Posted on No Comments on DEMONSTRATION OF CLOSED-LOOP STIMULATION PLATFORM WITH HARDWARE ACCELERATED ARTIFACT REMOVAL

DEMONSTRATION OF CLOSED-LOOP STIMULATION PLATFORM WITH HARDWARE ACCELERATED ARTIFACT REMOVAL

The lab has developed and validated a closed-loop stimulation platform capable of updating the stimulation montage at intervals as brief as 10 milliseconds, accompanied by a latency of only 20 milliseconds. However, the frequent updates to the montage generate biomimetic and varying electrical artifacts, thereby obscuring the neural response. Our research team has designed a proprietary integrated chip that leverages the spatial correlation inherent in the artifacts to facilitate hardware-accelerated removal. It enables uninterrupted, real-time monitoring of neural responses during active stimulation.

The lab member is validating the recording system’s capability to record emulated neural signals concurrently with biomimetic stimulation, specifically dynamic stimulation. The upper trace displays on the screen represents the raw input data, which includes stimulation artifacts and an oscillatory signal representing the underlying neural response. The bottom trace displays the neural response extracted by the proprietary recording system.

Posted on

Demonstration of Closed-loop Operation with Our Implantable Device

 

The blue and pink traces on the oscilloscope represents two of the many channels that are generating stimulation pulses, initially at 30Hz and 60Hz at different phase delays respectively. The green trace represents ADC recording that is wirelessly transmitted to a laptop running the closed-loop controller in real-time. Toggling the device’s analog recording sensor, the closed-loop controller on the laptop detects the change and modulates the stimulation protocols, communicating this back to the SoC chip wirelessly. The last portion of the video shows the new stimulation patterns of the blue and pink traces, at 30Hz and 20Hz, and phase delays respectively. In summary, we have demonstrated our system’s capability to simultaneously perform real-time recording and stimulation. Accordingly, it is feasible to implement closed-loop operations using our implantable system. A sophisticated controller is under development.

Click to view the Demo

The figure displays a demonstration of our closed-loop system performing real-time, multi-channel recording and stimulation. The green and yellow traces are EMG signals fed into the recording sensors of the chip. The information is transmitted wirelessly as input to a control algorithm that detects EMG envelopes over the chip’s multiple recording channels. The decision algorithm modulates the chips stimulation parameters to 60Hz (blue trace) and 20Hz (pink trace) when a large envelope is detected in the yellow EMG signal; and 20Hz (blue trace) and 30Hz (pink trace) when a large envelope is detected in the green EMG signal.

 

Posted on

Brain to Brain Communication

Our Brain-to-Brain communication research focuses on advancing the precision of non-invasive recording and stimulation of the brain cortex. The precision is increased by the development of advanced optimization algorithms combined with accurate tissue modeling. The models are augmented by our development of hardware stimulation and recording systems to allow independent fine-tuned stimulation of large amount of channels and fast data transmission between multiple brain-interface devices in real-time. Our bold aim is to step beyond a brain-computer interface and instead make human brain to brain connection possible, thus making scientific fiction a reality.

 

Posted on

Stimulation Artifact Cancellation

2016-07-01_15_04_54_02sec-pageLive Stimulation Artifact Cancellation

 

Recovering neural responses from electrode recordings is fundamental for understanding the dynamics of neural networks. This effort is often obscured by stimulus artifacts in the recordings, which result from stimuli injected into the electrode-tissue interface. Stimulus artifacts, which can be orders of magnitude larger than the neural responses of interest, can mask short-latency evoked responses. Furthermore, simultaneous neural stimulation and recording on the same electrode generates artifacts with larger amplitudes compared to a separate electrode setup, which inevitably overwhelm the amplifier operation and cause unrecoverable neural signal loss. This paper proposes an end-to-end system combining hardware and software techniques for actively cancelling stimulus artifacts, avoiding amplifier saturation, and recovering neural responses during current-controlled in-vivo neural stimulation and recording. The proposed system is tested in-vitro under various stimulation settings by stimulating and recording on the same electrode with a superimposed pre-recorded neural signal. Experimental results show that neural responses can be recovered with minimal distortion even during stimulus artifacts that are several orders greater in magnitude.

 

Artifact Cancellation fig1_systemOverview copy
System-Level Overview

 

Publication

“A Hybrid Hardware and Software Approach for Cancelling Stimulus Artifacts During Same-electrode Neural Stimulation and Recording,” Stanislav Culaclii, Brian Kim, Yi-Kai Lo, and Wentai Liu, accepted by EMBC 2016.

Posted on

Neural Signal Processing and Telemetry

Recording from a large number of neurons produces vast quantities of data that is highly difficult to extract and interpret due to noise and aggregation of multiple neural signals in a single recording site. In addition to conventional techniques like spike sorting, component analysis, de-noising detection, we are focusing on the cellular and molecular levels to understand the fundamentals of neural signals and behavior of large group of neurons. We found that the traditional deterministic channel models are insufficient for the description of the activities of real neurons. We are currently developing the stochastic kinetic modeling of sodium channel and its validation with the measurement of gate current due to transmembrane protein movements in the order of several picoamps with the bandwidth of several hundred of MHz.

