Machine-Learning Based Automatic and Real-time Detection of Mouse Scratching Behaviors

Abstract

Scratching is a main behavioral response accompanied by acute and chronic itch conditions, and has been quantified as an objective correlate to assess itch in studies using laboratory animals. Scratching has been counted mostly by human annotators, which is a time-consuming and laborious process. It has been attempted to develop automated scoring methods using various strategies, but they often require specialized equipment, costly software, or implantation of device which may disturb animal behaviors. To complement limitations of those methods, we have adapted machine learning-based strategy to develop a novel automated and real-time method detecting mouse scratching from experimental movies captured using monochrome cameras such as a webcam. Scratching is identified by characteristic changes in pixels, body position, and body size by frame as well as the size of body. To build a training model, a novel two-step J48 decision tree-inducing algorithm along with a C4.5 post-pruning algorithm was applied to three 30-min video recordings in which a mouse exhibits scratching following an intradermal injection of a pruritogen, and the resultant frames were then used for the next round of training. The trained method exhibited, on average, a sensitivity and specificity of 95.19% and 92.96%, respectively, in a performance test with five new recordings. This result suggests that it can be used as a non-invasive, automated and objective tool to measure mouse scratching from video recordings captured in general experimental settings, permitting rapid and accurate analysis of scratching for preclinical studies and high throughput drug screening.

Modelling epilepsy in the mouse: challenges and solutions

“…Machine-learning approaches have been used to generate video-based classifiers that can recognise and score a range of rodent behaviours in real time or offline. The repertoire of such detectable behaviours is constantly expanding and currently includes locomotion and climbing ( Bains et al, 2018 ), eating and rearing ( Aniszewska et al, 2014 ), and more subtle behaviours such as scratching ( Park et al, 2019 ) and head bobbing ( Baker, 2011 ). Video-based classifiers for seizures have not yet been developed, but many of the behaviours for which classifiers currently exist – such as rearing, scratching and head bobbing – are seen in rodent seizures ( Kandratavicius et al, 2014 ) and could form the basis for future classifiers.…”

Emerging Methods to Objectively Assess Pruritus in Atopic Dermatitis

“…To mitigate these issues, new tools are under development that offer less intrusive but more comprehensive and automated methods of visual scratching observation. One key area of development is combining infrared video recordings with machine-learning technologies that can be programmed to “learn” scratching movements [20], including unique movements of an individual person. In this way, a machine can take the place of a human observer, offering the benefits of decreased variations in evaluator judgement or visual fatigue, providing more objective and consistent measurements across patients.…”

Variation of the Performance of Machine-Learning Based Image Classifier in Automated Detection of Itch-Induced Scratch

“…This MBIC approach is non-invasive, inexpensive and suitable for objective counting of scratching with overall accuracy similar to or better than existing automated counting methods [11][12][13]. Moreover, it requires neither an attachment nor a surgical implantation of a device to mouse [11].…”

“…To this end, several machine-learning base experimental approaches have been developed to address this issue [8][9][10]. Very recently, Park and his colleagues [11] reported an automated real-time approach using machine learning-based image classifier (MBIC), which can be of practical use for the quantitative analysis of mouse scratching recorded with commercially available video cameras. This MBIC approach is non-invasive, inexpensive and suitable for objective counting of scratching with overall accuracy similar to or better than existing automated counting methods [11][12][13].…”