Episode 24, Imaging
Worms on the Chip – Machine Learning Meets Toxicology
Worms on the Chip – Machine Learning Meets Toxicology
Show notes
In this episode, we explore a high-tech twist on developmental toxicology. Researchers have combined microfluidic engineering with machine learning to automate the analysis of thousands of C. elegans for chemical toxicity testing — no anaesthetics or low-res imaging required.
Using the vivoChip device and a custom ML model called vivoBodySeg, the team:
Captures 3D images of ~1000 worms from 24 populations at once
Achieves near-human segmentation accuracy (Dice score: 97.8%)
Measures subtle toxicity effects like changes in body size and gut autofluorescence
Identifies EC10 and LOAEL values with high precision
Uses few-shot learning to adapt the model to new worm shapes and sizes
This platform slashes analysis time by 140× and sets a new benchmark for high-throughput New Approach Methodologies (NAMs) in toxicology.
📖 Based on the research article:
“Machine learning-based analysis of microfluidic device immobilised C. elegans for automated developmental toxicity testing”
Andrew DuPlissis, Abhishri Medewar, Evan Hegarty, et al.
Published in Scientific Reports (2025)
🔗 https://doi.org/10.1038/s41598-024-84842-x
🎧 Subscribe to the Worm Podcast for more stories where whole-organism biology meets cutting-edge tech!
This podcast is generated with artificial intelligence and curated by Veeren. If you’d like your publication featured on the show, please get in touch.
📩 More info:
📧 veeren.chauhan@nottingham.ac.uk