A Dynamic Contour Evolution Algorithm for Cell Segmentation and Synaptic Tracking under Occlusion

Area of Science:

• Cell biology
• Bioinformatics
• Computational biology

Background:

• Cell migration is crucial for biological processes but challenging to quantify due to cell variability and occlusion.
• Dysregulation of cell migration is linked to various diseases.

Purpose of the Study:

• To develop a robust algorithm for accurate single-cell segmentation and tracking.
• To enable precise quantification of cellular dynamics under diverse conditions.

Main Methods:

• A novel framework combining Dynamic Profile Evolution (DPE) contour deformation network and a graph neural network (GNN).
• Algorithm designed to handle morphological variability and cell occlusion for reliable tracking.
• Validation on HT22 cell datasets demonstrating improved recognition and identity continuity.

Main Results:

• The DPE-GNN framework accurately segments and tracks cells, outperforming existing methods.
• Analysis of over 4,000 cells revealed stable morphology and migration patterns under normal conditions.
• Oxidative stress was observed to reduce cell motility and alter migration trajectories.

Conclusions:

• The developed framework provides a versatile tool for precise quantification of cellular dynamics.
• Enables long-term live-cell monitoring and high-content analysis of cell behaviors.
• Offers insights into how environmental factors like oxidative stress affect cell migration patterns.