Preprint Highlight: Event-driven acquisition for content-enriched microscopy.

Abstract

Biological processes act over a broad range of temporal scales. Imaging fast-occurring and rare events would benefit from an on-the-fly adaptation of the acquisition frame rate according to the biological process' dynamics. This study presents an event-driven acquisition framework, where real-time machine-learning recognition of the onset of specific biological events triggers faster image acquisition. This framework was applied to follow mitochondrial fissions and bacterial divisions, achieving better temporal resolution during the events of interest along with minimized photo damage. Open code and a Micro-Manager plugin are provided, enabling adaptation to different microscopes and biological systems. The integration of real-time detection of precursors to specific events followed by tailored image acquisition will enable the enhanced characterization of fast and transient biological processes.