Rapid, open-source, and automated quantification of the head twitch response in C57BL/6J mice using DeepLabCut and Simple Behavioral Analysis
bioRxiv Preprint Server April 28, 2025 Alexander D. Maitland, Nicholas R. Gonzalez, Donna Walther et al. 1 citation preprint
A new automated method using open-source machine learning toolkits, DeepLabCut and SimBA, accurately quantifies the head twitch response (HTR) in mice from experimental videos. The approach, trained and validated on videos of C57BL/6J mice given various psychedelic drugs, performed best at 50% video resolution and 120 frames per second (precision 95.45%, recall 95.56%, F1 score 95.51%) and also worked well at lower frame rates. When applied to bufotenine, a tryptamine derivative, elevated HTRs occurred only after blocking serotonin 1A receptors (ED50 = 0.99 mg/kg, max counts = 24). HTR counts from the automated method strongly correlated with visual scoring and semi-automated software (r = 0.98–0.99). The method offers a modular, noninvasive, open-source alternative to existing techniques.