Novel semi-dry electrodes for brain–computer interface applications

Fei Wang^1,2,5, Guangli Li^3,5, Jingjing Chen², Yanwen Duan^3,4 and Dan Zhang¹

¹ Department of Psychology, School of Social Sciences, Tsinghua University, Beijing 100084, People’s Republic of China

² Department of Biomedical Engineering, School of Medicine, Tsinghua University, Beijing 100084, People’s Republic of China

³ School of Chemistry and Molecular Sciences, Wuhan University, Hubei 430072, People’s Republic of China

⁴ Wuhan Greentek Pty., Hubei 430074, People’s Republic of China

E-mail: yduan@whu.edu.cn and dzhang@tsinghua.edu.cn

Received 7 December 2015, revised 30 April 2016

Accepted for publication 6 June 2016

Published 5 July 2016

Abstract

Objectives. Modern applications of brain–computer interfaces (BCIs) based on electroencephalography rely heavily on the so-called wet electrodes (e.g. Ag/AgCl electrodes) which require gel application and skin preparation to operate properly. Recently, alternative ‘dry’ electrodes have been developed to increase ease of use, but they often suffer from higher electrode– skin impedance and signal instability. In the current paper, we have proposed a novel porous ceramic-based ‘semi-dry’ electrode. The key feature of the semi-dry electrodes is that their tips can slowly and continuously release a tiny amount of electrolyte liquid to the scalp, which provides an ionic conducting path for detecting neural signals. Approach. The performance of the proposed electrode was evaluated by simultaneous recording of the wet and semi-dry electrodes pairs in five classical BCI paradigms: eyes open/closed, the motor imagery BCI, the P300 speller, the N200 speller and the steady-state visually evoked potential-based BCI. Main results. The grand-averaged temporal cross-correlation was 0.95 ± 0.07 across the subjects and the nine recording positions, and these cross-correlations were stable throughout the whole experimental protocol. In the spectral domain, the semi-dry/wet coherence was greater than 0.80 at all frequencies and greater than 0.90 at frequencies above 10 Hz, with the exception of a dip around 50 Hz (i.e. the powerline noise). More importantly, the BCI classification accuracies were also comparable between the two types of electrodes. Significance. Overall, these results indicate that the proposed semi-dry electrode can effectively capture the electrophysiological responses and is a feasible alternative to the conventional dry electrode in BCI applications.

Keywords:

electroencephalography (EEG), brain–computer interface (BCI), semi-dry electrode, event-related potentials, rhythmic brain activity