Analog Lab: Electroencephalography-on-a-chip
We are developing integrated circuits and systems for sensing, processing and classifying scalp electroencephalography (EEG) signals. Small, ultra-low power and computationally-sophisticated chips are the key to practical brain-computer-interfaces (BCI).
(a) Current scalp EEG-based brain-computer interfaces consist of a headset, which records scalp EEG and transmits it to a laptop for analysis. (b) A BCI chip with millimeter dimensions can harvest about 1 µW of power. The analog components of the proposed chip are highlighted in the dashed box. Conventional instrumentation amplifiers for EEG consume several µWs of power. We are researching amplifier architectures that exploit nonlinear feedforward control to bring the power consumption down to sub-µW levels. A spike generator circuit and an asynchronous ADC together perform a feature-to-digital conversion. The spike generator consumes nWs of power, and the ADC power consumption is fairly negligible.