Further reading
RodentFlow deliberately uses a classic tracker: one animal, body center, no training (why). When an experiment needs more, such as body parts, several interacting animals, or automatic classification of behaviors like grooming or rearing, the tools below are the current open-source standard. Their per-animal outputs can be imported into RodentFlow for statistics and figures (how).
Overview
- Luxem K, Sun JJ, Bradley SP, et al. Open-source tools for behavioral video analysis: setup, methods, and best practices. eLife 12, e79305 (2023). doi:10.7554/eLife.79305 A practical review covering recording setups, pose estimation, behavior quantification, and how to validate and report automated analyses. A good first read before choosing any tool.
Pose estimation: tracking body parts
- Pereira TD, Tabris N, Matsliah A, et al. SLEAP: A deep learning system for multi-animal pose tracking. Nature Methods 19 (2022). doi:10.1038/s41592-022-01426-1
- Lauer J, Zhou M, Ye S, et al. Multi-animal pose estimation, identification and tracking with DeepLabCut. Nature Methods 19 (2022). doi:10.1038/s41592-022-01443-0
- Ye S, Filippova A, Lauer J, et al. SuperAnimal pretrained pose estimation models for behavioral analysis. Nature Communications 15, 5165 (2024). doi:10.1038/s41467-024-48792-2 Pretrained models, including a top-view mouse model, that can be used with little or no labeling of your own frames.
From poses to behaviors
- Weinreb C, Pearl JE, Lin S, et al. Keypoint-MoSeq: parsing behavior by linking point tracking to pose dynamics. Nature Methods 21 (2024). doi:10.1038/s41592-024-02318-2 Unsupervised discovery of behavioral "syllables" from keypoint tracking.
- Goodwin NL, Choong JJ, Hwang S, et al. Simple Behavioral Analysis (SimBA) as a platform for explainable machine learning in behavioral neuroscience. Nature Neuroscience 27 (2024). doi:10.1038/s41593-024-01649-9 Supervised classifiers for user-defined behaviors, with explanations of what drives each classification.
How these relate to RodentFlow
| Need | Use |
|---|---|
| Zone times, entries, distance, latencies, freezing for one animal, on the phone | RodentFlow alone |
| Nose-based exploration, head direction, rearing, gait from video | A pose estimator (SLEAP, DeepLabCut), then import per-animal values |
| Several animals in one arena (social tests) | A multi-animal pose tracker |
| Automatic labels for grooming, attack, mounting … | SimBA or Keypoint-MoSeq on top of pose tracking |
| Group statistics and figures for any of the above | Import into RodentFlow's Groups tab |
This list is deliberately short and limited to work published from 2022 on.