Automating superresolution microscopy image analysis with a novel deep learning architecture to probe intracellular trafficking

Abstract

Optical imaging is integral to health care for diagnostics and disease mapping. With the advent of superresolution microscopy, we can now visualize the biology of health and disease at the sub-cellular level. Such cutting-edge fluorescence microscopy methods generate large amounts of high precision images and videos, demanding high-level image processing and data analysis automation. For instance, optical video microscopy of mammalian cells with high spatial (60-250 nm) and temporal (20 frames/seconds) resolution imaging system yields hundreds of gigabytes of data, depicting the fast dynamics of the nano-sized sub-cellular compartments.