Data from: Segmentation of dense and multi-species bacterial colonies using models trained on synthetic microscopy images

This repository contains the relevant data and code for the article "Segmentation of dense and multi-species bacterial colonies using models trained on synthetic microscopy images" by Vincent Hickl, Abid Khan, René M. Rossi, Bruno F. B. Silva, and Katharina Maniura-Weber (arXiv version 3): Raw synthetic images of rod-shaped and circular cells Confocal microscopy images of densely packed colonies of Pseduomonas aeruginosa, and of mixed colonies of P. aeruginosa and Staphylococcus aureus Brigthfield microscopy images of rod-shaped and circular cells Synthetic images processed by cycleGAN, used as training data for segmentation models. A selection of trained segmentation models for dense Pseudomonas monolayers, and rod-shaped/circular cells in both confocal and brightfield microscopy images. Manually annotated experimental test images used to compare the segmentation of the synthetic model to other models from the literature, and to calculate PQ vs IoU cutoff curves. The code for processing raw synthetic images using CycleGAN can be found at https://github.com/abid1214/cyclegan Further relevant code can be found at https://github.com/vhickl/synth-bacteria-segmentation, including code to: generate raw synthetic images of rod-shaped and spherical bacteria analyze orientational order in densely-packed colonies of rod-shaped cells quantitatively compare segmentation masks of bacterial colonies Please contact the authors ([e-mail]) if there are any issues or questions regarding this data.