Segmentation and analysis of mother machine data: SAM

Abstract

Time-lapse imaging of bacteria growing in micro-channels in a controlled environment has been instrumental in studying the single cell dynamics of bacterial growth. This kind of a microfluidic setup with growth chambers is popularly known as mother machine [1]. In a typical experiment with such a set-up, bacterial growth can be studied for numerous generations with high resolution and temporal precision using image processing. However, as in any other experiment involving imaging, the image data from a typical mother machine experiment has considerable intensity fluctuations, cell intrusion, cell overlapping, filamentation etc. The large amount of data produced in such experiments makes it hard for manual analysis and correction of such unwanted aberrations. We have developed a modular code for segmentation and analysis of mother machine data (SAM) for rod shaped bacteria where we can detect such aberrations and correctly treat them without manual supervision. We track cumulative cell size and use an adaptive segmentation method to avoid faulty detection of cell division. SAM is currently written and compiled using MATLAB. It is fast ([~] 15 min/GB of image) and can be efficiently coupled with shell scripting to process large amount of data with systematic creation of output file structures and graphical results. It has been tested for many different experimental data and is publicly available in Github.