Brain Topogr DOI 10.1007/s10548-015-0435-5

ORIGINAL PAPER

Novel Multipin Electrode Cap System for Dry Electroencephalography P. Fiedler1 • P. Pedrosa2 • S. Griebel3 • C. Fonseca2 • F. Vaz4 • E. Supriyanto5 F. Zanow6 • J. Haueisen1,7

•

Received: 20 February 2015 / Accepted: 29 April 2015 Ó Springer Science+Business Media New York 2015

Abstract Current usage of electroencephalography (EEG) is limited to laboratory environments. Self-application of a multichannel wet EEG caps is practically impossible, since the application of state-of-the-art wet EEG sensors requires trained laboratory staff. We propose a novel EEG cap system with multipin dry electrodes overcoming this problem. We describe the design of a novel 24-pin dry electrode made from polyurethane and coated with Ag/AgCl. A textile cap system holds 97 of these dry electrodes. An EEG study with 20 volunteers compares the 97-channel dry EEG cap with a conventional 128-channel wet EEG cap for resting state EEG, alpha activity, eye blink artifacts and checkerboard pattern reversal visual evoked potentials. All volunteers report a good cap fit and good wearing comfort. Average impedances are below

150 kX for 92 out of 97 dry electrodes, enabling recording with standard EEG amplifiers. No significant differences are observed between wet and dry power spectral densities for all EEG bands. No significant differences are observed between the wet and dry global field power time courses of visual evoked potentials. The 2D interpolated topographic maps show significant differences of 3.52 and 0.44 % of the map areas for the N75 and N145 VEP components, respectively. For the P100 component, no significant differences are observed. Dry multipin electrodes integrated in a textile EEG cap overcome the principle limitations of wet electrodes, allow rapid application of EEG multichannel caps by non-trained persons, and thus enable new fields of application for multichannel EEG acquisition. Keywords Biopotential electrode  EEG  ECG  Dry electrode

& P. Fiedler [email protected] 1

Institute of Biomedical Engineering and Informatics, Technische Universita¨t Ilmenau, 98693 Ilmenau, Germany

2

Departamento de Engenharia Metalu´rgica e de Materiais, Faculdade de Engenharia, Universidade do Porto, 4200-465 Porto, Portugal

3

Department of Mechanism Technology, Technische Universita¨t Ilmenau, 98693 Ilmenau, Germany

4

Centro de Fı´sica, Universidade do Minho, 4710-057 Braga, Portugal

5

IJN-UTM Cardiovascular Engineering Centre, Universiti Teknologi Malaysia, 81300 Johor Bahru, Malaysia

6

eemagine Medical Imaging Solutions GmbH, 10243 Berlin, Germany

7

Department of Neurology, Biomagnetic Center, Jena University Hospital, 07747 Jena, Germany

Introduction Electroencephalography (EEG) is an important tool for research and diagnostics of the human brain function. Due to improved electronics and sensor concepts various new fields of applications have recently been introduced, including brain computer interfaces (Daly and Wolpaw 2008; Nicolas-Alonso and Gomez-Gil 2012; Hwang et al. 2013), ambient assisted living (D’Angelo et al. 2010; Askamp and van Putten 2014), EEG analysis during professional sports (Thompson et al. 2008; di Fronso et al. 2015), rehabilitation (Daly and Wolpaw 2008; Steinisch et al. 2013) as well as mobile, ubiquitous and long-term monitoring (Jordan 2004; Debener et al. 2012; De Vos and Debener 2014; Askamp and van Putten 2014; Michel et al. 2015). However, these applications often require multichannel setups

123

Brain Topogr

in order to facilitate comprehensive signal processing, analysis and interpretation. Besides continuous efforts for the development of alternative sensor technologies, commercially available multichannel EEG systems with more than 64 channels are still based on conventional silver/silver-chloride (Ag/AgCl) electrodes in combination with electrolyte gels or pastes (wet electrodes). Hence, their application requires welltrained staff to perform time-consuming, extensive preparation (skin abrasion, gel application, impedance optimization), cleaning and disinfection (Teplan 2002). The electrolyte gel limits the overall measurement time due to drying and can cause skin irritation, as well as hair damage (Teplan 2002; Searle and Kirkup 2000). In addition, its use also increases the risk of measurement errors due to electrode-to-electrode gel bridges (Greischar et al. 2004). These drawbacks and the process-associated patient stress limit current routine EEG analysis applications and render conventional, gel-based EEG systems inapplicable for outof-the-lab environments. The disadvantages of wet electrodes lead to novel sensor concepts for bioelectric signal acquisition, including fully capacitive sensors, infrared non-contact sensors, dry and quasi-dry contact electrodes (recent reviews by Liao et al. 2012; Chi et al. 2012; Lopez-Gordo et al. 2014). However, none of these concepts has so far been applied to multichannel EEG with more than 64 channels. We recently proposed multipin-shaped dry-contact electrodes that pass optimally through the hair layer (Fiedler et al. 2014, 2015). In this paper, we present a novel 97-channel EEG cap with novel flexible, multipin-shaped, dry electrodes. We compare its performance with a standard wet electrode EEG cap in a study on 20 volunteers.

