of the head [34]. Theyreported a decreasing threshold from frontal via temporal areasto occipital areas. This finding very well complies with ourown findings of generally higher comfort rating values at Fpzposition compared to T4 position.

According to Eq. 1 our investigated forces of 1, 2, 3, and4 N correspond to respective contact pressure values of approx.0.4, 0.9, 1.3, and 1.7 kg/cm2. Hence, with exception to 4 N,all applied forces are well below the pain thresholds of therespective head positions reported by Cuadrado et al. [34] forhealthy volunteers. Furthermore, the fact that no dependencywas determined between comfort evaluations at the beginningand after the measurements, indicate that the normal forcescan be applied over periods exceeding 1 hour.

The slightly improved comfort rating values observed forelectrodes with shore hardness A98 may correspond to thereduced ability of these electrodes to bend and adapt tolocal head curvature. Such lack of adaption can cause aninhomogeneous pressure distribution among the electrode pinsand thus pressure hotspots will decrease the perceived com-fort. In contrast, while the soft electrodes of shore hardnessA80 and A70 improve comfort, the order of magnitude andvariation of the impedances are increased, hence requiringhigher contact pressures. Additionally, the A70 electrodestend to bend on top of the hair layer thus compromising theability to pass through the hair. Consequently, for our proposedmultipin shape a shore hardness around A90 seems to be thebest compromise between comfort and applicability.

With regard to the required homogeneous and stable normalforce, the findings of this study comply very well withour previous observations when using multipin electrodes inmultichannel cap setups [28], [29]. In summary, a normalforce of min. 2 N (0.9 kg/cm2 contact pressure) in combinationwith a substrate shore hardness of A90 provides low and stableimpedances while maintaining a good wearing comfort of themultipin electrodes with 30 pins.

While it is possible to decrease the overall dimension ofthe electrode by decreasing the pin number, the decreasingcontact surface also increases contact pressure. A decreasingpin number might in practice also occur if an electrode isslightly tilted and not all pins reach the scalp. We conclude thatthis will have a small influence on the impedance. However,it leads to a less comfortable situation for the subject, a factwhich we can practically confirm from several hundreds ofdry EEG cap applications. Consequently, force application onelectrodes should ensure normal direction with respect to the


Fig. 9. Boxplot of the comfort rating results for all normal forces and shore hardness values: a) results at frontal position Fpz; b) results at temporalposition T4. Averaged values for both ratings and all volunteers are displayed. All measurements were done with electrodes comprising 30 pins. Theresulting contact pressure values are given on the secondary abscissa.

scalp. Electrodes with 30 and 19 pins both provide a goodcompromise between overall electrode dimensions and low,stable electrode-skin impedance without extensive influence ofvariation of normal force on contact pressure. Furthermore, ourresults may contribute to optimized electrode support designof other types of pin shaped dry electrodes.

Our values for the required contact pressure for low andstable electrode-skin impedances are higher than the onesreported by Cömert and Hyttinen [35]. The different materials,the different mechanical setups, the additional moistening andthe different testing sites on the human body can explain thelower values for the electrodes in [35].

As an outcome of this study, our future cap-designs for drymultipin electrodes will facilitate the required normal forcelevel by implementing specific textile cuts and elastic pas-sive adduction structures. Furthermore, integrated adductionmechanisms [28] will be investigated to enable position-selective force application. Additionally, position-specific elec-trode pin designs (e.g. reduced length for frontal electrodepositions) may contribute to a further increase of the overallwearing comfort in multichannel setups. An optimized mul-tichannel cap system ensuring a stable, reliable and repro-ducible electrode-skin contact will contribute to increase dryEEG signal quality and reduce artifacts arising from relativemovements between electrode and skin. In combination withrecent state-of-the-art online processing and artifact detectionalgorithms [7], [36], [37] the applicability of dry EEG forclinical and non-clinical applications will be further improved.

V. CONCLUSION

We investigated the influence of substrate flexibility, pinnumber, normal force, and resulting contact pressure on theelectrode-skin impedance and perceived wearing comfort ofmultipin electrodes. We identified optimal values for bothparameters that will be implemented in future dry electrodecap designs.

REFERENCES

[1] K. G. Jordan, “Emergency EEG and continuous EEG monitoringin acute ischemic stroke,” J. Clin. Neurophysiol., vol. 21, no. 5,pp. 341–352, 2004.

