Regresar

A Fine Dry-Electrode Selection to Characterize Event-Related Potentials in the Context of BCI

Abstract:

A brain-computer interface (BCI) detects brain activity and converts it to external commands, facilitating the interaction with external devices. One way to implement a BCI is through event-related potentials (ERP), which are positive or negative voltage deflections detected by electroencephalography (EEG) through conductive electrodes. A very promising technology of dry electrodes has been used in recent years, which is much easier and faster to install; useful also for daily life applications. But the disadvantage is that its signal-to-noise ratio is lower compared to traditional wet electrodes technology. Thus, we hypothesized that an appropriate selection of dry electrodes allows the recovery of much more information than traditional standard electrodes and therefore improves the BCI performance. This work shows the importance of electrode selection to obtain a better detection of the ERPs of the EEG signal with a minimum number of electrodes in a personalized manner. To illustrate this problem, we designed a BCI experiment based on P300-ERPs with a dry electrodes wireless EEG system and we evaluated its performance with two electrode selection methodologies designed for this purpose in 12 subjects. The experimental analysis of this work shows that our electrode selection methodology allows the P300-ERPs to be detected with greater precision than a standard electrode set choice. Besides, this minimum electrode selection methodology allows dealing with the well-known problem of inter- and intrasubject variability of the EEG signal, thus customizing the optimal selection of electrodes for each individual. This work contributes to the design of more friendly BCIs through a reduction in the number of electrodes, thus promoting more precise, comfortable, and lightweight equipment for real-life BCI applications.