the material should be a compromise: while the material must be stiff enough to rapidly and reliably passthrough the hair layer, couple with the scalp and avoid extensive bending (e.g. causing variable pin shortcuts); upon adduction, the substrate should be compliant in order to provide patient comfort and enable flexible adaption to local head curvature. We selected the substrate material based on thermoset PU (Biresin U1419, Sika, Germany) with a shore A hardness of 98. Ag/AgCl is the gold standard material for wet electrodes and has outstanding electrochemical characteristics (Searle and Kirkup 2000; Fiedler et al. 2014; Pedrosa et al. 2015b). Thus, it has been selected for the electrically conductive coating of the multipin electrodes. After cleaning of the uncoated PU substrates, Silver (Ag) coatings were applied during a multi-phase chemical coating process and subsequently chlorinated. The chemical process produces welladherent and dense coatings. The compact Ag/AgCl layer exhibits a thickness of approx. 2 lm, anchored on silver roots that extend up to 7 lm into the PU substrate. The design of the multipin substrate as well as a photograph of an exemplary Ag/AgCl coated electrode are shown in Fig. 1. Dry and Conventional Cap Cap systems with integrated electrodes enable rapid application and reproducible positioning for a large set of electrodes. Furthermore, flexibility of the cap material ensures a good fit and is intended to provide homogeneous adduction at all electrode positions. A reliable adduction of dry electrodes is more relevant than for wet electrodes due

Materials and Methods Dry Electrodes The shape of the dry-contact electrodes must enable the passing through of the hair layer and provide sufficient contact surface, reliability and stability, as well as patient comfort (Fiedler et al. 2014). The developed electrode design consists of 24 single conic pins with circular tops of 1 mm in diameter and a height of 6 mm. These pins are integrated onto a common circular base plate. The centerto-center distances of the pins are 2.5 mm. The design is an improved version of our previously presented multipin electrodes (Fiedler et al. 2014; Pedrosa et al. 2015a). A polyurethane (PU) substrate material enables free shape specification and easy production due to its moldability. PU materials are biocompatible and the shore hardness can be adjusted. The selected shore hardness of

123

Fig. 1 Electrode design and substrate material: a, b side views of the design scheme and photographs of the Ag/AgCl coated electrode, respectively; c, d top views of the scheme and photographs of the electrode

Brain Topogr

to the absence of gel or paste bridging the distance between electrode and scalp. The novel dry EEG cap comprises 97 electrodes, which were integrated into the fabrics of a double-layer textile EEG cap (Waveguard EEG caps, ANT B.V., Enschede, The Netherlands). The number of dry electrodes was limited by additional space requirements for the electrode fixation mechanism. For the acquisition of wet EEG reference data, a conventional cap system with 128 Ag/AgCl electrodes (Waveguard-128, equidistant layout, ANT B.V., Enschede, The Netherlands) was used. In both caps, the electrodes have been arranged to form a quasi-equidistant triangular layout as shown in Fig. 2 in order to reduce the influence of polar average reference (PAR) effects in the signal comparison and validation (Jungho¨fer et al. 1999). The dry multipin cap prototype and the commercial cap are also shown in Fig. 2, turned inside out.

In-vivo Tests The in vivo tests were performed sequentially for the dry multipin cap and the conventional wet cap for an overall number of 20 volunteers with an average age of 30 ± 5 years. Ethics committee approval was acquired prior to the study. Informed consent was obtained from all individual participants included in the study. The volunteers had a healthy skin condition, normal nutrition, average head circumference of 58.1 ± 1.2 cm, and a hair length ranging from bald patches up to 30 cm with an average of 6 ± 5 cm. No skin preparation was performed prior to the application of the caps. After application of the conventional cap, the hair was moved aside and EEG gel (ECI Electro-Gel, Electro-Cap International Inc., OH, USA) was applied at all electrode positions using a syringe equipped with a blunt cannula. For the wet cap we used the

