Related Experiment Video
Updated: Aug 6, 2026
07:36
C. elegans Tracking and Behavioral Measurement
Published on: November 17, 2012
PumpKin: A machine-learning pipeline for automatically tracking localized kinematics in freely moving C. elegans
Erin Shappell1,2, Debra Buggs3, Jennah Walcott4
1Interdisciplinary Program in Bioengineering, Georgia Institute of Technology, Atlanta, GeorgiaUnited States of America.
Plos Computational Biology
|July 17, 2026
Summary
We developed PumpKin, a novel method for tracking single-body part movements in C. elegans, improving accuracy for behaviors like pharyngeal pumping. This tool reliably measures feeding rates, outperforming human annotators in precision.
Area of Science:
- Neuroscience
- Animal Behavior Analysis
- Biophysics
Background:
- Understanding neural encoding of sensory information into behavior is a key neuroscience goal.
- Tracking animal pose is crucial for studying behavior, but current methods struggle with single-keypoint tracking accuracy.
- Localized behaviors, like C. elegans feeding, are sensitive to noise from other body parts, posing a challenge for existing tracking techniques.
Purpose of the Study:
- To develop a fast, automated method for reliably measuring pharyngeal pumping in freely moving Caenorhabditis elegans (C. elegans).
- To address the limitations of existing pose tracking techniques in accurately isolating and quantifying single-body part movements.
- To provide a robust tool for analyzing feeding behavior dynamics across various experimental conditions.
Main Methods:
- Combined a state-of-the-art object detector (Faster Region-based Convolutional Neural Network) with a tunable noise filter.
- Developed a technique named PumpKin for automated and reliable measurement of C. elegans pumping.
- Validated PumpKin's performance across eight different experimental conditions, including varying satiety and genetic backgrounds.
Main Results:
- PumpKin demonstrated high speed, averaging 0.4 seconds per frame.
- The method accurately estimated average pumping rates, showing positive correlation with expert annotator estimates.
- PumpKin provided reliable instantaneous pumping rate dynamics, exceeding human-human agreement in accuracy.
- A basal pumping rate of 0.5 Hz was observed across all tested C. elegans groups off food.
Conclusions:
- PumpKin effectively isolates and estimates single-body part motion during locomotion, overcoming limitations of current tracking methods.
- The technique shows potential for generalization to other systems requiring localized behavior analysis.
- This tool offers a significant advancement for studying feeding behavior and its modulation in C. elegans.
