Dissertation

VLSI Circuits for Bidirectional Neural Interfaces

Greenwald, Elliot

2016

Overview

Medical devices that deliver electrical stimulation to neural tissue are important clinical tools that can augment or replace pharmacological therapies. The success of such devices has led to an explosion of interest in the field, termed neuromodulation, with a diverse set of disorders being targeted for device-based treatment. Nevertheless, a large degree of uncertainty surrounds how and why these devices are effective. This uncertainty limits the ability to optimize therapy and gives rise to deleterious side effects. An emerging approach to improve neuromodulation efficacy and to better understand its mechanisms is to record bioelectric activity during stimulation. Understanding how stimulation affects electrophysiology can provide insights into disease, and also provides a feedback signal to autonomously tune stimulation parameters to improve efficacy or decrease side-effects. The aims of this work were taken up to advance the state-of-the-art in neuro-interface technology to enable closed-loop neuromodulation therapies. Long term monitoring of neuronal activity in awake and behaving subjects can provide critical insights into brain dynamics that can inform system-level design of closed-loop neuromodulation systems. Thus, first we designed a system that wirelessly telemetered electrocorticography signals from awake-behaving rats. We hypothesized that such a system could be useful for detecting sporadic but clinically relevant electrophysiological events. In an 18-hour, overnight recording, seizure activity was detected in a pre-clinical rodent model of global ischemic brain injury. We subsequently turned to the design of neurostimulation circuits. Three critical features of neurostimulation devices are safety, programmability, and specificity. We conceived and implemented a neurostimulator architecture that utilizes a compact on-chip circuit for charge balancing (safety), digital-to-analog converter calibration (programmability) and current steering (specificity). Charge balancing accuracy was measured at better than 0.3%, the digital-to-analog converters achieved 8-bit resolution, and physiological effects of current steering stimulation were demonstrated in an anesthetized rat. Lastly, to implement a bidirectional neural interface, both the recording and stimulation circuits were fabricated on a single chip. In doing so, we implemented a low noise, ultra-low power recording front end with a high dynamic range. The recording circuits achieved a signal-to-noise ratio of 58 dB and a spurious-free dynamic range of better than 70 dB, while consuming 5.5 μW per channel. We demonstrated bidirectional operation of the chip by recording cardiac modulation induced through vagus nerve stimulation, and demonstrated closed-loop control of cardiac rhythm.

• Publisher: ProQuest Dissertations & Theses
• Subject: Biomedical engineering / Electrical engineering / Health sciences / Neurosciences / Physiology
• ISBN: 1687987661, 9781687987662