WhiskEras: A New Algorithm for Accurate Whisker Tracking

Abstract

Rodents engage in active touch using their facial whiskers: they explore their environment by making rapid back-and-forth movements. The fast nature of whisker movements, during which whiskers often cross each other, makes it notoriously difficult to track individual whiskers of the intact whisker field. We present here a novel algorithm, WhiskEras, for tracking of whisker movements in high-speed videos of untrimmed mice, without requiring labeled data. WhiskEras consists of a pipeline of image-processing steps: first, the points that form the whisker centerlines are detected with sub-pixel accuracy. Then, these points are clustered in order to distinguish individual whiskers. Subsequently, the whiskers are parameterized so that a single whisker can be described by four parameters. The last step consists of tracking individual whiskers over time. We describe that WhiskEras performs better than other whisker-tracking algorithms on several metrics. On our four video segments, WhiskEras detected more whiskers per frame than the Biotact Whisker Tracking Tool. The signal-to-noise ratio of the output of WhiskEras was higher than that of Janelia Whisk. As a result, the correlation between reflexive whisker movements and cerebellar Purkinje cell activity appeared to be stronger than previously found using other tracking algorithms. We conclude that WhiskEras facilitates the study of sensorimotor integration by markedly improving the accuracy of whisker tracking in untrimmed mice.

Real-Time Closed-Loop Feedback in Behavioral Time Scales Using DeepLabCut

“…In the last 20 years, multiple research groups have reported increasingly sophisticated, easy-to-use tools for automated annotation of video data of animal behavior ( Knutsen et al, 2005 ; Voigts et al, 2008 ; Perkon et al, 2011 ; Clack et al, 2012 ; Ohayon et al, 2013 ; Giovannucci et al, 2018 ; Dominiak et al, 2019 ; Vanzella et al, 2019 ; Betting et al, 2020 ; Petersen et al, 2020 ). One natural extension of this ability has been to apply these algorithms for on-line, closed-loop paradigms, where changes in behavior of the animal are detected as rapidly as possible, and the behavior is used to modify or manipulate the brain, the virtual environment or the context of behavior.…”

“…One challenge that real-time tracking with video data poses is the computational cost of tracking objects from video frames. Fortunately, the power of processors and algorithms has increased and this has enabled processing of individual frames in <10 ms ( Knutsen et al, 2005 ; Voigts et al, 2008 ; Perkon et al, 2011 ; Clack et al, 2012 ; Ohayon et al, 2013 ; Cao et al, 2018 ; Dominiak et al, 2019 ; Betting et al, 2020 ; Petersen et al, 2020 ). By comparison, little progress has been made to enhance acquisition latency of frames, i.e., the time required to transfer video frames between devices.…”

“…Traditionally, video data have been analyzed manually. More recently, various algorithms have been developed for automating movement detection ( Knutsen et al, 2005 ; Voigts et al, 2008 ; Perkon et al, 2011 ; Clack et al, 2012 ; Ohayon et al, 2013 ; Giovannucci et al, 2018 ; Dominiak et al, 2019 ; Vanzella et al, 2019 ; Betting et al, 2020 ; Petersen et al, 2020 ). With the development of DeepLabCut, a marker-less pose-estimation toolkit based on deep learning ( Mathis et al, 2018 ), computer vision approaches are being used for monitoring poses of animals and for tracking the movement of virtually any part of the body.…”

Purkinje Cell Activity Resonation Generates Rhythmic Behaviors at the Preferred Frequency of 8 Hz

“…The whisker movements were described as the average angle of all trackable whiskers per frame. For a subset of data, we used a new custom written tracking algorithm: WhiskEras [ 48 ], ( , accessed on 12 May 2022).…”

Leptin deficiency-caused behavioral change – A comparative analysis using EthoVision and DeepLabCut

“…However, these advanced analyses are not applicable for many behavioral research laboratories, as they usually lack necessary hardware and needed know-how. To overcome this issue, a growing number of research groups develop methods that enable an analysis of motion and associated-behavior more efficiently, and above all, automatically (Knutsen et al, 2005;Voigts et al, 2008;Perkon et al, 2011;Clack et al, 2012;Ohayon et al, 2013;Giovannucci et al, 2018;Dominiak et al, 2019;Vanzella et al, 2019;Betting et al, 2020;Petersen et al, 2020). Recently, Deep Learning (DL) applications have been used in behavior imaging data analysis.…”