Abstract
Chemobehavioural phenotypic analysis using small aquatic model organisms is becoming an important toolbox in aquatic ecotoxicology and neuroactive drug discovery. The analysis of the organisms’ behavior is usually performed by combining digital video recording with animal tracking software. This software detects the organisms in the video frames, and reconstructs their movement trajectory using image processing algorithms. In this work we investigated the impact of video file characteristics, video optimization techniques and differences in animal tracking algorithms on the accuracy of quantitative neurobehavioural endpoints. We employed larval stages of a free-swimming euryhaline crustacean Artemia franciscana,commonly used for marine ecotoxicity testing, as a proxy modelto assess the effects of video analytics on quantitative behavioural parameters. We evaluated parameters such as data processing speed, tracking precision, capability to perform high-throughput batch processing of video files. Using a model toxicant the software algorithms were also finally benchmarked against one another. Our data indicates that variability in video file parameters; such as resolution, frame rate, file containers types, codecs and compression levels, can be a source of experimental biases in behavioural analysis. Similarly, the variability in data outputs between different tracking algorithms should be taken into account when designing standardized behavioral experiments and conducting chemobehavioural phenotyping.
Highlights
The acquisition of organism-wide high-dimensional phenotypic data has undergone an evolution to existing molecular omics paradigms over the last decade (Houle, Govindaraju & Omholt, 2010)
For all tested animal-tracking algorithms the computing time was proportionally smaller with an increased digital compression levels (Fig. 1A, Table 1)
Due to intrinsic variability of video file parameters, such as resolution, frame rate, file containers, codecs and compression levels we sought to benchmark if any of those parameters can significantly affect performance and accuracy of animal tracking algorithms
Summary
The acquisition of organism-wide high-dimensional phenotypic data (phenomics) has undergone an evolution to existing molecular omics paradigms over the last decade (Houle, Govindaraju & Omholt, 2010). Manifestations of behavioural phenotypes can transcend multiple levels of biological structure and function; providing causative links between genetic, subcellular and physiological processes (Chapman, 2007; Gerhardt, 2007; Hellou, 2011) This can deliver an overarching strategy to identify functional effects of chemicals with particular biological and/or ecological effects (Faimali et al, 2017; Zhang et al, 2019). It is postulated that highthroughput chemobehavioural phenomics using simplified digital fingerprints of behaviors can become an important toolbox for discovery of novel neuroactive medicines, screening for neurotoxic effects as well as identify specific nerve poisons and their antidotes (Bruni, Lakhani & Kokel, 2014; Kokel & Peterson, 2008)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
