Wireless, broadband, and implantable neuroprostheses Borton, David Allenson creator 2012 xxxvii, 267 p. born digital Thesis (Ph.D. -- Brown University (2012) Nurmikko, Arto Director Donoghue, John Reader Tripathi, Anubhav Reader Darling, Eric Reader Denison, Timothy Reader Brown University. Biomedical Engineering sponsor theses This thesis highlights our advances across many disciplines developing a fully implantable, low power, broadband, and wireless neural recording platform for use in persons with neuromotor disease and disconnect. Two systems meeting this goal are developed. First, an externally powered (RF inductive coupling at 13.56MHz), low profile (3mm), and polymeric-packaged (polydimethylsiloxane) 16 and 100 channel neural interface technology using an innovative low power (9mW) optoelectronic semiconductor laser (VCSEL, 852nm) for transcutaneous data telemetry is described (EPM). The EPM consumes just 29mW and is shown through finite element modeling and experimental measurements to be safely powered and generate heat below FDA regulations (<2C above body temperature). Second, an implantable device incorporating an inductively charged (2MHz) lithium-ion battery and long-range (3 meters) RF data telemetry (3.2-3.8GHz OOK) into a hermetically sealed titanium casing consuming only 90.6mW (>6hour operation) is described (IPM). Both systems leverage a custom low-power amplifier ASIC (6mW, 3.6µVrms RTI noise) integrated with 510k-approved microelectrode neural sensor array, detecting signals from 0.1Hz-7.8kHz (broadband). The EPM and IPM are verified on benchtop and validated in vivo in a non-human primate animal model and a novel animal model in swine for device verification and validation. Broadband neural data was wirelessly collected from 16 animals (11 swine, 5 primates) totaling over 1600 implant days. In addition, a development effort was also initiated to bring neural signal processing and decoding electronics onto a wearable platform (BrainPhone). Neural data was recorded from an IPM and fed into the embedded platform where it was processed, validating the system architecture. Finally, we suggest from our recordings in two monkeys that brain states under standard anesthetic agents (isoflurane, propofol, and ketamine) are classifiable in a low dimensional representation of spiking rate. The addition of spiking information may provide a richer description of brain state under anesthesia than current methods and thus enhance the understanding of anesthetic mechanisms. These advancements represent a major milestone in implantable neuroprosthetic device evolution, and present a safe, information-rich, and untethered platform for clinical and neuroscientific research. Neuroengineering Brain-Machine Interface Neuroprosthetics NeurosciencesBrain-computer interfaces RPB 20121023 engEnglish10.7301/Z0CR5RNKIn CopyrightCollection is open for research.dissertations