Fig. 2 Compared cap systems with equidistant electrode layouts: a Novel textile cap system with 97 dry multipin electrodes turned inside out, b conventional wet cap system with 128 electrodes turned inside out, c, d 2D topographic plots of the 97 dry and 128 wet electrodes, respectively

123

Brain Topogr

integrated patient ground electrode (additional electrode between the positions RA2, RA3, RR5, and RR6), while for the dry cap we placed a conventional Ag/AgCl ring electrode with EEG gel at the right mastoid position. Hence, in the current study the ground electrode for both the wet and the dry cap are conventional wet electrodes in order to provide a stable ground connection between amplifier and volunteer. EEG acquisition was performed using a commercial unipolar 128-channel biosignal amplifier (Refa-Ext, ANT B.V., Enschede, The Netherlands) with common average reference. For each volunteer and cap system, four different kinds of test signals were recorded: resting state EEG, alpha activity, eye blink artifacts, and a visual evoked potential (VEP). For all EEG episodes, the sampling rate was constant at 2048 samples/second. The VEP test consisted of a pattern reversal checkerboard paradigm with 300 trials and was performed according to the ISCEV 2010 standard (Odom et al. 2010). Alpha activity was provoked by asking the subject to close his/her eyes. The eye blinks were externally triggered at a rate of approx. 0.3 Hz. The data acquisition was performed using ASA Software (ANT B.V., Enschede, The Netherlands). Electrode–skin impedances were measured for all electrodes directly after cap application and prior to EEG acquisition. An additional impedance measurement was performed after the EEG acquisition for eight volunteers. For the conventional wet electrodes, an impedance threshold of 20 kX was set, while no threshold was set for the dry electrodes. In each case, the dry cap was tested prior to the wet cap in order to avoid any gel or cleaning influences on the skin or hair condition and consequently the dry electrode test results. Both tests were performed on the same day with a pause of 4 h ± 0.8 h between both sessions. In order to minimize influences of volunteer attention on the test results, we asked the volunteers to evaluate their attention level based on the Stanford sleepiness scale (Hoddes et al. 1972) prior to each EEG acquisition. All tests were performed at room temperature and average air humidity of 45 ± 5 %. Data Conditioning and Comparison All data processing was performed using custom MATLAB algorithms (The Mathworks, Natick, USA). The EEG signals were filtered using a 24 dB Butterworth band pass with cut-off frequencies at 1 and 40 Hz, as well as a 36 dB notch filter at 50 Hz. Artifact contaminated data were manually identified and excluded. Data episodes of 30 s for resting state and alpha activity were manually selected and the mean power spectral density was calculated using the

123

Welch estimation method. Visual inspection of all EEG data served for classifying operational electrodes. Nonoperational electrodes both had a zero line or high artifact contamination and were excluded from all analysis. After filtering the raw VEP data, artifact contaminated trials were manually identified and excluded before averaging the remaining trials. Subsequently, the global field power in time domain (GFPt) over all channels was calculated according to Eq. 1 (Lehmann and Skrandies 1984). U corresponds to the measured voltage amplitudes of the whole data sequence of 1024 samples (500 ms) of an electrode i or j out of the overall number of electrodes (m). rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 1 Xm Xm  GFPt ¼ Ui  Uj ð1Þ i¼1 j¼1 2m The GFPt describes the spatial standard deviation within the electrode set (max. 97 dry multipin electrodes or max. 128 conventional wet electrodes) at a distinct point in time. Assuming equal electrode distribution over the head, the GFPt is neither depending on the number nor the position of the electrodes. Thus, it enables a direct comparison between the respective dry and wet VEP results of the volunteers. For a quantitative evaluation of the wet and dry results, we calculated the root mean square deviation (RMSD) and the Spearman’s rank correlation (COR) between the respective wet and dry GFPt curves of each volunteer, according to Eqs. 2 and 3, respectively. The variables GFPtwi and GFPtdi correspond to the ith out of n data samples of the respective compared GFPt curves. sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Pn 2 i¼1 ðGFPtwi  GFPtdi Þ RMSD ¼ ð2Þ n   Pn  i¼1 GFPtwi  GFPtw GFPtdi  GFPtd COR ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 Pn  2ffi Pn  GFPtw  GFPtw GFPtd  GFPtd i i i¼1 i¼1

ð3Þ While the RMSD allows the comparison of absolute amplitudes of the GFPt during the analyzed EEG sequence, the correlation enables comparison of the general GFPt evolution over time. In addition to comparing wet and dry recordings, we calculated RMSD and correlation parameters for 10 split groups of 10 randomly selected volunteers in order to evaluate the variability within each cap type. The relative channel failure probability was determined for each dry multipin cap recording and volunteer. Therefore, three independent recording sequences of 120 s containing spontaneous EEG have been manually analyzed for each of the 20 volunteers. A channel failure was defined as a single channel’s signal with considerable artifacts during more than 20 % of an analyzed EEG sequence.

