Automated cell lineage reconstruction using label-free 4D microscopy.

Area of Science:

• Developmental biology
• Computational biology
• Bioimaging

Background:

• Lineage tracing is crucial for understanding metazoan embryo development, especially in invariant organisms like C. elegans.
• Current methods rely on fluorescence microscopy and manual tracing, limiting throughput and requiring transgenic approaches.
• Automating cell detection and tracking in label-free 3D imaging remains a significant challenge.

Purpose of the Study:

• To develop a deep learning pipeline for automated cell detection and tracking in label-free 3D time-lapse imaging.
• To overcome limitations of manual annotation and fluorescence-based methods in lineage tracing.
• To enable high-throughput developmental studies without genetic modification.

Main Methods:

• Developed embGAN, a deep learning pipeline for label-free 3D cell detection and tracking.
• Trained embGAN without manual data annotation, ensuring scale invariance and generalization across labs and instruments.
• Benchmarked performance using cell lineage tracing in C. elegans embryos.

Main Results:

• embGAN achieves robust cell detection and tracking in dense embryonic tissues.
• The pipeline demonstrates high scale invariance and generalizes well to diverse imaging conditions.
• Near state-of-the-art performance was achieved in cell detection and tracking for C. elegans embryos.

Conclusions:

• embGAN provides an automated solution for cell lineage tracing in label-free 3D imaging.
• This method facilitates high-throughput developmental studies, reducing reliance on fluorescent reporters and transgenics.
• The tool advances the field of computational developmental biology by enabling efficient analysis of complex embryonic processes.