Coordinate-based neural representations for computational adaptive optics in two-photon microscopy

By measuring and correcting sample-induced aberrations, adaptive optics (AO) enables noninvasive imaging of subcellular structures in living organisms with two-photon (2P) fluorescence microscopy. We will introduce CoCoA-2P, a self-supervised machine-learning algorithm capable of simultaneously estimating aberrations and recovering 3D structural information from a single 2P image stack without requiring external training datasets. We will showcase the applications of CoCoA-2P for high-resolution in vivo structural imaging of the mouse brain and eye lenses.