 

Selected Publications

  1. “A 12-channel 6mW Wireless Neural Recording IC with On-the-fly Spike Sorting and UWB Transmitter,” M. Chae, W. Liu, Z. Yang, T. Chen, J. Kim, M. Sivaprakasam, and M. R. Yuce, To appear in International Solid-State Circuits Conference, February 2008.
  2. “An Integrated Multi-Channel Neural Recording System,” M. Chae, W. Liu, and M. Sivaprakasam, BMES Annual Fall Meeting, September 2007.

 

 

Collaborators

  1. Huntington Medical Research Institutes
  2. Arizona State University
Posted on

Non-Invasive Functional Magnetic Stimulation

Neural stimulation is commonly accomplished by a voltage or current pulse through a microelectrode. Ideally, a method would exist which inherently had zero net charge transfer, required only simple driver circuitry and was completely isolated from the tissue to reduce circuit failure due to corrosion and fouling by protein deposition. Magnetic stimulation achieves these goals. The presence of scar tissue or deposited proteins is irrelevant because the magnetic permeability of tissue is near that of free space. Excitation arises from the magnetic field which generates a current across the membrane of the cell which changes the resting potential of the neuron and triggers an action potential (the fundamental signal generation mechanism neurons employ). We are currently studying the fundamental mechanisms of magnetic stimulation, developing models and verifying through in vitro experiments.
Magstim

Magrecording

 

 

Selected Publications

  1. “Functional Magnetic Stimulation for Epiretinal Prosthesis”, E. Basham, W. Liu, and M. Sivaprakasam, Abstract B254, Association of Research in Vision and Ophthalmology Annual Meeting, May 2005.

 

 

Collaborators

  1. Santa Clara University
Posted on

Microstimulation Platform for Neural Systems

Functional Electrical Stimulation of biological tissue has a wide range of applications ranging from pain relief to neural prostheses. Flexibility, small size and low power operation, safety are key requirements in microstimulation systems. Custom integrated circuits for microstimulation face the challenge of having to support relatively high stimulation voltages for the current CMOS technology, while still needing maintain low power operation and achieve a high degree of miniaturization. In addition, the experimental nature of the evolving microstimulation applications demands a high degree of flexibility and versatility. We are working on several microstimulation applications through our collaborators that have a varying degree of requirements.
Stimchip

Wearablestim

Selected Publications

  1. “Challenges for Integrated Circuits in Implantable Devices,” W. Liu and M. Sivaprakasam, Volume: 20, Future Fab International, January 2006.
  2. “Microelectronics Design for Implantable Wireless Biomimetic Microelectronic Systems,” W. Liu, M. Sivaprakasam, G. Wang, M. Zhou, J. Granacki, J. LaCoss, and J. Wills, Volume: 24, Pages: 66 – 74, IEEE Engineering in Medicine and Biology Magazine, September 2005.
  3. “A Variable Range Bi-Phasic Current Stimulus Driver Circuitry for an Implantable Retinal Prosthetic Device,” M. Sivaprakasam, W. Liu, M. S. Humayun, and J. D. Weiland, Volume: 41, Pages: 763 – 771, IEEE Journal of Solid State Circuits, March 2005.

Collaborators

  1. University of Southern California
  2. Long Beach Veteran Affairs
  3. Stanford University
  4. Huntington Medical Research Institutes
Posted on

Low Power Integrated Neural Recording System

Advances in micro electrode arrays (MEA) have enabled neuroscientists and researchers in biomedical engineering to take advantage of a large number of channels, and this has made it possible to pursue a variety of neuroprosthetic applications such as treating spinal cord injuries, deep brain stimulation to treat Parkinson’s disease. This also requires neural recording electronics that are small enough to be integrated close to the MEA and consume less power so that they can be safely placed close to the tissue. Though several efforts have been made by the research community to minimize the power and area of each individual circuit block, almost no attention was paid to find the trade-offs among those circuit blocks to achieve an optimal design. We are currently developing optimal design methodologies for integrated neural recording systems that allow the optimal design for a given set of specifications and constraints. These are verified by custom chip fabrication and verification. Wireless telemetry is essential for recording neural signals from the subject in its natural environment. We are currently developing wireless transceivers for short range telemetry for neural signals. Power and area are the key parameters for the transceivers to be implantable. We are investigating different architectures and the associated tradeoffs between computation and communication.
Neuralrecording2

 

 

Selected Publications

  1. “A 12-channel 6mW Wireless Neural Recording IC with On-the-fly Spike Sorting and UWB Transmitter,” M. Chae, W. Liu, Z. Yang, T. Chen, J. Kim, M. Sivaprakasam, and M. R. Yuce, International Solid-State Circuits Conference, February 2008.
  2. “An Integrated Multi-Channel Neural Recording System,” M. Chae, W. Liu, and M. Sivaprakasam, BMES Annual Fall Meeting, September 2007.
  3. “A Wideband Telemetry Unit for Multi-Channel Neural Recording,” M. R. Yuce, W. Liu, M. Chae, and J. Kim, IEEE International Conference on Ultra-Wideband, September 2007.

 

 

Collaborators

  1. Huntington Medical Research Institutes
  2. Arizona State University