[2] V. Michel, L. Mazzola, M. Lemesle, and L. Vercueil, “Long-term EEGin adults: Sleep-deprived EEG (SDE), ambulatory EEG (Amb-EEG) andlong-term video-EEG recording (LTVER),” Clin. Neurophysiol., vol. 45,no. 1, pp. 47–64, 2015.

[3] S. Debener, F. Minow, R. Emkes, K. Gandras, and M. De Vos, “Howabout taking a low-cost, small, and wireless EEG for a walk?” Psy-chophysiology, vol. 49, no. 11, pp. 1617–1621, 2012.

[4] J. Askamp and M. J. van Putten, “Mobile EEG in epilepsy,” Int. J. Psy-chophysiol., vol. 91, no. 1, pp. 30–35, 2014.

[5] M. De Vos and S. Debener, “Mobile EEG: Towards brain activitymonitoring during natural action and cognition,” Int. J. Psychophysiol.,vol. 91, no. 1, pp. 1–2, 2013.

[6] S. Debener, R. Emkes, M. De Vos, and M. Bleichner, “Unobtrusiveambulatory EEG using a smartphone and flexible printed electrodesaround the ear,” Sci. Rep., vol. 5, Nov. 2015, Art. no. 16743.

[7] T. R. Mullen et al., “Real-time neuroimaging and cognitive monitoringusing wearable dry EEG,” IEEE Trans. Biomed. Eng., vol. 62, no. 11,pp. 2553–2567, Nov. 2015.

[8] T. Thompson, T. Steffert, T. Ros, J. Leach, and J. Gruzelier, “EEGapplications for sport and performance,” Methods, vol. 45, no. 4,pp. 279–288, 2008.

[9] E. Filho, M. Bertollo, C. Robazza, and S. Comani, “The jugglingparadigm: A novel social neuroscience approach to identify neuropsy-chophysiological markers of team mental models,” Frontiers Psychol.,vol. 6, p. 799, May 2015.

[10] G. Cheron et al., “Brain oscillations in sport: Toward EEG biomarkersof performance,” Frontiers Phsychol., vol. 7, p. 246, Feb. 2016.

[11] J. J. Daly and J. R. Wolpaw, “Brain–computer interfaces in neurologicalrehabilitation,” Lancet Neurol., vol. 7, no. 11, pp. 1032–1043, 2008.

[12] S. Comani et al., “Monitoring neuro-motor recovery from stroke withhigh-resolution EEG, robotics and virtual reality: A proof of concept,”IEEE Trans. Neural Syst. Rehabil. Eng., vol. 23, no. 6, pp. 1106–1116,Nov. 2015.

[13] L. F. Nicolas-Alonso and J. Gomez-Gil, “Brain computer interfaces, areview,” Sensors, vol. 12, no. 2, pp. 1211–1279, 2012.

[14] H.-J. Hwang, S. Kim, S. Choi, and C.-H. Im, “EEG-based brain-computer interfaces: A thorough literature survey,” Int. J. Human-Comput. Interact., vol. 29, no. 12, pp. 814–826, 2013.

[15] A. Searle and L. Kirkup, “A direct comparison of wet, dry and insu-lating bioelectric recording electrodes,” Physiol. Meas., vol. 21, no. 2,pp. 271–283, 2000.

[16] M. Teplan, “Fundamentals of EEG measurement,” Meas. Sci. Technol.,vol. 2, no. 2, pp. 1–11, 2002.

[17] R. Merletti, “The electrode–skin interface and optimal detection ofbioelectric signals,” Physiol. Meas., vol. 31, no. 10, pp. 154–156, 2010.


[18] L. L. Greischar et al., “Effects of electrode density and electrolytespreading in dense array electroencephalographic recording,” Clin. Neu-rophysiol., vol. 115, no. 3, pp. 710–720, 2004.

[19] D. M. Alschuler, C. E. Tenke, G. E. Bruder, and J. Kayser, “Identifyingelectrode bridging from electrical distance distributions: A survey ofpublicly-available EEG data using a new method,” Clin. Neurophysiol.,vol. 125, no. 3, pp. 484–490, 2014.

