DaphTrack: A deep learning-based multidimensional behavior analysis system for neonatal Daphnia magna

Area of Science:

• Environmental Toxicology
• Aquatic Ecology
• Computational Biology

Background:

• Pollutant monitoring is crucial for ecosystem health assessment.
• Daphnia magna (D. magna) is a sensitive aquatic indicator species used in environmental toxicology.
• Current methods struggle to detect and track neonatal D. magna, limiting behavioral response analysis to contaminants.

Purpose of the Study:

• To develop DaphTrack, a deep learning system for automated detection, counting, and behavioral analysis of neonatal D. magna.
• To establish a comprehensive behavioral parameter matrix for quantifying locomotor characteristics.
• To validate DaphTrack's accuracy using the marine antifouling biocide DCOIT.

Main Methods:

• Utilized the YOLO11n object detection framework and an optimized ByteTrack++ tracking algorithm.
• Developed a behavioral parameter matrix including speed, acceleration, trajectory density, and inter-individual distance.
• Assessed the acute toxic effects of 4,5-Dichloro-2-n-octyl-4-isothiazolin-3-one (DCOIT) on neonatal D. magna behavior.

Main Results:

• Achieved high identification accuracy (96.5%) and robust multi-object tracking (IDF1 = 88.8%) for neonatal D. magna.
• The behavioral parameter matrix effectively quantified locomotor characteristics.
• Demonstrated DaphTrack's capability to assess the toxicological effects of DCOIT on D. magna behavior.

Conclusions:

• DaphTrack significantly improves the accuracy and efficiency of neonatal D. magna identification and behavioral analysis.
• The system provides a novel tool for understanding toxicological mechanisms of aquatic contaminants.
• Offers a new perspective for environmental risk assessment in aquatic ecosystems.