A Novel Methodology for Characterizing Cell Subpopulations in Automated Time-lapse Microscopy

Area of Science:

• Cell biology
• Bioimage analysis
• Computational biology

Background:

• Time-lapse imaging (biomovies) captures cell colony dynamics in microfluidic chambers.
• Analyzing biomovies for cell growth and lineage is challenging due to high cell density, diversity, and image noise.
• Existing methods struggle with accurate single-cell detection, limiting subpopulation analysis.

Purpose of the Study:

• To develop a robust method for characterizing cell subpopulations within biomovies.
• To shift data analysis from individual cells to groups with similar characteristics (e.g., fluorescence).
• To automate the extraction of cell lineage and growth patterns.

Main Methods:

• A three-step framework: preprocessing, particle tracking, and patch lineage construction.
• Preprocessing enhances signal-to-noise ratio and spatial alignment of biomovie frames.
• Cells are abstracted as 'particles' for tracking, followed by linking trajectories and filtering artifacts.
• Patch lineage construction groups related particle trajectories based on proximity and fluorescence to identify subpopulations.
• The CYCASP framework generates a complete graph of patch lineages, encoding spatio-temporal colony development.

Main Results:

• The CYCASP method successfully separates cell colonies into distinct subpopulations.
• It enables timely interpretation of colony growth dynamics.
• Demonstrated computation time of less than 5 minutes for biomovies and simulated films.

Conclusions:

• CYCASP offers an efficient and accurate approach to biomovie analysis, overcoming limitations of single-cell tracking.
• The particle and patch abstractions facilitate the characterization of cellular groups and their lineage.
• This method significantly advances the analysis of cell colony behavior over time.