Brain Topogr

Statistics We compared the datasets obtained with the wet and dry EEG caps. Due to the different numbers and positions of the electrodes, we analyzed the following parameters: PSD, GFPt, 2D interpolated spatial distributions of the alphaband PSD and the three main components of the VEP. The PSD was analyzed on the level of single frequencies (1–40 Hz in 1 Hz steps) as well as EEG bands (mean of the PSDs within each EEG band). All listed parameters exhibited a non-normal distribution, which was confirmed by Kolmogorov–Smirnov tests (alpha-level 0.05). The statistical significance of parameter differences was tested by means of a Wilcoxon-Mann– Whitney U Test with alpha-level set to 0.01.

Results Application and Comfort The average time-to-recording was measured between the start of the application of the cap and the start of the first EEG test sequence. The required time for manual optical tracking of the single electrode positions was excluded due to its dependence of electrode type and number. For the novel 97 electrode cap the average time-to-recording was 7 min in comparison to the conventional 128 electrode cap with an average of 68 min. Before the start of the tests and after the end of the session, all volunteers reported a good cap fit and good wearing comfort. The reported attention levels varied between 2 and 4 and showed no systematic difference between the wet and dry recordings (respective means of 2.8 ± 0.7 and 2.7 ± 0.7). After removing the each electrode cap, the scalp of the volunteers was manually inspected for skin marks. In both cap types marks were found at the electrode-scalp contacts, namely the electrode pin tips (dry cap) and the silicone gel rings (wet cap). However, the volunteers reported no unpleasant skin sensation or pain, and the contact marks vanished after approx. 30 min. Impedances and Reliability The impedance values averaged over all 20 volunteers for the dry multipin electrodes are shown in Fig. 3. Two electrodes (R7 and R9) were damaged during the process of the study (white circles in Fig. 3). Out of the overall 97 electrodes, 92 electrodes show average impedance below 150 kX. Furthermore, 51 electrodes exhibit average impedances below 80 kX. The maximum observed average impedance was 164 kX. In

general, electrodes at frontal, temporal, and lower occipital positions show impedance levels below 100 kX and a lower standard deviation. In comparison, the impedance level as well as the standard deviation is higher at central and parietal positions. Within the 60 EEG test episodes, an average number of 83 % of the 97 electrodes was operational with adequate EEG signal quality (visual inspection); while in the worst case 76 % of the 97 electrodes were operational. The highest number of absolute channel failures occurred at electrodes Lm, Rm, and RE4 with 23, 21, and 18 respective failures during the 60 tests. As for the impedances, low occurrence of failures are associated with frontal, temporal, and lower occipital positions, while the failures increase at central and parietal positions. No considerable impedance drifts were observed for the dry electrodes during the overall measurement time of approx. 2 h. In average the impedance decreased by 6 kX with a standard deviation of the impedance change of 23 kX. Spontaneous EEG Figure 4 shows overlay plots of the mean power spectral density (PSD) and topographic mappings of the PSD for 30 s of recorded EEG containing prominent alpha activity. The mean PSD shown in Fig. 4a represents the grand average over all volunteers showing alpha activity (16/20 volunteers) and all channels, which are operational. The alpha activity is clearly represented by increased PSD between 8 and 13 Hz with main peaks at 9.4 and 10.6 Hz. The respective PSD values of the wet and dry electrodes are 20 ± 30 lV2/Hz and 19 ± 29 lV2/Hz for the main peak, as well as 9 ± 9 and 7 ± 6 lV2/Hz for the secondary peak. Overall, dry and wet PSDs for alpha and resting EEG show a similar spectral signal characteristic for the investigated frequency range of 1–40 Hz. However, the dry multipin electrodes show slightly increased PSD values for frequencies below 8 Hz compared to the wet signals. The statistical test showed no significant differences for the EEG band analysis. In case of the per-frequency analysis, we obtained significant differences for frequencies between 4 and 6 Hz with p-values ranging from 0.005 to 0.008. The topographic mappings in Fig. 4b, c show the 2D linear interpolated, cumulated PSD of the alpha band (8–13 Hz) for the wet and dry recordings, respectively. The color maps of the spatial mappings were normalized to the maximum of the wet and dry datasets. Furthermore, missing information for nonoperational channels was linearly interpolated based on the surrounding electrodes data. The maximum alpha activity is clearly visible in the lower parietal and occipital area. Both maps show similar spatial distributions. The main activity in the mapping for the dry

