Abstract
Purpose :
High-density electrophysiology probes, such as Neuropixels, are useful tools for recording the activity of hundreds of neurons simultaneously with high spatiotemporal precision. This capability promises to revolutionize approaches to systems and population analyses in the central visual system. With the growing adoption of these technologies, a major limitation in the analysis pipeline can be the characterization of single neurons and their correlations to individual stimuli. In order to utilize high-density electrophysiology for visual neuroscience, it is vital to correlate individual neuronal activity with respect to visual space: that is, characterize each neuronal receptive field (RF). The mapping of RFs presents unique challenges due to the variable nature of visual responses. There are a number of approaches to high-throughput RF mapping in the literature, though there is variation in data preprocessing, significance testing, and RF characterization. To date, there is no published comparison between high-throughput mapping methods to inform best practices. More importantly, the lack of standardization across the field limits quantitative rigor, conclusion generalization, and experimental replicability. We hypothesize that each RF estimation method would result in statistically indistinguishable RF parameters.
Methods :
We compared the center location and spatial extent of three receptive field mapping methods (spatiotemporal noise, sweep mapping, masked gratings) and their respective analytical approaches on neurons in dorsolateral geniculate nucleus of the thalamus (dLGN) and primary visual cortex (V1) using multi-Neuropixels recordings of simultaneous activity in hundreds of single neurons in mice.
Results :
A single, two-Neuropixels recording yielded a total of 439 units. From these units, we found 60 RFs using spatiotemporal noise, 42 using sweep mapping, and 177 using masked gratings.
Conclusions :
We hypothesize that each RF estimation method would result in statistically indistinguishable RF parameters. Preliminary results suggest significant differences between these methods in total RF yield, RF size, and RF center. These measurements are vital for subsequent analyses correlating pairwise and population activity between functional networks in the visual hierarchy.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.