The hitoe^® (Toray Inc., Tokyo, Japan) electrode was prepared from electroconductive textile fabric which is made of nano-fiber yarn (fiber diameter 700 nm; polyester) coated with PEDOT-PSS polymer thermobonding composition (Fig. 1; 6–7, 8). For assessing the electric property of the electroconductive textile, various types of hitoe^® pad (4 cm × 4 cm, 4 cm × 8 cm and 10 cm × 2.8 cm) were cut from PEDOT-PSS-coated textile.

Two types of wearable electrodes were sewed (C3fit IN-Pulse^®, GOLDWIN Inc., Toyama, Japan): T shirts for men and brassiere for women (Fig. 1; 1 and 3). ECG electrodes made of hitoe^® patch were sewn to the reverse of the cloth (Fig. 1; 2 and 4). Both the types of wear were compressive to the skin for sticking snugly to the electrodes. The size of electrodes was 8 cm × 4 cm for men’s T shirts and 10 cm × 2.8 cm for women’s brassiere, where the distance between the center of each electrode is 20 cm (Fig. 1; 5). These embedded bipolar electrodes were placed on V5R and V5, which correspond to lead I [15]. An insulated electroconductive lead ribbon was made of coated silver wires which connected between a hitoe^® patch and a terminal of ECG device (Fig. 1; 6–14, 15).

DC resistance, AC impedance, offset voltage, internal noise, and polarization voltage for 5 s after a defibrillator discharge at 200 V were measured on hitoe^® pads in dry condition. DC resistance (sheet resistance), which represents horizontal electroconductivity of the material, was measured by four-point needle electro-conductive meter (Loresta-AX, MCP-T370, Mitsubishi Chemical Analytech, Japan). For AC impedance measurement, a pair of electrodes faced each other and layered over 200 g weight were connected to the bio-electrode impedance meter at a 10-Hz sine wave (Melon Technos, Japan). Offset voltage (twin electrodes were faced to each other) and defibrillator recovery potential were taken by digital multifunctional meter (DM2561A, NF Corporation, Japan). Two hundred voltage loaded to the defibrillator was applied to these electrodes according to the schematics and protocols of ANSI. Bio-signal waveform conformation, electrode’s combined offset instability, and internal noises were measured by polygraph (AP-1124, Miyuki-giken Japan) or neuro-physical recorder (MEB-5504 Nihon Koden Japan), respectively, at adequate frequency range (0.01–1000 Hz). These measurements are mandated for disposal electrodes by the American National Standards Institute (ANSI)/Association for the Advancement of Medical Instrumentation (AAMI) [16].

The electrical contact impedances to the subjects’ skin were measured simultaneously on textile (hitoe^® 4 cm × 8 cm two electrodes were fixed inside the underwear, Fig. 1) and conventional gel-type (round shape Ag–AgCl with electrolyte liquid gel, 25 mm in diameter, with adhesive pad 5 cm, Vitrode R, Nihon Koden, Japan) electrodes on 74 volunteers using a bio-electrode impedance meter at a 10-Hz sine wave (Melon Technos, Japan) (protocol 1). Each electrode pad was immersed in 500 μl of a 30% glycerol/0.9% NaCl solution to wet the surface 1 h before use, according to the instruction of the electrodes. Subject’s skin was pretreated with alcoholic wiper for conventional electrodes, but not for textile electrodes.

Four smart shirts with textile electrodes were washed repeatedly with (n = 2) or without (n = 2) a mesh laundry net according to the Japan Industrial Standard (JIS) protocol [17]. Briefly, one washing cycle consisted of 5 min of gentle machine washing at 30 °C (JIS #105) or slightly harder machine washing at 40 °C (JIS #103) using synthetic detergent, followed by two rounds of a machine rinse for 2 min each with clean water and drying at a temperature below 80 °C. This cycle was repeated more than 150 times. Between each machine wash cycle, a 5-min spin-dry procedure was interposed for dehydration. After each washing cycle, the sheet resistance of each pad was measured using the in-line four-point probe technique at dried condition.

Forty-seven men and 19 women enrolled to study the ECG signal quality during supine rest, sitting, and physical exercises (Table 1). None of the participants had any history of cardiovascular disease or skin disease. All subjects took their chest measurements to prepare a wearable device fitted tightly to each subject’s thorax. In five subjects, this smart shirt was still loose; therefore, safety pins were used to make the shirt fit more tightly to their thorax.

