A Low-Cost, Autonomous Gait Detection and Estimation System for Analyzing Gait Impairments in Mice

With the advancement in imaging technology, many commercial systems have been developed for performing motion analysis in mice. However, available commercial systems are expensive and use proprietary software. In this paper, we describe a low-cost, camera-based design of an autonomous gait acquisition and analysis system for inspecting gait deficits in C57BL/6 mice. Our system includes video acquisition, autonomous gait-event detection, gait-parameter extraction, and result visualization. We provide a simple, user-friendly, step-by-step detailed methodology to apply well-known image processing techniques for detecting mice footfalls and calculating various gait parameters for analyzing gait abnormalities in healthy and neurotraumatic mice. The system was used in a live animal study for assessing recovery in a mouse model of Parkinson's disease. Using the videos acquired in the study, we validate the performance of our system with receiver operating characteristic (ROC) and Hit : Miss : False (H : M : F) detection analyses. Our system correctly detected the mice footfalls with an average H : M : F score of 92.1 : 2.3 : 5.6. The values for the area under an ROC curve for all the ROC plots are above 0.95, which indicates an almost perfect detection model. The ROC and H : M : F analyses show that our system produces accurate gait detection. The results observed from the gait assessment study are in agreement with the known literature. This demonstrates the practical viability of our system as a gait analysis tool.