A flexible framework for automated STED super-resolution microscopy

Abstract

Super-resolution microscopy enables the observation of cells at unprecedented detail but usually entails high light exposure and slow imaging. Thus, often only a few manually selected regions are imaged, limiting the ability to capture the distribution of quantitative features in a population of cells in an unbiased fashion. An exciting strategy to circumvent these limitations are imaging pipelines in which informative regions are detected on-the-fly by software and imaged automatically. Point-scanning methods like STimulated Emission Depletion (STED), in particular, can be sped up by selective imaging of small regions. Here, I present autoSTED, a flexible Python-based framework to construct automated imaging pipelines for STED microscopy. Instead of fixed acquisition loops defined at the beginning of an experiment, autoSTED employs a priority queue of acquisition tasks. After each image acquisition, callback functions can trigger actions like adding new tasks based on data, enabling dynamic and adaptive imaging. Complex experimental pipelines can be built from easily exchangeable building-blocks or expanded through custom code, facilitating integration of state-of-the-art computer vision methods. autoSTED can drastically speed up super-resolved imaging of subcellular structures and enables autonomous operation of a microscope for days with minimal hands-on time and bias.