123

Brain Topogr

Fig. 3 Electrode-skin impedances and channel failures during the dry multipin cap tests: a Mean, and b standard deviation of the absolute electrode–skin impedance calculated over all 20 volunteers; c relative

channel failure probability manually determined during 60 recordings of 120 min overall length for all 20 volunteers. The two white circles indicate two damaged electrodes

Fig. 4 Comparison of the grand average of recordings of 30 s of alpha activity for the 16 out of 20 volunteers showing alpha activity: a power spectral density (PSD, Welch estimation) spectral overlay plot calculated; solid lines indicate mean over all channels and

volunteers, while dotted lines represent mean ± standard deviation; b, c 2D interpolated topographic mapping of the PSD per channel for the conventional and dry caps, respectively; grey markers indicate areas of significant differences between wet and dry topographies

electrodes seems to be slightly shifted towards occipital electrode positions. The statistical tests showed significant differences for 0.97 % of the area of the maps. Small differences are concentrated in the left central-parietal area at the area of LL7, LA4, and LA5 electrodes. A larger area of differences is located in the lower right head area around electrode LE4. Time domain overlay plots of spontaneous EEG containing externally triggered eye blink artifacts are shown in

Fig. 5a, b for recordings of 5 and 50 s, respectively. They were acquired by frontal electrodes at L1, LL1, R1, and RR1 positions. Similar signal shape is visible for both types of electrodes in the detailed 5 s recording, as well as in the 50 s long-time recording. For the eye blinks shown, the maximum amplitude of the dry electrode recordings is slightly higher than that observed in the wet electrode recordings. The baseline signals between the eye blinks are very similar for dry and wet electrodes.

123

Brain Topogr

Fig. 5 Overlay plot of spontaneous EEG recordings of an exemplary volunteer recorded by the frontal electrodes at L1, LL1, R1, and RR1 positions and filtered 1–40 Hz: a overlay of 5 s, and b overlay of 50 s

Evoked Potentials In Fig. 6, the results of the checkerboard pattern reversal VEP are shown for a period of 100 ms pre-stimulus up to 400 ms post-stimulus. The results represent the grand average over all volunteers. Amplitudes and latencies of the main components are similar in the butterfly plots (Fig. 6a,

Fig. 6 Grand average of a pattern reversal visual evoked potential (VEP): a, b butterfly plot of conventional and dry electrode channels, respectively; c, d global field power; e, f 2D linear interpolated topographic mapping of the three main components of the VEP,

b) as well as the GFPt (Fig. 6c, d). The GFPt amplitude of the main components are 9.8 lV at 73.3 ms, 24.0 lV at 129 ms, and 12.5 lV at 196.4 ms post-stimulus for the conventional wet cap and 8.3 lV at 74.8 ms, 20.5 lV at 125.6 ms, and 9.6 lV at 196.9 ms for the dry multipin cap. The RMSD and correlation were calculated between the respective wet and dry GFPt for each individual volunteer.

normalized to the respective maximum amplitude; grey markers indicate areas of significant differences between wet and dry topographies

123

Brain Topogr

The mean RMSD between both cap types calculated over all volunteers is 2.9 ± 0.8 lV and the mean correlation is 0.8 ± 0.2. For the 10 split groups of 10 randomly selected volunteers the mean RMSD values are 2.6 ± 1.5 and 2.1 ± 1.3 lV for the wet and dry caps, respectively. The mean correlation of the split groups is 0.9 ± 0.1 for both caps. Figure 6e, f, respectively, depict 2D linear interpolated topographic mappings of the three main components N75, P100 and N145 (at the aforementioned maximum signal amplitude) of the wet and dry VEP recordings. The spatial maps show the distribution and orientation expected for those VEP components. Furthermore, the wet and dry results are overall very similar. The wet P100 distribution seems to be slightly more homogeneous than the dry P100 distribution. Furthermore, the extension of the negative potential distribution of the dry N75 component seems to be slightly increased in the occipital area. For the GFPt courses of all volunteers, we found no significant differences between the wet and dry recordings. The topographic maps show significant differences of 3.52 and 0.44 % of the map areas for the N75 and N145 VEP components, respectively. The differences appear in the left temporal (N75), as well as left and right central-temporal (N145) area of the head. For the P100 component no significant differences were determined.

