DeepFly3D, a deep learning-based approach for 3D limb and appendage tracking in tethered, adult Drosophila
- EPFL, Switzerland
Abstract
Studying how neural circuits orchestrate limbed behaviors requires the precise measurement of the positions of each appendage in 3-dimensional (3D) space. Deep neural networks can estimate 2-dimensional (2D) pose in freely behaving and tethered animals. However, the unique challenges associated with transforming these 2D measurements into reliable and precise 3D poses have not been addressed for small animals including the fly, Drosophila melanogaster. Here we present DeepFly3D, a software that infers the 3D pose of tethered, adult Drosophila using multiple camera images. DeepFly3D does not require manual calibration, uses pictorial structures to automatically detect and correct pose estimation errors, and uses active learning to iteratively improve performance. We demonstrate more accurate unsupervised behavioral embedding using 3D joint angles rather than commonly used 2D pose data. Thus, DeepFly3D enables the automated acquisition of Drosophila behavioral measurements at an unprecedented level of detail for a variety of biological applications.
Data availability
All data generated and analyzed during this study are included in the DeepFly3D GitHub site: https://github.com/NeLy-EPFL/DeepFly3D and in the Harvard Dataverse.
-
aDN-GAL4 UAS-CsChrimsonHarvard Dataverse, doi:10.7910/DVN/S4L4KX.
-
MDN-GAL4 UAS-CsChrimsonHarvard Dataverse, doi:10.7910/DVN/8SUC9U.
Article and author information
Author details
Funding
Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung (175667)
- Daniel Morales
- Pavan Ramdya
Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung (181239)
- Daniel Morales
- Pavan Ramdya
EPFL (iPhD)
- Semih Günel
Microsoft Research (JRC Project)
- Helge Rhodin
Swiss Government Excellence Postdoctoral Scholarship (2018.0483)
- Daniel Morales
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
Copyright
© 2019, Günel 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
-
- 8,094
- views
-
- 888
- downloads
-
- 130
- 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)
DeepFly3D, a deep learning-based approach for 3D limb and appendage tracking in tethered, adult Drosophila
eLife 8:e48571.
https://doi.org/10.7554/eLife.48571
Further reading
-
- Neuroscience
Significant technical challenges exist when measuring synaptic connections between neurons in living brain tissue. The patch clamping technique, when used to probe for synaptic connections, is manually laborious and time-consuming. To improve its efficiency, we pursued another approach: instead of retracting all patch clamping electrodes after each recording attempt, we cleaned just one of them and reused it to obtain another recording while maintaining the others. With one new patch clamp recording attempt, many new connections can be probed. By placing one pipette in front of the others in this way, one can ‘walk’ across the mouse brain slice, termed ‘patch-walking.’ We performed 136 patch clamp attempts for two pipettes, achieving 71 successful whole cell recordings (52.2%). Of these, we probed 29 pairs (i.e. 58 bidirectional probed connections) averaging 91 μm intersomatic distance, finding three connections. Patch-walking yields 80–92% more probed connections, for experiments with 10–100 cells than the traditional synaptic connection searching method.
-
- Neuroscience
Identical stimuli can be perceived or go unnoticed across successive presentations, producing divergent behavioral outcomes despite similarities in sensory input. We sought to understand how fluctuations in behavioral state and cortical layer and cell class-specific neural activity underlie this perceptual variability. We analyzed physiological measurements of state and laminar electrophysiological activity in visual area V4 while monkeys were rewarded for correctly reporting a stimulus change at perceptual threshold. Hit trials were characterized by a behavioral state with heightened arousal, greater eye position stability, and enhanced decoding performance of stimulus identity from neural activity. Target stimuli evoked stronger responses in V4 in hit trials, and excitatory neurons in the superficial layers, the primary feed-forward output of the cortical column, exhibited lower variability. Feed-forward interlaminar population correlations were stronger on hits. Hit trials were further characterized by greater synchrony between the output layers of the cortex during spontaneous activity, while the stimulus-evoked period showed elevated synchrony in the feed-forward pathway. Taken together, these results suggest that a state of elevated arousal and stable retinal images allow enhanced processing of sensory stimuli, which contributes to hits at perceptual threshold.
