Neuropixels NXT: Integrated Silicon Probes for Large Scale Extracellular Recording in Rodents and Primates

点击翻译按钮获取中文摘要

1 Project Summary: The BRAIN 2025 Report, with the goal to “produce a dynamic picture of the functioning brain 2 by developing and applying improved methods for large-scale monitoring of neural activity” is directly 3 addressed by this application. The Neuropixels probe (Jun et al., Nature 2017) demonstrated that a high 4 channel count Si shank with continuous, dense, programmable sites yielded a large capacity increase in 5 monitoring of neural activity (100's vs. 10's of units for typical devices). Neuropixels NXT will increase 6 substantially the utility of that high capacity shank by optimizing and deploying a new platform technology for 7 neural recording, direct-to-digital. Neuropixels, while revolutionary over prior art, has an integrated probe 8 “base” for amplification and digitization and a digital interface board (head stage) that both reside above the 9 brain. Together, the crowding of these components above the brain limits the number of shanks that can be 10 used in one experimental animal. Follow-on projects are already underway to relieve this crowding and weight 11 such that we expect 2 Neuropixels probes (768 channels of recording capacity) could be used on a freely 12 moving, tethered mouse. While this represents a substantial improvement, it remains far short of the brain wide 13 recording that satisfies the above stated ambition to “produce a dynamic picture of the functioning brain”. 14 Building on the demonstrated performance of the Neuropixels shank, we propose to optimize the base 15 electronics so that a probe with ~768 channels would have a base of < 5mm2 (an 18x shrinkage of the base 16 area per channel compared to Neuropixels, and 6x shrinkage compared to Neuropixels 2.0), require no extra 17 digital interface board (head stage), and instead have a 6-conductor micro-cable between the probe and a high 18 speed data hub for digital transmission that can handle multiple probes simultaneously. This smaller probe 19 base will allow it to be made narrow so that each probe will act as a “unit shank” that can be tiled side-to-side 20 into a “comb”, and combs stacked into 2 dimensional arrays of shanks, yielding dense 3 dimensional arrays of 21 programmable recording sites. We project capacities of >16 shanks (9600 sites/channels, all active) in a 22 mouse, 40 shanks (40,000+ programmable sites, 26,000+ channels) in a rat, and up to 100 shanks (450,000 23 programmable sites, 76.800 channels) in a non-human primate. The work plan will first optimize: 1) the design 24 of the direct-to-digital recording channel pixel for size/noise tradeoff, and 2) the data hub to allow up to 12 25 probes to be served by a single output cable. With this information, a family of “unit shanks” will be produced, 26 ready for sale to the research community, which can be fixtured into a wide variety of recording geometries. 27 These developments in an environment capable of long term manufacture and affordable costs will ensure 28 wide availability.