Validation of melomind(TM) signal quality: a proof of concept resting-state and ERPs study

Abstract

Wearable EEG systems have become accessible to researchers and clinicians over the last decade, thus requiring neurotechnology companies to seek for outstanding EEG signal quality. Here, we show that the melomind headset equipped with dry electrodes (myBrain Technologies, Paris, France) allows the recording of reliable electro-cortical dynamics as compared to a wet-based standard-EEG system (actiCAP, Brain Products GmbH, Gilching, Germany). EEGs were acquired simultaneously from the two systems while thirteen subjects underwent a visual oddball, a steady-state visually-evoked potentials (SSVEPs) and two resting-state (RS) tasks. RS were acquired with eyes-closed and eyes-open (2 minutes each) and repeated twice (before and after the cognitive tasks). During the oddball task, participants responded on a gamepad when a target-stimulus was displayed. In the SSVEPs, visual responses were elicited at 15 and 20 Hz through a series of 15-second stimuli presented 5 times each. The power of theta- [4-8 Hz], alpha- [8-13 Hz], and beta- [13-30 Hz] band was extracted from the two RS. The signal-to-noise-ratio in the 15 ({+/-} 1) and 20 ({+/-} 1) Hz range was computed from the SSVEPs. The shape of the N2/P300 complex was analysed from the oddball task. Strong correlations resulted between the parameters obtained from the two EEG systems (0.53 < Pearsons r < 0.97). Bland and Altman analysis revealed small dissimilarities between the two systems, with values laying in the 95% confidence interval in all the tasks. Our results demonstrate that the melomind is an affordable solution to reliably assess humans electro-cortical dynamics at-rest and during cognitive tasks, thus paving the way to its use in neuroscience studies and brain-computer interfaces.