Optical Oscilloscope: Real-time, high-throughput, volumetric voltage imaging

Neuroscientists lack tools to monitor the voltage of thousands of neurons in 3D and in real time. This is limiting because neurons communicate electrically, and network-scale voltage dynamics drive both cognitive function and dysfunction. Neuroscientists currently use multiphoton scanning to image through scattering in brain tissue, but scanning is too slow to capture small, fast voltage signals across large populations of neurons. We will develop high-throughput, real-time, volumetric voltage and vascular imaging based on light-field microscopy (LFM). LFM enables scanless volume acquisition, but scattering and computational load currently limit its use. We will overcome these barriers by developing optics-aware deep neural networks trained on one- photon light fields and scattering-robust two-photon volumes, and implement them on graphics processing units and field-programmable gate arrays for the real-time, kilohertz readout required for closed-loop control. These unique capabilities will enable discovery science including mapping neurovascular dysfunction in Alzheimer’s disease, uncovering network-scale learning rules, and probing how hippocampal replay shapes cortical dynamics and connectivity. The Optical Oscilloscope will deliver low-latency, cellular-resolution voltage imaging at network scale, enabling new insight into how neural circuits function and remodel during learning and disease. Key words: voltage imaging, deep learning, light field microscopy, two-photon microscopy, neural circuits, real-time, closed-loop