UC Irvine

Restoring the ability to walk in individuals with chronic spinal cord injury is an ambitious goal that has been rigorously pursued in recent years. While there have been a number of attempts to enable walking in those patients, no universally accepted biomedical solution exists to address this grand challenge to this day. Brain-machine interfaces (BMIs) are one of the promising platforms to restore motor and sensory functions in people with paraplegia. However, fully-implantable BMIs must address a number of critical issues to become clinically viable, in particular excessive overall power dissipation and inadequate interference resilience in neural recording.

In this dissertation, an energy-efficient electrocorticography (ECoG) array architecture for fully-implantable BMIs is presented. A novel dual-mode analog signal processing method is introduced that extracts useful neural features from high-γ band (80-160 Hz) at the early stages of signal acquisition. This approach utilizes a distinct optimized signal pathway based on power envelope extraction to achieve significant power savings for digitization and processing. A prototype incorporating a 32-channel ultra-low power signal acquisition front-end was fabricated in 180nm CMOS process and successfully tested in-vitro and in-vivo.

Next, an ultra-low power mixed-signal neural data acquisition system is presented. The dual-mode data acquisition system enables a novel low-power hybrid-domain neural decoding architecture for implantable BMIs with high channel count. The fully-integrated custom chip implemented in 180nm CMOS process achieves excellent performance with significant back-end power-saving advantage compared to prior works. The fabricated prototype was further evaluated with in-vivo human tests at bedside, in addition to electrical characterization.

Finally, common-mode interference phenomenon in multi-channel biosignal recording systems employing a shared-reference scheme is studied. While it is well-understood that a shared-reference scheme causes impedance mismatch at the input terminals of bioamplifier, and thus limits the maximum achievable common-mode rejection ratio (CMRR), a theoretical study that can provide quantitative assessment of this source of degradation is still lacking. This section provides an equivalent electrical circuit model of the input interface consisting of an electrode array and bioamplifiers, followed by a complete analysis to formulate the CMRR degradation.