Direct readout of spiking and subthreshold voltage activity from genetically-specified neural populations will facilitate major advances in our understanding of neural computation at the network level. For this reason, developing a genetically encoded voltage indicator (GEVI) with adequate speed, membrane localization, and brightness to report action potentials in mammalian cells has been a major goal in neuroscience for the past two decades. Recently, we developed an opsin-based GEVI, called Archon, which exhibits good localization in neurons of multiple species, several fold improved brightness over previous opsin-based reporters, order-of-magnitude improvements in voltage sensitivity and photo bleaching over GFP-like reporters, and compatibility with optogenetic control. However, action potential imaging, even in a single Archon-expressing neuron, requires a state of the art sCMOS camera to meet sample-rate and signal-to-noise (SNR) requirements. These devices are large, power-hungry, bandwidth intensive, and support a limited field of view at maximal frame rate. This precludes their integration into head-mountable devices and greatly limits their ability to image large numbers of cells. The goal of this proposal is to create a head mountable miniature microscope (?miniscope?) that enables high speed (>1000 frames per second; FPS) voltage imaging of neural populations in freely moving animals. To do this, we will adapt our recently developed integrated compressed sensing CMOS image sensor with pixel-wise exposure (?CS-PCE Camera?) for GEVI imaging. The CS- PCE camera's key innovation is to permit independent exposure of each pixel on the sensor array instead of exposing the entire array in lock-step with a frame clock. This enables CS reconstruction of high speed video from samples acquired much more slowly than the Shannon/Nyquist rate. In this proposal, we demonstrate that a 1000 FPS video of Archon readout of neuronal action potentials can be accurately reconstructed from a CS-PCE camera operated at 100 FPS. We demonstrate that using long pixel-wise exposure and slow readout saves power, reduces system size, and increases the SNR while maintaining action potential detectability compared to state of the art sCMOS imaging devices. Taken together, we aim to provide a genetically-targetable replacement for microelectrode-based recordings, which has been a long sought after goal in systems neuroscience research. To evaluate device performance, we will use the CS-PCE miniscope to record hippocampal place-cell sequence replay, which exceeds the temporal bandwidth of calcium-based functional imaging techniques and is currently only accessible using electrical recordings.
Electrical signaling is the basis of communication and computation in neural circuits. The purpose of this project is to develop a fundamentally new type of image sensor that allows direct optical readout of voltage signals from many neurons simultaneously in freely moving animals. This technology will provide key insights into how neural circuits operate normally and how disease states that affect voltage signaling lead to circuit disfunction.
|Hu, Sile; Ciliberti, Davide; Grosmark, Andres D et al. (2018) Real-Time Readout of Large-Scale Unsorted Neural Ensemble Place Codes. Cell Rep 25:2635-2642.e5|