Computational volumetric one-photon voltage imaging in rodents

Rockefeller University, New York NY

Investigators

Abstract

Project Summary To understand how the brain develops and accomplishes its unique computational and cognitive capabilities through the dynamics of three-dimensional networks of neurons remains a key goal of contemporary neuroscience. To inform accurate modeling of network dynamics, tools will be indispensable that enable recording of neuronal activity at both, the temporal scale at which neuronal communication and computation is performed, and a spatial scale that encompasses the ensemble of neurons participating in a given function. Recently, increasingly efficient genetically encoded voltage indicators (GEVIs) have been developed that allow optical readout of electrical neuronal activity at a millisecond timescale. However, recording of GEVI activity has remained limited to small fields-of-view and few axial planes, either due to low signal and speed achievable in scanning multiphoton methods or due to the low signal-to-background ratio that hampers existing one-photon modalities. To develop optical recording techniques capable of capturing the full electrical dynamics of large ensembles of neurons therefore remains an unmet need. This project aims to address this need by designing and demonstrating a novel ultra-high-speed volumetric imaging system capable of capturing the full dynamics of genetically encoded voltage indicators from thousands of neurons in the cortex and hippocampus of awake, behaving rodents. This will be achieved through a combination of structured one-photon excitation with advanced light field detection. The proposed technique will achieve a temporal resolution sufficient to capture the timing of single action potentials from thousands of neurons. The voltage dynamics of active neurons will be extracted and demixed from background and crosstalk by an advanced machine learning algorithm that exploits the presence of unscattered and weakly scattered light in the targeted depth range. The proposed approach is designed to deliver a time resolution approaching electrical recordings while also determining the 3D positions of all active neurons in a volume, in a manner that is immediately parallelizable to cover larger lateral areas at minimal invasiveness. If successful, this project will therefore contribute to enabling optical electrophysiology at the level of entire neuronal systems. 1