Accurate Tracking of Locomotory Kinematics in Mice Moving Freely in Three-Dimensional Environments.

Abstract

Marker-based motion capture (MBMC) is a powerful tool for precise, high-speed, three-dimensional tracking of animal movements, enabling detailed study of behaviors ranging from subtle limb trajectories to broad spatial exploration. Despite its proven utility in larger animals, MBMC has remained underutilized in mice due to the difficulty of robust marker attachment during unrestricted behavior. In response to this challenge, markerless tracking methods, facilitated by machine learning, have become the standard in small animal studies due to their simpler experimental setup. However, trajectories obtained with markerless approaches at best approximate ground-truth kinematics, with accuracy strongly dependent on video resolution, training dataset quality, and computational resources for data processing. Here, we overcome the primary limitation of MBMC in mice by implanting minimally invasive markers that remain securely attached over weeks of recordings. This technique produces high-resolution, artifact-free trajectories, eliminating the need for extensive post-processing. We demonstrate the advantages of MBMC by resolving subtle drug-induced kinematic changes that become apparent only within specific behavioral contexts, necessitating precise three-dimensional tracking beyond simple flat-surface locomotion. Furthermore, MBMC uniquely captures the detailed spatiotemporal dynamics of harmaline-induced tremors, revealing previously inaccessible correlations between body parts and thus significantly improving the translational value of preclinical tremor models. While markerless tracking remains optimal for many behavioral neuroscience studies in which general posture estimation suffices, MBMC removes barriers to investigations demanding greater precision, reliability, and low-noise trajectories. This capability significantly broadens the scope for inquiry into the neuroscience of movement and related fields.