Single‐cell segmentation in bacterial biofilms with an optimized deep learning method enables tracking of cell lineages and measurements of growth rates

Abstract

Bacteria often grow into matrix‐encased three‐dimensional (3D) biofilm communities, which can be imaged at cellular resolution using confocal microscopy. From these 3D images, measurements of single‐cell properties with high spatiotemporal resolution are required to investigate cellular heterogeneity and dynamical processes inside biofilms. However, the required measurements rely on the automated segmentation of bacterial cells in 3D images, which is a technical challenge. To improve the accuracy of single‐cell segmentation in 3D biofilms, we first evaluated recent classical and deep learning segmentation algorithms. We then extended StarDist, a state‐of‐the‐art deep learning algorithm, by optimizing the post‐processing for bacteria, which resulted in the most accurate segmentation results for biofilms among all investigated algorithms. To generate the large 3D training dataset required for deep learning, we developed an iterative process of automated segmentation followed by semi‐manual correction, resulting in >18,000 annotated Vibrio cholerae cells in 3D images. We demonstrate that this large training dataset and the neural network with optimized post‐processing yield accurate segmentation results for biofilms of different species and on biofilm images from different microscopes. Finally, we used the accurate single‐cell segmentation results to track cell lineages in biofilms and to perform spatiotemporal measurements of single‐cell growth rates during biofilm development.

In Vivo Microrheology Reveals Local Elastic and Plastic Responses Inside 3D Bacterial Biofilms

“…Single-Cell Segmentation and Analysis: For 3D segmentation of all individual cells in biofilm images, a trained model of StarDist OPP, which is a cell segmentation algorithm based on a convolutional neural network (CNN), was used. [53] The StarDist OPP model was trained with a dataset involving 3D slices of V. cholerae biofilms imaged by the microscope system. By using StarDist OPP, the accuracy of cell segmentation for 3D bacterial biofilm volume image is above 80%, which is higher than any other single-cell segmentation method achieved on the biofilm image data.…”

“…In step 1 of this algorithm, the three 3D biofilm images of the time series were segmented using the trained CNN model that is part of the StarDist OPP algorithm. [ 53 ] This segmentation results in a labeled object for each individual cell. In step 2 of the algorithm, one cell at the bottom surface of the biofilm is tracked manually.…”

“…To achieve this, we coupled highly reproducible microfluidic biofilm cultivation systems and flow control [52] with recent advances in 3D live-cell microscopy and highly-accurate deep-learning-based image analysis for cell detection. [53][54][55][56] Our experimental platform enabled us to track the individual cells in 3D biofilms of Vibrio cholerae during and after exposure to strong fluid shear, for a wide range of different shear rate amplitudes, exposure periods, and biofilm sizes. To reconstruct the internal material properties of biofilms, we developed a computational inference method for estimating elastic forces from cell tracking data.…”

“…To achieve this, we coupled highly reproducible microfluidic biofilm cultivation systems and flow control [ 52 ] with recent advances in 3D live‐cell microscopy and highly‐accurate deep‐learning‐based image analysis for cell detection. [ 53–56 ]…”

GND-PCA Method for Identification of Gene Functions Involved in Asymmetric Division of C. elegans

“…In recent years, using deep learning in feature detection and image processing has become a hot topic [35]. Jelli E. et al applied deep learning to segment cells in 3D images [36]. Greenwald N. F. et al constructed TissueNet using deep learning techniques to segment tissue types in [37].…”