Trajectory analysis of in micro-droplets.

In order to study biofilm formation in microdroplets, we use microfluidics technology to make the droplets and confocal microscopy to capture bacterial movement and biofilm formation in the droplets. We develop a multi-target tracking methodology, using a YOLOv5 detector to identify cells and a DeepSORT algorithm to track cell movements. We find that bacteria with autonomous migration and biofilm-forming ability prefer to cluster and swarm near the microdroplet surface, rather than in the droplet interior. Bacterial mobility depends on phenotype and spatial location within the droplet. The motile cells move about 3.5 times faster than the matrix-producing cells. When the cells are near the wall of the droplet, the direction of the motion of motile cells is along that wall. When the cells are inside the droplet, the direction of the motion of motile cells is disordered, i.e., there is no clear directional or goal-oriented movement. This contrast increases the cell contact probability and facilitates the formation of a biofilm in the droplet. Furthermore, we develop a mathematical model to describe the motion behavior of in microdroplets, which is useful for exploring the influence of motility on biofilm formation.