In ECG recordings using a Holter ECG recorder (Kenz Cardy 303 Pico+^®, SUZUKEN Co., Ltd., Nagoya, Japan) (Fig. 1; 6–13), we studied the textile electrodes first. On wearing the smart shirt, the electrode pad was wetted with a few drops of glycerol (C₃H₈O₃). Skin abrader, gel, and adhesive were not applied. ECG recordings were conducted sequentially in a supine position for 30 s, sitting at a comfortable position for 30 s, a trunk rotation (the left and right shoulders are alternately protruded forwardly, just like twisting movement) in a sitting position on the chair for 60 s, and stepping for 90 s. The timing of each twisting or stepping exercise was adjusted to metronomic sounds (1.0 Hz). At the end of the first protocol, the subjects’ skin was observed for any signs of irritation.

Ten minutes after the above protocol, each volunteer’s skin was cleaned with 0.2% chlorhexidine gluconate and abraded gently with a skin preprocessor; then, conventional gel-type electrodes (Cardyrode-P^®, SUZUKEN) were applied and fixed firmly with gel and adhesive on modified lead I. ECG recordings were repeated on the same four conditions described above. At the end of the protocol, we again observed skin conditions at the sites, where the electrodes were attached.

The recorded whole ECG signals were printed out using the Holter ECG analyzer (Kenz Cardy Analyzer 05^®, SUZUKEN Co., Ltd., Nagoya, Japan) in 66 volunteers (Protocol 1) to determine the shapes of P, QRS, and T waves and to identify artifacts. Tiny spikes, notches, large baseline swings, a widened isoelectric line, loss of ECG signal, and electromagnetic interference (EMI) [17] were defined as artifacts.

ECG waveforms of P, ORS, and T for each subject were reviewed by two of three certified electrophysiologists (board-certified members of the Japanese Heart Rhythm Society: K.Y., Y.H., and H.M.), and the ECG signal quality was scored in each protocol from 1 to 3 points. Point “3” (a top quality) was assigned for stable recordings lacking motion artifacts, Point “2” for a satisfactory recording with a motion artifact, and point “1” for a poor recording such as with any loss of the ECG waveforms due to a large shift in the isoelectric level and electromyogram (EMG). The scores of the two electrophysiologists were added, and the patients’ ECG quality was categorized as ‘excellent’ for 6 points, ‘good' for 4–5 points, and ‘bad or poor' for 2–3 points. From the consensus of above three cardiologists, ‘excellent’ and ‘good’ are defined as clinically acceptable recordings.

The ratio of the QRS voltage to a baseline shift was quantified and termed the signal voltage- (QRS voltage)-to-artifact voltage ratio or the extent of the drift level of the isoelectric line (SAR). In the present study, the SAR was calculated using a modification of Meziane’s method (Fig. 2) [9].

All data are expressed as the means ± SD. The data management and statistical analyzes were performed using JMP^® (Version 9, SAS Institute Inc.). The differences in average impedances between electrode types were assessed by a two-tailed Student’s t test. We used Wilcoxon’s signed rank test for comparison of the SAR and the Chi-square test for evaluation of the ECG quality. A p value <0.05 was considered significant.

The electrode characteristics of the present textile electrodes were accorded well with the ANSI/AAMI guidelines (Table 2). The electrode-skin combined impedances (Protocol 1) were 135 ± 35, and 219 ± 17 kΩ in the textile and conventional electrodes (n = 74), respectively (p < 0.01).

A sharp increase in the electrical resistance of the pads was noted during repeated washings under the slightly hard condition (JIS, #103) without the use of a mesh laundry bag (Fig. 3). Gentle machine-washing (JIS, #105) using a mesh laundry bag was better for maintaining electrical conductivity of the textile shirt than washing without the use of a bag. Up to 50 machine-washing cycles with the use of a mesh laundry bag, electrical conductivity was maintained.

As shown in the representative tracings (Fig. 4), P, QRS, and T waves were comparable between the textile and conventional electrodes. No heart beats were missing from any of the ECG recordings of either the textile or conventional electrodes under the supine, sitting, or stepping condition. SAR was better in conventional than textile electrodes and was worst in twist among four situations (Table 3). The quality scores of each waveform during the stepping exercise were excellent in more than 88% of the subjects in either textile or conventional electrode types (Fig. 5).

High-frequency noise and a baseline shift were noted during twist movements in both the textile and conventional electrodes. The P wave was clearly detectable in supine, sitting, and stepping conditions, as represented in Fig. 4b. In twisting condition, P waves were detected in 17% of cases and others were lost in artifacts.

SAR was better in men than women in either textile or conventional electrodes and was comparable in supine, sitting, or stepping (Table 4). The twisting motion caused the most deleterious effect on the ECG waveform among four conditions in both textile and conventional electrodes in either men or women. Approximately two-thirds of the recordings were categorized as ‘good’ in the textile group, and the proportions of poor recordings (i.e., QRS and T waves that were not recognizable) were 19% and 33% for conventional and textile electrodes, respectively. Noise was detected in 41% of the recordings of men during twisting with textile electrodes, while that was 6% with conventional electrodes (Fig. 4).