[20] S. A. Kingsley, S. Sriram, A. Pollick, and J. Marsh, “Photrode opticalsensor for electrophysiological monitoring,” Aviation Space Environ.Med., vol. 74, no. 11, p. 1215, 2003.

[21] M. S. Fernandes et al., “Hydrogel-based photonic sensor for a biopoten-tial wearable recording system,” Biosensors Bioelectron., vol. 26, no. 1,pp. 80–86, Sep. 2010.

[22] M. Gerloff, M. Oehler, S. Mitschke, and M. Schilling, “Lightweightcapacitive 8 channel-EEG-helmet,” Biomed. Tech. Proc., vol. 55,pp. 136–144, Oct. 2010.

[23] S. M. Lee, J. H. Kim, H. J. Byeon, Y. Y. Choi, K. S. Park, and S.-H. Lee,“A capacitive, biocompatible and adhesive electrode for long-term andcap-free monitoring of EEG signals,” J. Neural Eng., vol. 10, no. 3,p. 036006, 2013.

[24] L.-D. Liao et al., “Biosensor technologies for augmented brain–computer interfaces in the next decades,” Proc. IEEE, vol. 100,no. Special Centennial Issue, pp. 1553–1566, May 2012.

[25] Y. M. Chi, Y.-T. Wang, Y. Wang, C. Maier, T.-P. Jung, andG. Cauwenberghs, “Dry and noncontact EEG sensors for mobile brain–computer interfaces,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 20,no. 2, pp. 228–235, Mar. 2012.

[26] M. A. Lopez-Gordo, D. Sanchez-Morillo, and F. P. Valle, “Dry EEGelectrodes,” Sensors, vol. 14, no. 7, pp. 12847–12870, 2014.

[27] P. Fiedler et al., “Comparison of three types of dry electrodes forelectroencephalography,” Acta IMEKO, vol. 3, no. 3, pp. 33–37, 2014.

[28] P. Fiedler et al., “Multichannel EEG with novel Ti/TiN dry electrodes,”Sens. Actuators A, Phys., vol. 222, pp. 139–147, Jan. 2015.

[29] P. Fiedler et al., “Novel multipin electrode cap system for dry electroen-cephalography,” Brain. Topogr., vol. 28, no. 5, pp. 647–656, 2015.

[30] P. Pedrosa et al., “Ag:TiN-coated polyurethane for dry biopotentialelectrodes: From polymer plasma interface activation to the first EEGmeasurements,” Plasma Process. Polym., vol. 13, no. 3, pp. 341–354,2015.

[31] R. Oostenveld and P. Praamstra, “The five percent electrode systemfor high-resolution EEG and ERP measurements,” Clin. Neurophysiol.,vol. 112, no. 4, pp. 713–719, 2001.

[32] J. Scott and E. C. Huskisson, “Graphic representation of pain,” Pain,vol. 2, no. 2, pp. 175–184, 1976.

[33] M. Friedman, “The use of ranks to avoid the assumption of normalityimplicit in the analysis of variance,” J. Amer. Statist. Assoc., vol. 32,no. 200, pp. 675–701, Dec. 1937.

[34] M. L. Cuadrado et al., “Pressure pain sensitivity of the scalp in patientswith nummular headache: A cartographic study,” Cephalalgia, vol. 30,no. 2, pp. 200–206, 2010.

[35] A. Cömert and J. Hyttinen, “Investigating the possible effect ofelectrode support structure on motion artifact in wearable bio-electric signal monitoring,” BioMed Eng OnLine, vol. 14, p. 44,May 2015.

[36] I. Daly, R. Scherer, M. Billinger, and G. Müller-Putz, “FORCe: Fullyonline and automated artifact removal for brain-computer interfacing,”IEEE Trans. Neural. Syst. Rehabil. Eng., vol. 23, no. 5, pp. 725–736,Sep. 2015.

[37] U. Graichen, R. Eichardt, P. Fiedler, D. Strohmeier, F. Zanow, andJ. Haueisen, “SPHARA—A generalized spatial Fourier analysis formulti-sensor systems with non-uniformly arranged sensors: Applicationto EEG,” PLoS ONE, vol. 10, no. 4, p. e0121741, 2015.

contact pressure and flexibility of multipin dry eeg ... · and also with the biomagnetic center,...

Documents