The assessment of motor function recovery in spinal cord injury (SCI) mouse models traditionally relies on semi-quantitative methods like the Basso Mouse Scale (BMS), which are prone to inter-observer variability and are labor-intensive. To address these limitations, this study aimed to develop and validate novel, objective kinematic metrics for a more precise and automated evaluation of motor function. Using the DeepLabCut software package to analyze locomotion videos of SCI mice, we applied Permutation Feature Importance (PFI) to identify key joints associated with motor recovery. Based on this data-driven approach, we engineered four new indicators: max horizon-hip-hindpaw angular velocity to quantify hindlimb movement speed and range, average horizon-hindpaw-hindpaw tip angle to assess foot posture (dorsal versus plantar contact), neck swing times to evaluate trunk stability at later recovery stages, and step difference to measure forelimb-hindlimb coordination. Our results demonstrate that these novel metrics exhibit strong correlations with manual BMS scores and capture distinct aspects of motor recovery. Furthermore, integrating these indicators into a Random Forest regression model improved the sensitivity of automatic assessment of BMS compared with models using only previously established metrics. In conclusion, these novel kinematic indicators provide a more objective, sensitive, and efficient framework for assessing motor function in SCI mice, thereby enhancing experimental reproducibility and offering a powerful tool for preclinical research.
Nong et al. (Tue,) studied this question.