|
Inherent variation amongst human brains causes difficulty in implementing electroencephalography (EEG) into a universal brain-machine interface (BMI). Existing EEG systems suffer from an inconsistent signal quality, burdensome preparation time, and discomfort caused by rigid wires and metal electrodes in a hair cap. Additionally, leading classification methods require training on a per-subject or per-session basis. Although recent machine learning techniques offer a simpler EEG arrangement with fewer electrodes, these EEG devices still involve intrusive and bulky headgear, equipped with separate non-portable electrical hardware.
Here, we introduce a fully portable, wireless, flexible scalp electronics on a soft elastomeric membrane, representing a comfortable and ergonomic wearable BMI. These imperceptible soft electronics incorporate an ultrathin nanomembrane electrode on non-hair-bearing skin, flexible conductive electrodes on the hair-bearing scalp, and low-profile, skin-conformal electronics on the neck for fully portable, wireless data acquisition. Analytical and computational studies establish the fundamental design criteria of the flexible, skin-like hybrid electronics (SKINTRONICS), enabling seamless, portable EEG recording with significantly enhanced signal quality over commercial systems. Newly designed time-domain analysis with deep convolutional neural networks allows real-time, highly accurate classification of steady-state visually evoked potentials from only two channels. This portable scalp system with six human participants achieves a high accuracy (94.54 ± 0.90%) for a corresponding information transfer rate of 122.1 ± 3.53 bits per minute. In vivo study of the fully portable BMI, enabled by the SKINTRONICS and deep-learning algorithm, shows precise, low-latency control of a wireless wheelchair, motorized vehicle, and keyboard-less presentation via two-channel EEG, which demonstrates potential implications for wide range applications of the new class of EEG-based universal BMI technologies.
|
