Computational optical sectioning by phase-space imaging with an incoherent multiscale scattering model

Optical sectioning is essential for fluorescence imaging in thick tissue to extract in-focus information from noisy background. Traditional methods achieve optical sectioning by rejecting the out-of-focus photons at a cost of photon efficiency, resulting in a tradeoff between sectioning capability and detection parallelization. Here, we show phase-space imaging with an incoherent multiscale scattering model can achieve computational optical sectioning with ~20 dB improvement for signal-to-background ratio in scattering medium, while maximizing the detection parallelization by imaging the entire volume simultaneously. We validated the superior performance by imaging various biological dynamics in Drosophila embryos, zebrafish larvae, and mice.