Discussion We developed and successfully validated the function of a novel 97-electrode multipin electrode cap for dry, passive EEG acquisition. We compared the novel cap to a conventional 128 channel gel-based Ag/AgCl electrode cap with regard to time-to-recording, electrode–skin impedance and EEG signal quality in time and frequency domain. Our findings prove a comparable signal quality for both systems. The time-to-recording of 7 min is a reduction of 89 % in comparison to the conventional 128-channel cap and a reduction of 86 % when normalized to the number of electrodes per cap. Consequently, the preparation costs are similarly reduced by not only the costs of consumables (gels, pastes) but also laboratory staff costs. The new cap can even be applied by the volunteers or patients themselves. With regard to drying effects of the electrolyte materials in wet electrodes, the principle application time of dry electrodes is not limited. The measured electrode–skin impedances with average values below 150 kX for 94 % of the channels in combination with the stable open circuit potential known to be attributed to silver and silver/silver-chloride electrodes (Fiedler et al. 2014; Pedrosa et al. 2015b) render the

123

electrodes compatible with commercial amplifiers without the need for active pre-amplification or impedance converters. The impedance differences between hairless (frontal) and hairy (temporal and occipital) positions are negligible. Hence, the proposed electrode shape proved to be able to pass reliably and reproducibly through the hair layer in a multichannel setup in a study on 20 volunteers, which is in accordance with our previous findings (Fiedler et al. 2010, 2014). Furthermore, the distribution of impedances on the head emphasizes that high impedances are mainly related to insufficient adduction at central and parietal positions rather than to electrode shape or electrochemical characteristics. The average number of 83 % of operational electrodes is in accordance with the generally low impedance level. Likewise, the distribution of channel failures is similar and may again be attributed to insufficient contact pressure and contact stability. 55.4 % of non-operational electrodes were isoelectric (no contact or impedance above 256 kX), while 44.6 % of these electrodes are attributed to artifact related exclusion. Unfortunately, two electrodes (R7, R9) have been damaged at the electrode-cable interface during the study, accounting for 10.9 % of the overall non-operational electrodes. These electrodes have thus been excluded from further analysis. Future work will include an improved electrode-cable interface. With regard to signal characteristics and quality, no systematic differences have been observed in the PSD for resting state EEG, alpha activity and eye blink recordings. Differences in specific frequencies (4–6 Hz) might be attributable to the sequential recording setup and thus, intraindividual differences. Amplitude differences of single channels can be caused by intra-individual EEG signal variation, as well as slight individual variations of the cap and thus electrode positions. Small differences in spatial distribution of VEP and alpha activity patterns can also be attributed to differences in electrode density and distribution, as shown in Fig. 2c, d. The areas of significant differences for alpha and VEP topographical maps are located at electrodes with increased impedances and failure probability. Consequently, unstable signal contacts at these positions may have caused the apparent deviations. Furthermore, the amount of differences in the lower occipital area is influenced by the low electrode density and thus large interpolation area. Moreover, the significant differences for alpha and VEP topographical maps are partly at positions with low signal amplitude. The current study focuses on the proof of principle for the application of the novel dry electrode technology in the standard EEG range (1–40 Hz) as well as under lab conditions. According to our experience and in line with the electrochemical properties of our dry electrodes, dry and wet electrodes signals are expected to show no differences

Brain Topogr

for frequencies above 40 Hz. The applicability of our dry electrodes below 1 Hz remains to be investigated. The computed RMSD values between both compared VEP results were below 10 % with respect to the absolute amplitudes. This finding, in combination with the high correlation coefficient of [0.8 between wet and dry GFPt as well as the results of the split group comparison, indicate that most of the differences between the wet and dry EEG recordings are caused by inter-trial variability. Further influences include electrode density, environmental noise as well as a small PAR effects caused by the different electrode density of both caps (Jungho¨fer et al. 1999). A conventional wet electrode was applied for the ground electrode in order to provide a stable and reproducible ground connection. Alternatively, a self-adhesive hydrogelbased electrode can be applied in order to increase ease of application. However, we are currently working on an integrated dry ground electrode approach at a hairless position that allows for complete elimination of gels or hydrogels and their corresponding drawbacks. During the EEG tests under laboratory conditions we did not observe extensive movement artifacts for the dry electrode cap. The order of magnitude of such artifacts was comparable to those of the conventional wet cap. However, dry electrodes may be more susceptible to movement artifacts, especially when strong accelerations occur (e.g. during running), due to the absence of gel at the electrode– skin interface. The order of magnitude, the global influence of movement artifacts, their reliable automatic detection as well as interpolation of bad channels and EEG sequences remain to be investigated. In contrast to dry EEG with only a few channels, dry high-density EEG can provide a basis for comprehensive post-processing algorithms such as spatial harmonics analysis (Graichen et al. 2015) in order to allow for the calculation of artifact-free EEG even under poor recording conditions. Current electrode design limits the overall number of electrodes to approx. 100 electrodes due to an additional rim of 2 mm around the pin arrangement. However, this rim and hence the limitation of electrode number is solely related to the current electrode fixation mechanism. An adaptation of the current design will allow eliminating the fixation rim, consequently reducing the overall electrode size and further increasing the electrode density up to more than 130 electrodes. Equidistant electrode arrangement was selected to minimize PAR effects (Jungho¨fer et al. 1999) and furthermore allow future application of automatic artifact detection and channel interpolation algorithms. However, dry electrode layouts are not restricted to equidistant arrangements, but can also be implemented according to conventional and extended 10–20 layouts (Klem et al. 1999). The electrode adduction force and the ability of the electrode to pass through the hair are the remaining major

