In recent decades, wearable health monitors have grown from crude heart rate sensors toall-in-one devices that can track steps, motion, location, arrhythmia, electrocardiogram,
blood oximetry, and calorie usage. More recent work has focused on developing wearables
that provide non-invasive neural recording (electroencephalography - EEG) for focus, stress,
and drowsiness monitoring. As wrist-worn wearables run out of features to add, head/ear
worn EEG wearables offer new ways to provide users with helpful and actionable wellness
information. The problem blocking widespread neural wearables is that existing devices
are bulky, uncomfortable, require single-use wet electrodes, and often require training in
everyday scenarios.
This thesis details an end-to-end design process for low-profile, dry-electrode neural recordinghearables that can record EEG inside the ear. Starting from the EEG signal basics, this
work will walk through the modeling, design, and verification of the three parts of an Ear
EEG system: the electrodes, the neural recording system, and the downstream processing
and classification software. Particular focus is placed on maximizing user comfort and the
ease of manufacture without hurting system performance.
All of these topics will be based on constructing a practical in-ear EEG device based on mul-tiple dry electrodes, a user-generic design, and a lightweight wireless interface for streaming
data and device programming. Different earpiece manufacturing processes will be showcased
for prototyping (using 3D printing and electroless plating) and production at scale (using
vacuum forming and spray coating). The performance of this system will be evaluated with
human subject trials that recorded spontaneous and evoked physiological signals, eye-blinks,
alpha rhythm, auditory steady-state response (ASSR), and drowsiness detection.
