Segmentation, tracking and cell cycle analysis of live-cell imaging data with Cell-ACDC

Area of Science:

• Cell Biology
• Biotechnology
• Bioinformatics

Background:

• High-throughput live-cell imaging generates vast datasets, posing analysis challenges.
• Existing analytical pipelines struggle with data volume and require manual correction.
• Deep learning has advanced cell segmentation and tracking but comprehensive tools are lacking.

Purpose of the Study:

• To introduce Cell-ACDC, an open-source, GUI-based framework for live-cell imaging data analysis.
• To provide state-of-the-art deep learning models for cell segmentation and tracking.
• To enable intuitive, semi-automated cell cycle annotation.

Main Methods:

• Development of Cell-ACDC, a Python-based graphical user interface (GUI).
• Integration of deep learning models for cell segmentation (mammalian and yeast).
• Implementation of cell tracking algorithms and a semi-automated cell cycle annotation workflow.

Main Results:

• Cell-ACDC facilitates analysis of live-cell imaging data without programming.
• Demonstrated independence of mTOR activity from cell volume in hematopoietic stem cells.
• Observed higher p38 activity in smaller cells, suggesting a role in size regulation.
• Found decreasing histone Htb1 concentrations with replicative age in S. cerevisiae.

Conclusions:

• Cell-ACDC offers a user-friendly platform for applying deep learning to live-cell imaging.
• The tool allows for visualization and correction of segmentation and tracking errors.
• Its open-source and modular design supports integration of new analysis methods.