Anatomical Identification of Extracellularly Recorded Cells in Large-Scale Multielectrode Recordings

Abstract

This study combines for the first time two major approaches to understanding the function and structure of neural circuits: large-scale multielectrode recordings, and confocal imaging of labeled neurons. To achieve this end, we develop a novel approach to the central problem of anatomically identifying recorded cells, based on the electrical image: the spatiotemporal pattern of voltage deflections induced by spikes on a large-scale, high-density multielectrode array. Recordings were performed from identified ganglion cell types in the macaque retina. Anatomical images of cells in the same preparation were obtained using virally transfected fluorescent labeling or by immunolabeling after fixation. The electrical image was then used to locate recorded cell somas, axon initial segments, and axon trajectories, and these signatures were used to identify recorded cells. Comparison of anatomical and physiological measurements permitted visualization and physiological characterization of numerically dominant ganglion cell types with high efficiency in a single preparation.

Optogenetic therapy: High spatiotemporal resolution and pattern recognition compatible with vision restoration in non-human primates

“…Importantly, we demonstrate in this paper that light intensities below the 600 nm radiation safety limit can activate the transfected retina very efficiently. 33 Exposure limits are, by nature, conservative. They concern here a continuous illumination, as the higher risk is the photochemical hazard, which depends on the total illumination received per 48 hours.…”

Global motion processing by populations of direction-selective retinal ganglion cells

“…Soma position on the array was used to identify quadruplets (Fig 2) and pairwise distances between cells ( Fig 3A). Soma position was estimated from the Electrical Image (EI) on the array 11,65 . The EI consisted of the average voltage on each electrode preceding a spike 66 .…”

Validation of neural spike sorting algorithms without ground-truth information

“…For the sake of simplicity, we will restrict our attention here to stationary data, but note that the framework for quality assessment that we introduce below does not depend on stationarity Unfortunately, despite major progress in algorithms and software, spike sorting remains a labor-intensive task with a substantial manual component, both in the clustering and quality assessment phases. Furthermore, in the last two decades, the number of channels from which it is becoming possible to record simultaneously has grown from one (or a few) to thousands [6,18,27,31]. As a result, it has become an urgent matter to accelerate the data analysis, to automate the clustering algorithms, and to develop statistical protocols that can provide robust estimates of the accuracy of the neuronal identification.…”