The aim of the research was to design an advanced analytical tool for the precise characterization of microbial aggregates from biofilms formed on food-processing surfaces. The approach combined imaging flow cytometry with a machine learning-based interpretation protocol. Biofilm samples were collected from three diagnostic points of the food-processing lines at two independent time points. The samples were investigated for the complexity of microbial aggregates and cellular metabolic activity. Thus, aggregates and singlets of biofilm-associated microbes were simultaneously examined for the percentages of active, mid-active, and nonactive (dead) cells to evaluate the physiology of the microbial cells forming the biofilm structures. The tested diagnostic points demonstrated significant differences in the complexity of microbial aggregates. The significant percentages of the bacterial aggregates were associated with the dominance of active microbial cells, e.g., 75.3% revealed for a mushroom crate. This confirmed the protective role of cellular aggregates for the survival of active microbial cells. Moreover, the approach enabled discriminating small and large aggregates of microbial cells. The developed tool provided more detailed characteristics of bacterial aggregates within a biofilm structure combined with high-throughput screening potential. The designed methodology showed the prospect of facilitating the detection of invasive biofilm forms in the food industry environment.
Keywords: biofilm dispersal; bioimaging; food-processing; machine learning; single-cell analysis.