DeepEthogram, a machine learning pipeline for supervised behavior classification from raw pixels
- Harvard Medical School, United States
- Boston Children's Hospital, United States
- Massachusetts General Hospital, United States
- Champalimaud Center for the Unknown, Portugal
Abstract
Videos of animal behavior are used to quantify researcher-defined behaviors-of-interest to study neural function, gene mutations, and pharmacological therapies. Behaviors-of-interest are often scored manually, which is time-consuming, limited to few behaviors, and variable across researchers. We created DeepEthogram: software that uses supervised machine learning to convert raw video pixels into an ethogram, the behaviors-of-interest present in each video frame. DeepEthogram is designed to be general-purpose and applicable across species, behaviors, and video-recording hardware. It uses convolutional neural networks to compute motion, extract features from motion and images, and classify features into behaviors. Behaviors are classified with above 90% accuracy on single frames in videos of mice and flies, matching expert-level human performance. DeepEthogram accurately predicts rare behaviors, requires little training data, and generalizes across subjects. A graphical interface allows beginning-to-end analysis without end-user programming. DeepEthogram's rapid, automatic, and reproducible labeling of researcher-defined behaviors-of-interest may accelerate and enhance supervised behavior analysis.
Data availability
Code is posted publicly on Github and linked in the paper. Video datasets and human annotations are publicly available and linked in the paper.
Article and author information
Author details
Christopher D Harvey
Funding
National Institutes of Health (R01MH107620)
- Christopher D Harvey
National Science Foundation (GRFP)
- Nivanthika K Wimalasena
Fundacao para a Ciencia ea Tecnologia (PD/BD/105947/2014)
- Tomás Cruz
Harvard Medical School Dean's Innovation Award
- Christopher D Harvey
Harvard Medical School Goldenson Research Award
- Christopher D Harvey
National Institutes of Health (DP1 MH125776)
- Christopher D Harvey
National Institutes of Health (R01NS089521)
- Christopher D Harvey
National Institutes of Health (R01NS108410)
- Christopher D Harvey
National Institutes of Health (F31NS108450)
- James P Bohnslav
National Institutes of Health (R35NS105076)
- Clifford J Woolf
National Institutes of Health (R01AT011447)
- Clifford J Woolf
National Institutes of Health (R00NS101057)
- Lauren L Orefice
National Institutes of Health (K99DE028360)
- David A Yarmolinsky
European Research Council (ERC-Stg-759782)
- M Eugenia Chiappe
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
Ethics
Animal experimentation: All experimental procedures were approved by the Institutional Animal Care and Use Committees at Boston Children's Hospital (protocol numbers 17-06-3494R and 19-01-3809R) or Massachusetts General Hospital (protocol number 2018N000219) and were performed in compliance with the Guide for the Care and Use of Laboratory Animals.
Copyright
© 2021, Bohnslav 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
-
- 13,891
- views
-
- 1,318
- downloads
-
- 110
- 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)
DeepEthogram, a machine learning pipeline for supervised behavior classification from raw pixels
eLife 10:e63377.
https://doi.org/10.7554/eLife.63377
Further reading
-
- Neuroscience
Quantitative information about synaptic transmission is key to our understanding of neural function. Spontaneously occurring synaptic events carry fundamental information about synaptic function and plasticity. However, their stochastic nature and low signal-to-noise ratio present major challenges for the reliable and consistent analysis. Here, we introduce miniML, a supervised deep learning-based method for accurate classification and automated detection of spontaneous synaptic events. Comparative analysis using simulated ground-truth data shows that miniML outperforms existing event analysis methods in terms of both precision and recall. miniML enables precise detection and quantification of synaptic events in electrophysiological recordings. We demonstrate that the deep learning approach generalizes easily to diverse synaptic preparations, different electrophysiological and optical recording techniques, and across animal species. miniML provides not only a comprehensive and robust framework for automated, reliable, and standardized analysis of synaptic events, but also opens new avenues for high-throughput investigations of neural function and dysfunction.
-
- Neuroscience
Mesolimbic dopamine encoding of non-contingent rewards and reward-predictive cues has been well established. Considerable debate remains over how mesolimbic dopamine responds to aversion and in the context of aversive conditioning. Inconsistencies may arise from the use of aversive stimuli that are transduced along different neural paths relative to reward or the conflation of responses to avoidance and aversion. Here, we made intraoral infusions of sucrose and measured how dopamine and behavioral responses varied to the changing valence of sucrose. Pairing intraoral sucrose with malaise via injection of lithium chloride (LiCl) caused the development of a conditioned taste aversion (CTA), which rendered the typically rewarding taste of sucrose aversive upon subsequent re-exposure. Following CTA formation, intraoral sucrose suppressed the activity of ventral tegmental area dopamine neurons (VTADA) and nucleus accumbens (NAc) dopamine release. This pattern of dopamine signaling after CTA is similar to intraoral infusions of innately aversive quinine and contrasts with responses to sucrose when it was novel or not paired with LiCl. Dopamine responses were negatively correlated with behavioral reactivity to intraoral sucrose and predicted home cage sucrose preference. Further, dopamine responses scaled with the strength of the CTA, which was increased by repeated LiCl pairings and weakened through extinction. Thus, the findings demonstrate differential dopamine encoding of the same taste stimulus according to its valence, which is aligned to distinct behavioral responses.
