A YOLO based software for automated detection and analysis of rodent behaviour in the open field arena

Rodent models are important in mechanistic studies of the physiological and pathophysiological determinants behaviour. The Open Field Test (OFT) is one most commonly utilised tests to assess rodent behaviour a novel open environment. key variables assessed an OFT general locomotor activity exploratory behaviours can be manually or by automated systems. Although several systems exist, they often expensive, difficult use, limited type video that analysed. Here we describe machine-learning algorithm – dubbed Cosevare uses trained YOLOv3 DNN identify track movement mice open-field arena. We validated Cosevare's capacity accurately locomotive 10 videos, comparing outputs generated with analysis 5 manual scorers. Behavioural differences between control those diet-induced obesity (DIO) were also documented. found based tracker accurate at identifying tracking within arena instances variable backgrounds. Additionally, kinematic spatial-based demonstrated highly consistent scoring locomotion, centre square duration (CSD) entries (CSE) Automated was able distinguish behavioural healthy DIO mice. study easily mouse field supports machine learning as potential future alternative for assessment animal wide range species differing environments tests. • Freely available algorithms used create software successful test determining observed model obesity.