Phase-contrast time-lapses of seven bacterial species growing in microfluidic mother machine traps

This dataset and software accompany the article "Rapid label-free identification of seven bacterial species using microfluidics, single-cell time-lapse phase-contrast microscopy, and deep learning-based image and video classification" for reproducing the results.

In the study, deep-learning models are trained to classify phase-contrast videos (time-lapses) of bacteria growing in microfluidic chip traps. The dataset consists of lab isolates of the species Pseudomonas aeruginosa, Escherichia coli, Klebsiella pneumoniae, Acinetobacter baumannii, Enterococcus faecalis, Proteus mirabilis, and Staphylococcus aureus. The video clips have around 30 frames each, captured during one hour of growth (2 minutes between each frame). The whole dataset consists of around 620,000 images from 19,500 traps.

Additionally, the package contains software to re-run the experiments, generate output metrics, and build the graphs in the article.