Deep learning enables structured illumination microscopy with low light levels and enhanced speed

• DOI: 10.1038/s41467-020-15784-x
• Publication Date: 2020-04-22T10:03:13Z
• Authors: Luhong Jin; Bei Liu; Fenqiang Zhao; Stephen Hahn; Bowei Dong; Ruiyan Song; Timothy C. Elston; Yingke Xu; Klaus M. Hahn
• Keywords: Photobleaching

Abstract Structured illumination microscopy (SIM) surpasses the optical diffraction limit and offers a two-fold enhancement in resolution over limited microscopy. However, it requires both intense multiple acquisitions to produce single high-resolution image. Using deep learning augment SIM, we obtain five-fold reduction number of raw images required for super-resolution generate under extreme low light conditions (at least 100× fewer photons). We validate performance neural networks on different cellular structures achieve multi-color, live-cell imaging with greatly reduced photobleaching.