This proposal describes a novel approach for recording from tens of thousands of neurons simultaneously using bundles of microwires that are each smaller than a human hair. To accomplish this, thousands of microwires are bundled together, with one end being inserted into the brain and the other being bonded to a high-speed image sensor chip that reads out the voltages on each wire. The sensor output can then be acquired to a PC using standard frame-grabber technology. This approach was developed as part of an international collaboration between the research laboratories of Andreas Schaefer (UCL, UK) and Nicholas Melosh (Stanford University, CA). The company Paradromics was created specifically to develop the academic prototype system into a robust commercial product. Significance - Neural circuits are composed of many thousands of neurons, and information is encoded within these networks by the relative timing of action potentials (1-4). In order to understand how neural circuits behave and how they impact human health and behavior, new techniques are necessary that can resolve the individual spiking activity of many thousands of neurons simultaneously (2, 5, 6). Here we propose a technology that is not only capable of sampling tens of thousands of channels, but can do so in vivo, at depths where non-invasive optical methods are infeasible. Innovation - Our approach is innovative because it combines two highly scalable technologies for the first time ever: high resolution CMOS camera sensors and microwire bundles (7). Our prototype device uses bundles of tens of thousands of wires, and a 640x512 sensor chip (8), but in principle we can scale up both bundles and sensors to millions of channels without any architectural changes. Approach - Here we propose to turn an exciting new laboratory prototype into a turnkey research tool for neuroscience labs around the world. To accomplish this, we propose a series of simple but important improvements to our existing prototype system that will enable us to increase the number of available recording channels by drastically improving bundle-sensor connectivity. We also propose to develop data acquisition software that will make using the system as simple and intuitive as possible. Market - We estimate there are approximately 5,000 active multi-channel recording systems, each with an average less than 100 channels. Our first product will have > 300,000 channels. Two of our systems will therefore be capable of recording from more neurons simultaneously than all of the other multi- channel recording systems in the world combined. Further, because we are harnessing already- developed CMOS technology, we think that we can provide these systems for a price comparable to existing multi-channel recording systems from major suppliers.

Public Health Relevance

Circuits in the brain involve many thousands of neurons working together in complex networks to accomplish tasks. Current techniques for recording in vivo brain activity can't resolve the individual activities of this many neurons, so many fundamental questions about how the brain works remain unanswered. Here we describe a new method for large-scale electrical recordings in the brain that allows for measuring the activity of unprecedented tens of thousands of neurons simultaneously. With the proposed funding, we will develop our existing prototype into a robust product that can be broadly disseminated and incorporated into regular neuroscience practice.

Agency
National Institute of Health (NIH)
Institute
National Institute of Mental Health (NIMH)
Type
Small Business Innovation Research Grants (SBIR) - Phase I (R43)
Project #
5R43MH110287-02
Application #
9254605
Study Section
Special Emphasis Panel (ZRG1-ETTN-C (10)B)
Program Officer
Grabb, Margaret C
Project Start
2016-04-05
Project End
2018-03-31
Budget Start
2017-04-01
Budget End
2018-03-31
Support Year
2
Fiscal Year
2017
Total Cost
$398,523
Indirect Cost
Name
Paradromics, Inc.
Department
Type
Domestic for-Profits
DUNS #
079841616
City
San Jose
State
CA
Country
United States
Zip Code
95112