Spatial resolution enhancement in photon-starved STED imaging using deep learning-based fluorescence lifetime analysis

Abstract

As a super-resolution imaging method, stimulated emission depletion (STED) microscopy has unraveled fine intracellular structures and provided insights into nanoscale organizations in cells. Although image resolution can be further enhanced by continuously increasing the STED-beam power, the resulting photodamage and phototoxicity are major issues for real-world applications of STED microscopy. Here we demonstrate that, with 50% less STED-beam power, the STED image resolution can be improved up to 1.45-fold using the separation of photons by a lifetime tuning (SPLIT) scheme combined with a deep learning-based phasor analysis algorithm termed flimGANE (fluorescence lifetime imaging based on a generative adversarial network). This work offers a new approach for STED imaging in situations where only a limited photon budget is available.