influences on the number of operational electrodes. If the right amount of adduction force is applied to the multipin electrodes, the round pin tops cause small pressure marks on the scalp, hence improving contact surface, reducing contact impedance and limiting relative movements and artifacts. Hence, a reproducible, homogeneous and stable adduction will contribute to a further increase of reliability of the dry cap system. Future work will include the development of an adapted textile cut and different dimensions of caps. Furthermore, integrated adduction mechanisms (Fiedler et al. 2015) and different electrode pin length for different head areas will be investigated. A dedicated study of the interrelation of contact pressure, substrate hardness, interfacial impedance and patient comfort will be performed separately and will contribute to the determination of the optimal contact pressure and electrode flexibility, comprising both stable scalp contact and patient comfort.

Conclusion In conclusion, we proposed a novel dry EEG cap system with multipin electrodes for rapid multichannel AC-EEG. Possible applications of this technology go beyond basic or clinical laboratory research and enable out-of-the-lab, ubiquitous, and long-term EEG acquisition. Acknowledgments This work was financially supported by the German Federal Ministry of Education and Research (03IPT605A), the German Academic Exchange Service (D/57036536), the Thu¨ringer Aufbaubank and the European Social Fund (2012 FGR 0014), and the European Union (FP7-PEOPLE Marie Curie IAPP project 610950).

References Askamp J, van Putten MJ (2014) Mobile EEG in epilepsy. Int J Psychophysiol 91(1):30–35. doi:10.1016/j.ijpsycho.2013.09.002 Chi YM, Wang YT, Wang Y, Maier C, Jung TP, Cauwenberghs G (2012) Dry and noncontact EEG sensors for mobile braincomputer interfaces. IEEE Trans Neural Syst Rehabil Eng 20(2):228–235. doi:10.1109/TNSRE.2011.2174652 D’Angelo LT, Parlow J, Spiessl W, Hoch S, Lueth TC (2010) A system for unobtrusive in-car vital parameter acquisition and processing. In: Proceedings of the 4th international conference on pervasive computing technologies for healthcare 2010, Munich, Germany, New York, pp 1–7 Daly JJ, Wolpaw JR (2008) Brain-computer interfaces in neurological rehabilitation. Lancet Neurol 7(11):1032–1043. doi:10.1016/ S1474-4422(08)70223-0 De Vos M, Debener S (2014) Mobile EEG: towards brain activity monitoring during natural action and cognition. Int J Psychophysiol 91(1):1–2. doi:10.1016/j.ijpsycho.2013.10.008 Debener S, Minow F, Emkes R, Gandras K, De Vos M (2012) How about taking a low-cost, small, and wireless EEG for walk? Psychophysiology 49:1449–1453. doi:10.1111/j.1469-8986. 2012.01471.x

