LFCT: A Benchmark Dataset for Low-Frame-Rate Cell Tracking in Long-Term Live-Cell Microscopy

Area of Science:

• Cell biology
• Microscopy
• Bioimage analysis

Background:

• Cell tracking in time-lapse microscopy is vital for understanding cellular dynamics.
• Current benchmarks often overlook low-frame-rate imaging, limiting algorithm evaluation for long-term studies.
• Sparse temporal sampling and large inter-frame motion pose significant challenges in live-cell imaging.

Purpose of the Study:

• To introduce the Low Frame-rate Cell Tracking dataset (LCFT) as a novel benchmark.
• To facilitate the development and evaluation of cell tracking algorithms under low-frame-rate conditions.
• To provide a comprehensive resource for robust cell tracking in long-term live-cell imaging.

Main Methods:

• Collected multi-day live-cell microscopy sequences from four human cell lines.
• Utilized phase-contrast and fluorescence imaging at 10× and 20× magnifications.
• Combined automated segmentation and tracking with extensive manual curation for ground-truth annotation.

Main Results:

• Developed the LCFT dataset featuring diverse cell lines and imaging conditions.
• Provided detailed ground-truth annotations including cell IDs, temporal links, lineage, and mitosis events.
• LCFT enables rigorous benchmarking of cell tracking algorithms for sparse temporal sampling.

Conclusions:

• The LCFT dataset is a valuable resource for advancing cell tracking methodologies.
• It specifically addresses the unmet need for evaluating algorithms in low-frame-rate, long-term imaging scenarios.
• LCFT will drive the development of more robust and accurate cell tracking solutions for biological research.