Creating a Pose-Estimation System for Use in Studying Motor Cortical Dynamics

Abstract

Currently, researchers attempt to better understand motor control by using stereotyped tasks in rodents. Using this method, researchers can correlate rat movements with activity occurring in the motor cortex. Our lab, specifically, uses a rodent paradigm where rats perform a supination task while simultaneously having population-level neuronal spiking activity recorded by a tetrode. This allows researchers to infer dynamics from stereotyped tasks but not from naturalistic tasks. In addition, using readings from a knob leaves room for error in recording. This project is an attempt to mitigate these issues with the introduction of a system that can quantify rat movement captured by two synchronous cameras. The system will use the novel pose-estimation software DeepLabCut to tag individual body parts, then find their trajectories, and finally piece together a 3D image of rat movement.