Abstract
Cell motility and predation are important for the dynamics of many multi-cellular ecosystems, such as the gut or the soil. Approaches to image cell dynamics in such complex systems are scant, and high-throughput analysis methods to segment and track single-cell behaviors are currently lacking. Here, we addressed these limitations by implementing a fast fluorescence microscopy technique enabling the high-resolution acquisition of cell movement over large areas and long time periods. Next, we applied deep learning to semantically segment two different bacteria species within complex micro-environments . We implemented a method to build single cell traces by combining the cell masks from different time points to follow the dynamics of single cells with high spatial and temporal resolutions and over long periods of time. We applied and validated these methods by characterizing the dynamics of Escherichia coli predation by Myxococcus xanthus .