Tim Denison

A 2

This paper describes a prototype micropower instrumentation amplifier intended for chronic sensing of neural field potentials (NFPs). NFPs represent the ensemble activity of thousands of neurons and code-useful information for both normal activity and disease states. NFPs are small - of the order of tens of muV- and reside at low bandwidths that make them susceptible to excess noise. Therefore, to ensure the highest fidelity of signal measurement for diagnostic analysis, the amplifier is chopper-stabilized to eliminate 1/f and popcorn noise. The circuit was prototyped in an 0.8 mum CMOS process and consumes under 2.0 muW from a 1.8 V supply. A noise floor of 0.98 muVrms was achieved over a bandwidth from 0.05 to 100 Hz; the noise-efficiency factor of 4.6 is one of the lowest published to date. A flexible on-chip high-pass filter is used to suppress front-end electrode offsets while maintaining relevant physiological data. The monolithic architect and micropower low-noise low-supply operation could help enable applications ranging from neuroprosthetics to seizure monitors that require a small form factor and battery operation. Although the focus of this paper is on neurophysiological sensing, the circuit architecture can be applied generally to micropower sensor interfaces that benefit from chopper stabilization.

\mu\hbox{W}

This paper describes an amplification and spectral processing IC for extracting key bioelectrical signals, or ldquobiomarkersrdquo, which are expressed in the brains field potentials. The intent is to explore using these biomarkers to drive prosthetic actuators or titrate therapy devices such as a deep-brain neurostimulator. The prototype IC uses 5 muW/channel to resolve signals on the order of 1 muVrms. The four channels on the device provide independent spectral analysis from DC to 1 kHz, with variable bandwidth and power filtering characteristics. The noise floor and flexible spectral processing support a broad range of potential applications including sleep staging, Parkinsons disease, detection of movement intention for neuroprosthesis, and detection of high frequency ldquofast ripplesrdquo for exploring seizure prediction. To fully demonstrate the ICs functionality, we include results from a prototype ldquoclosed-looprdquo neurostimulator implementing adaptive titration of therapy based on measured field potential activity.

100 nV/rtHz Chopper-Stabilized Instrumentation Amplifier for Chronic Measurement of Neural Field Potentials

An implantable bi-directional brain-machine interface (BMI) prototype is presented. With sensing, algorithm, wireless telemetry, and stimulation therapy capabilities, the system is designed for chronic studies exploring closed-loop and diagnostic opportunities for neuroprosthetics. In particular, we hope to enable fundamental chronic research into the physiology of neurological disorders, define key electrical biomarkers related to disease, and apply this learning to patient-specific algorithms for therapeutic stimulation and diagnostics. The ultimate goal is to provide practical neuroprosthetics with adaptive therapy for improved efficiency and efficacy.

A 5

An electric-field sensor is presented for applications such as xerography. The sensor architecture combines a vibrating MEMS structure with synchronous detection-based electronics. Prototyped in a MEMS process, the noise floor is 4.0V/m/radicHz and the INL is 20V/m over a range of +/-700kV/m, an order-of-magnitude improvement over existing MEMS devices

\mu

Modulation of neural activity through electrical stimulation of tissue is an effective therapy for neurological diseases such as Parkinsons disease and essential tremor. Researchers are exploring improving therapy through adjustment of stimulation parameters based upon sensed data. This requires classifiers to extract features and estimate patient state. It also requires algorithms to appropriately map the state estimation to stimulation parameters. The latter, known as the control policy algorithm, is the focus of this work. Because the optimal control policy algorithms for the nervous system are not fully characterized at this time, we have implemented a generic control policy framework to facilitate exploratory research and rapid prototyping of new neuromodulation strategies.