123

Brain Topogr Di Fronso S, Bertollo M, Comani S (2015) Neural markers of performance states in an Olympic Athlete: a case study in airpistol shooting. J Sport Sci Fiedler P, Brodkorb S, Fonseca C, Vaz F, Zanow F, Haueisen J (2010) Novel TiN-based dry EEG electrodes: Influence of electrode shape and number on contact impedance and signal quality. In: Panagiotis DB, Pallikarakis N (eds) IFMBE proceedings of the xii mediterranean conference on medical and biological engineering and computing, 2010 May 27–30, Chalkidiki, Greece. Springer, Berlin, pp 418–421 Fiedler P, Haueisen J, Jannek D, Griebel S, Zentner L, Vaz F, Zanow F, Haueisen J (2014) Comparison of three types of dry electrodes for electroencephalography. Acta Imeko 3(3):33–37 Fiedler P, Griebel S, Pedrosa P, Fonseca C, Vaz F, Zentner L, Zanow F, Haueisen J (2015) Multichannel EEG with novel Ti/TiN dry electrodes. Sens Actuator A 221:139–147. doi:10.1016/j.sna. 2014.10.010 Graichen U, Eichardt R, Fiedler P, Strohmeier D, Zanow F, Haueisen J (2015) SPHARA—A generalized spatial Fourier analysis for multi-sensor systems with non-uniformly arranged sensors: application to EEG. PLoS One Greischar LL, Burghy CA, van Reekum CM, Jackson DC, Pizzagalli DA, Mueller C, Davidson RJ (2004) Effects of electrode density and electrolyte spreading in dense array electroencephalographic recording. Clin Neurophysiol 115(3):710–720. doi:10.1016/j. clinph.2003.10.028 Hoddes E, Dement W, Zarcone V (1972) The development and use of the Stanford sleepiness scale (SSS). Psychophysiology 9:150 Hwang HJ, Kim S, Choi S, Im CH (2013) EEG-based brain-computer interfaces: a thorough literature survey. Int J Hum Comput Interact 29(12):814–826. doi:10.1080/10447318.2013.780869 Jordan KG (2004) Emergency EEG and continuous EEG monitoring in acute ischemic stroke. J Clin Neurophysiol 21(5):341–352 Jungho¨fer M, Elbert T, Tucker DM, Braun C (1999) The polar average reference effect: a bias in estimating the head surface integral in EEG recording. Clin Neurophysiol 110(6):1149–1155 Klem GH, Luders HO, Jasper HH, Elger C (1999) The ten-twenty electrode system ofthe International Federation. The International Federation of Clinical Neurophysiology. Electroencephalogr Clin Neurophysiol Suppl 52:3–6 Lehmann D, Skrandies W (1984) Spatial analysis of evoked potentials in man—a review. Progr Neurobiol 23(3):227–250

123

Liao LD, Lin CT, McDowell K, Wickenden AE, Gramann K, TzyyPing J, Li-Wei K, Jyh-Yeong C (2012) Biosensor technologies for augmented brain-computer interfaces in the next decades. Proc IEEE 100:1553–1566. doi:10.1109/JPROC.2012.2184829 Lopez-Gordo MA, Sanchez-Morillo D, Pelayo Valle F (2014) Dry EEG electrodes. Sensors 14(7):12847–12870. doi:10.3390/ s140712847 Michel V, Mazzola L, Lemesle M, Vercueil L (2015) Long-term EEG in adults: sleep-deprived EEG (SDE), ambulatory EEG (AmbEEG) and long-term video-EEG recording (LTVER). Clin Neurophysiol. doi:10.1016/j.neucli.2014.11.004 Nicolas-Alonso LF, Gomez-Gil J (2012) Brain computer interfaces, a review. Sensors 12(2):1211–1279. doi:10.3390/s120201211 Odom JV, Bach M, Brigell M, Holder GE, McCulloch DL, Tormene AP, Vaegan (2010) ISCEV standard for clinical visual evoked potentials (2009 update). Doc Ophthalmol 120(1):111–119. doi:10.1007/s10633-009-9195-4 Pedrosa P, Machado D, Fiedler P, Alves E, Barradas NP, Haueisen J, Vaz F, Fonseca C (2015a) Electrochemical and structural characterization of nanocomposite Ag y:TiN x thin films for dry bioelectrodes: the effect of the N/Ti ratio and Ag content. Electrochim Acta 153:602–611. doi:10.1016/j.electacta.2014.12. 020 Pedrosa P, Fiedler P, Lopes C, Alves E, Barradas NP, Haueisen J, Vaz F, Fonseca C (2015b) Ag:TiN-coated polyurethane for dry biopotential electrodes: from polymer plasma activation to the first EEG measurements. Plasma Process Polym. Searle A, Kirkup L (2000) A direct comparison of wet, dry and insulating bioelectric recording electrodes. Physiol Meas 21(2):271–283 Steinisch M, Tana MG, Comani S (2013) A post-stroke rehabilitation system integrating robotics, VR and high-resolution EEG imaging. IEEE Trans Neural Syst Rehabil Eng 21(5):849–859. doi:10.1109/TNSRE.2013.2267851 Teplan M (2002) Fundamentals of EEG measurement. Meas Sci Technol 2(2):1–11 Thompson T, Steffert T, Ros T, Leach J, Gruzelier J (2008) EEG applications for sport and performance. Methods 45(4):279–288. doi:10.1016/j.ymeth.2008.07.006