Tools and methods for high-throughput single-cell imaging with the mother machine.

Area of Science:

• Microscopy and Image Analysis
• Computational Biology
• Cell Biology

Background:

• High-throughput analysis of microscopy data, particularly from mother machine platforms for single-cell time-lapse imaging, faces significant bottlenecks in image processing.
• Existing mother machine image analysis pipelines have limited adoption by non-expert users.

Purpose of the Study:

• To introduce napari-MM3, a novel, modular, and interactive image analysis pipeline for mother machine data, integrated as a plugin for napari.
• To quantitatively compare napari-MM3 with existing pipelines (BACMMAN, DeLTA) and assess the impact of different analysis methods on single-cell data interpretation.

Main Methods:

• Development of napari-MM3, a software plugin for the napari multidimensional image viewer.
• Comparative analysis of mother machine datasets using napari-MM3, BACMMAN, and DeLTA.
• Investigation of parameter sensitivity, particularly thresholding and deep learning segmentation, on extracted single-cell physiological parameters.

Main Results:

• Key single-cell physiological parameter correlations and distributions are generally robust across different analysis pipelines.
• Minor adjustments in thresholding parameters can systematically influence extracted experimental parameters.
• Deep learning segmentation is susceptible to 'what you put is what you get' (WYPIWYG), where pixel-level training data variations bias spatial and temporal measurements.

Conclusions:

• napari-MM3 offers a user-friendly solution for mother machine image analysis, leveraging napari's interactivity.
• While analysis methods show robustness, careful parameter selection and awareness of deep learning biases are crucial for accurate single-cell data interpretation.
• The study provides valuable insights for researchers implementing mother machine-based high-throughput imaging and analysis.