Abstract
Most fluorescence microscopy operates at irradiance levels approaching solar flux, a regime that can profoundly perturb the physiology of living systems that have evolved to function in near darkness. Reducing irradiance to mitigate phototoxicity, however, causes detector readout noise to dominate the signal. Detectors originally developed for defense applications overcome this limitation through on-chip amplification, pushing imaging into the Poisson (photon-sparse) noise regime. In this talk, we will present a Poisson-limited imaging strategy that combines efficient light-sheet illumination with deep learning to achieve a 10,000-fold reduction in irradiance relative to state-of-the-art microscopy.