Identification of Attention State for Menu-Selection using In-Ear EEG Recording

Abstract

Conventional EEG devices have limitations for the use of Brain-Computer Interface (BCI) because they are uncomfortable to wear in daily life. Since most smartphone users use earphones, a novel Earphone-shaped EEG device, which measures EEG signals in the ear canal while maintaining functions of the earphone, can be powerful tools for BCI. In this report, the attention state recorded from in-ear EEG was discriminated from the resting state to use simple application of one-button menu selection. Power spectral densities (PSD) in eye-closed state, eye-open state, and attention state were compared using autoregressive (AR) Burg method. Using selected features from Fisher ratio, attention state was successfully classified from resting state with support vector machine (SVM). Based on this study, prototypes for stable recording and sound delivery are developing and real-time BCI application using earphone-shaped EEG device will be researched.