Ultra-high density electrodes improve detection, yield, and cell type identification in neuronal recordings

Abstract

To study the neural basis of behavior, we require methods to sensitively and accurately measure neural activity at single neuron and single spike resolution. Extracellular electrophysiology is a principal method for achieving this, but it has biases in the neurons it detects and it imperfectly resolves their action potentials. To overcome these limitations, we developed a silicon probe with significantly smaller and denser recording sites than previous designs, called Neuropixels Ultra (NP Ultra). This device measures neuronal activity at ultra-high densities (>1300 sites per mm, 10 times higher than previous probes), with 6 μm center-to-center spacing and low noise. This device effectively comprises an implantable voltage-sensing camera that captures a planar image of a neuron's electrical field. We introduce a new spike sorting algorithm optimized for these probes and use it to find that the yield of visually-responsive neurons in recordings from mouse visual cortex improves ~3-fold. Recordings across multiple brain regions and four species revealed a subset of unexpectedly small extracellular action potentials not previously reported. Further experiments determined that, in visual cortex, these do not correspond to major subclasses of interneurons and instead likely reflect recordings from axons. Finally, using ground-truth identification of cortical inhibitory cell types with optotagging, we found that cell type was discriminable with approximately 75% success among three types, a significant improvement over lower-resolution recordings. NP Ultra improves spike sorting performance, sampling bias, and cell type classification.

DREDge: robust motion correction for high-density extracellular recordings across species

“… a Motion estimation from spikes detected in imposed motion datasets from Neuropixels 1, 2 (i, ii) and Ultra probes (iii) 6;34 . DREDge’s motion traces (left column, blue) and motion-corrected spike rasters (middle column) match the quality of Kilosort’s (left column, red; right column) on NP1 and NP2 data (i, ii).…”

“…Fig. 15 were reported in Ye et al 34 , and feature a very dense electrode layout, with 384 sites arranged in a 64 × 6 grid with 6μm vertical and horizontal channel spacing. Here, we focused on recordings with zig-zag motion imposed by a similar methodology as discussed above; more details are available in the reference.…”

“…Further, unlike KS, DREDge was able to track the same imposed zig-zag motion, plus additional probe motion, in recordings made with the Neuropixels Ultra (NP Ultra 34 ) probe (Fig. 5.a, iii).…”

“…We applied DREDge to in vivo datasets from a variety of species and MEA types, including human Neuropixels recordings 15;16 , recordings in mice from the International Brain Lab’s large-scale reproducible electrophysiology experiment 32 , non-human primate recordings during probe insertion 33 , and mouse recordings using the experimental ultra high-density Neuropixels probe 34 , among others. Through these experiments, we demonstrate the usage and utility of DREDge along with some of the novel downstream analyses that it enables.…”

Development of wafer-scale multifunctional nanophotonic neural probes for brain activity mapping

“…15 Also, complementary techniques are still required to validate the identification of certain cell types through extracellular recordings. 16,17 These shortcomings can be overcome with optogenetics and functional fluorescence imaging, where genetic modifications make neurons of specific cell types interrogable with light. 8,9 For single-photon absorption, optical excitation typically uses a wavelength in the visible spectrum.…”