W/Channel Spectral Analysis IC for Chronic Bidirectional Brain–Machine Interfaces

A prototype bi-directional neural interface system is presented with closed-loop and embedded DSP capabilities. The system includes 32-electrode stimulation capability, eight multiplexed low-noise low-power bio-potential sensing channels with on-chip digital FFT, and a Cortex M3-based microcontroller for implementing closed-loop algorithms. The stimulation subsystem can provide a maximum stimulation current of 12mA with approximately 3% accuracy per channel. The sensing subsystem uses a fully differential chopping amplifier achieving a noise floor <;100nV/rtHz with approximately 1μA current from 2V supply. A 13bit continuous-time sigma-delta ADC is used to sample the amplifier output at the rate of 33kHz. The ADC consumes approximately 100nA while achieving an ENOB of 12b with a 250Hz date rate. A hardware implementation of a cached-FFT is included in the sensing IC to perform spectral analysis of the bio-potential signal. The resulting multiple spectral signatures can be selectively sent to the microcontroller allowing for algorithmic control of the stimulator system. The system partition was designed to minimize overall computational power while providing user flexibility. The sensing performance of the prototype has been demonstrated with 2-D cursor control in a non-human primate brain machine interface (BMI) model using 20 features simultaneously.

An implantable Bi-directional brain-machine interface system for chronic neuroprosthesis research

This paper describes a prototype architecture that tracks the power fluctuations in discrete frequency bands for a broad spectrum of neuronal biomarkers. The circuit merges chopper-stabilization with heterodyne signal processing to construct a low-noise amplifier with highly programmable, robust filtering characteristics. It is concluded that in our application, the area and modest noise penalty are offset by the benefit of a well-partitioned analog-to-digital boundary flexible enough to potentially extract a wide spectrum of biomarkers associated with disease.

A Self-Resonant MEMS-based Electrostatic Field Sensor with 4V/m/Hz Sensitivity

Implantable medical devices can provide chronic access to the nervous system. Implants containing embedded scientific instrumentation payloads (e.g. - sensors, classification, and control policy implementation) provide a unique opportunity for exploring diseased neural networks and how these neural networks may be better treated. Physically embedding payloads in an implant creates intertwined constraints such as power consumption, algorithmic computation limits and lack of flexibility, data storage, and the scale of sensing information. These limitations can be largely addressed with a combination of rechargeable batteries and high-bandwidth, secure, distance telemetry, which enables a distributed neural research system. Taking advantage of a distributed architecture helps facilitate scientific investigation in a more unconstrained environment. In this paper, we describe the design of an implantable research tool, discuss the prototype system architecture and its design details, and present preliminary bench verification and validation with human data drawn from representative use cases.

A flexible algorithm framework for closed-loop neuromodulation research systems

This paper describes the preliminary technology validation of a bi-directional neural interface in an awake large animal model (ovine). The device addresses the major requirements of a chronic research system, including operation within the implantable environment and electrical stimulation with concurrent bioelectric sensing. Preliminary chronic measurements of network dynamics demonstrate that a chronically stable bi-directional interface to the nervous system is achievable. This was shown through chronic impedance and evoked potential measurements in the thalamo-cortical circuit of Papez. Characterization of bioelectric sensing in the presence of stimulation was also performed through measurements of the noise floor in the presence and absence of stimulation. Further technology validation was performed by using the prototype to correlate activity within and between structures in the circuit of Papez in the presence and absence of stimulation.

A 32-channel modular bi-directional neural interface system with embedded DSP for closed-loop operation

Architectures capable of using an algorithm to modify actuation based on measured signals are often called “closed-loop” systems. While such systems are traditionally thought to rely on algorithms residing in device firmware, these may also reside outside the device in a host processor located physically nearby, or on a cloud-based architecture. In order to serve the potentially broad array of data processing modalities, we have developed an application programming interface (API). The API enables access to the sensing and stimulation capabilities of an implantable bi-directional neural interface. Systems using the API on different hardware/software platforms could measure neural signals, process signals in realtime, and modulate stimulation parameters using a variety of algorithms. This flexibility allows increased algorithm access and enables rapid prototyping for potentially improved technology solutions. The system performance was characterized using a signal generator to input square wave pulses to a Simulink model via the API. Closed-loop stimulation latencies of around 600ms were achieved.