Publication
Electrode Reduction for EEG-based Imagined Speech BCI Applications
Maurice Rekrut; Johannes Ihl; Tobias Jungbluth
In: 5th International Neuroergonomics Conference. Neuroergonomics Conference (NEC-2024), July 8-12, Bordeaux, France, Frontiers in Neuroergonomics, 2024.
Abstract
EEG-based imagined speech BCIs try to decode imagined speech from EEG activity. While several studies have proven the feasibility of this innovative BCI concept, current implementations usually rely on high-resolution EEG devices with 64+ channels. These setups make EEG-based imagined speech BCIs inconvenient for everyday use. To overcome this problem and to evaluate the necessity of such complex setups, we applied electrode reduction on three different imagined speech EEG datasets, to find a best suitable subset of electrodes. Our results show that the commonly used 64+ channel setups are oversized, and a comparable performance could be achieved with half of the electrodes for all three datasets. A common subset shared between subjects could however not be found.