A Stable-Contact-Oriented In-Ear EEG Sensor With Contact-Mechanics-Guided Electrode Placement
Abstract
Long-term ear electroencephalography (Ear-EEG) recording in naturalistic settings requires stable sensor-ear-canal contact. This study proposes a stable-contact-oriented in-Ear-EEG sensor that combines a flexible multiflange substrate, a fin-like stabilizer, and contact-mechanics-guided electrode placement. Impedance comparison in 30 participants verified the effectiveness of the simulation-guided site selection. In a 4-h continuous-wear experiment with 20 participants, contact impedance showed no upward drift, recovery-stage root mean square (rms) was close to the baseline level after motion perturbation, with no significant recovery-versus-baseline difference detected (p = 0.622), and no significant Rest-1/Rest-2 difference was detected in resting-state coefficient of variation (CV) after prolonged wear (left ear: p = 0.416; right ear: p = 0.936). Functional experiments showed that the sensor could record eyes-open/eyes-closed alpha modulation and decodable four-target steady-state visual evoked potential (SSVEP) responses. Using filter-bank task-discriminant component analysis (FB_TDCA), Ear-EEG achieved an accuracy of ( $68.4~\pm ~9.0$ ) % and an information transfer rate (ITR) of ( $18.6~\pm ~7.3$ ) bits/min, significantly outperforming the temporal scalp channels T7/T8. These results support the stable-contact performance and application-level usability of the proposed sensor during prolonged wear.