Scratch-AID: a deep-learning based system for automatic detection of mouse scratching behavior with high accuracy
- University of Pennsylvania, United States
- University of California, Davis, United States
- Howard Hughes Medical Institute, Harvard Medical School, United States
- Massachusetts Institute of Technology, United States
Abstract
Mice are the most commonly used model animals for itch research and for development of anti-itch drugs. Most labs manually quantify mouse scratching behavior to assess itch intensity. This process is labor-intensive and limits large-scale genetic or drug screenings. In this study, we developed a new system, Scratch-AID Automatic Itch Detection), which could automatically identify and quantify mouse scratching behavior with high accuracy. Our system included a custom-designed videotaping box to ensure high-quality and replicable mouse behavior recording and a convolutional recurrent neural network (CRNN) trained with frame-labeled mouse scratching behavior videos, induced by nape injection of chloroquine (CQ). The best trained network achieved 97.6% recall and 96.9% precision on previously unseen test videos. Remarkably, Scratch-AID could reliably identify scratching behavior in other major mouse itch models, including the acute cheek model, the histaminergic model, and a chronic itch model. Moreover, our system detected significant differences in scratching behavior between control and mice treated with an anti-itch drug. Taken together, we have established a novel deep learning-based system that is ready to replace manual quantification for mouse scratching behavior in different itch models and for drug screening.
Data availability
The training and test videos generated during the current study can be downloaded from DRYAD (https://doi.org/10.5061/dryad.mw6m9060s). The codes for model training and test can be downloaded from GitHub (https://github.com/taimeimiaole/Scratch-AID)
-
Data From: Scratch-AID: a deep-learning based system for automatic detection of mouse scratching behavior with high accuracyDryad Digital Repository, doi:10.5061/dryad.mw6m9060s.
Article and author information
Author details
Funding
National Science Foundation (DMS-1854770)
- Javier Arsuaga
National Institutes of Health (R01 NS083702)
- Wenqin Luo
National Institutes of Health (R34 NS118411)
- Long Ding
- Wenqin Luo
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
Ethics
Animal experimentation: Mice were housed in the John Morgan animal facility at the University of Pennsylvania. All animal treatments were conducted in accordance with protocols approved by the Institutional Animal Care and Use Committee and the guidelines of the National Institutes of Health (Protocol No. 804886).
Copyright
© 2022, Yu et al.
This article is distributed under the terms of the Creative Commons Attribution License permitting unrestricted use and redistribution provided that the original author and source are credited.
Metrics
-
- 3,555
- views
-
- 381
- downloads
-
- 3
- citations
Views, downloads and citations are aggregated across all versions of this paper published by eLife.
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Scratch-AID: a deep-learning based system for automatic detection of mouse scratching behavior with high accuracy
eLife 11:e84042.
https://doi.org/10.7554/eLife.84042
Further reading
-
- Neuroscience
Insulin plays a key role in metabolic homeostasis. Drosophila insulin-producing cells (IPCs) are functional analogues of mammalian pancreatic beta cells and release insulin directly into circulation. To investigate the in vivo dynamics of IPC activity, we quantified the effects of nutritional and internal state changes on IPCs using electrophysiological recordings. We found that the nutritional state strongly modulates IPC activity. IPC activity decreased with increasing periods of starvation. Refeeding flies with glucose or fructose, two nutritive sugars, significantly increased IPC activity, whereas non-nutritive sugars had no effect. In contrast to feeding, glucose perfusion did not affect IPC activity. This was reminiscent of the mammalian incretin effect, where glucose ingestion drives higher insulin release than intravenous application. Contrary to IPCs, Diuretic hormone 44-expressing neurons in the pars intercerebralis (DH44PINs) responded to glucose perfusion. Functional connectivity experiments demonstrated that these DH44PINs do not affect IPC activity, while other DH44Ns inhibit them. Hence, populations of autonomously and systemically sugar-sensing neurons work in parallel to maintain metabolic homeostasis. Accordingly, activating IPCs had a small, satiety-like effect on food-searching behavior and reduced starvation-induced hyperactivity, whereas activating DH44Ns strongly increased hyperactivity. Taken together, we demonstrate that IPCs and DH44Ns are an integral part of a modulatory network that orchestrates glucose homeostasis and adaptive behavior in response to shifts in the metabolic state.
-
- Neuroscience
The mushroom body (MB) is the center for associative learning in insects. In Drosophila, intersectional split-GAL4 drivers and electron microscopy (EM) connectomes have laid the foundation for precise interrogation of the MB neural circuits. However, investigation of many cell types upstream and downstream of the MB has been hindered due to lack of specific driver lines. Here we describe a new collection of over 800 split-GAL4 and split-LexA drivers that cover approximately 300 cell types, including sugar sensory neurons, putative nociceptive ascending neurons, olfactory and thermo-/hygro-sensory projection neurons, interneurons connected with the MB-extrinsic neurons, and various other cell types. We characterized activation phenotypes for a subset of these lines and identified a sugar sensory neuron line most suitable for reward substitution. Leveraging the thousands of confocal microscopy images associated with the collection, we analyzed neuronal morphological stereotypy and discovered that one set of mushroom body output neurons, MBON08/MBON09, exhibits striking individuality and asymmetry across animals. In conjunction with the EM connectome maps, the driver lines reported here offer a powerful resource for functional dissection of neural circuits for associative learning in adult Drosophila.
