A Real-Time Automated Deep Learning Workflow for Non-invasive High-Magnification Imaging of C. elegans

Area of Science:

• Aging research
• Neurobiology
• Model organism studies

Background:

• Caenorhabditis elegans (C. elegans) is a key model organism for aging and neurobiology.
• Non-invasive imaging is crucial for preserving natural physiology in C. elegans.
• High-magnification imaging of freely moving C. elegans is challenging due to motion blur and focal drift.

Purpose of the Study:

• To develop a real-time tracking workflow for imaging individual C. elegans.
• To enable stable, high-magnification imaging of freely moving worms.
• To support longitudinal phenotyping in aging and neurobiology studies.

Main Methods:

• A microfluidic platform with deep learning head detection was used.
• Image-based autofocus and motorized-stage feedback ensured stable imaging.
• The system allowed for repeated daily imaging of individual C. elegans throughout their lifespan.

Main Results:

• Stable imaging of freely moving C. elegans was achieved across multiple magnifications, including neuronal scale.
• The workflow integrated deep learning, autofocus, and stage feedback.
• Fluorescence images of freely moving worms were comparable to immobilized ones.

Conclusions:

• The developed real-time tracking workflow facilitates non-invasive, longitudinal imaging of C. elegans.
• This method overcomes challenges of high-magnification imaging in freely moving organisms.
• The platform supports advanced aging and neurobiology research in C. elegans without compromising physiology.