A low-cost system for hindlimb kinematic analysis of rodents performing robotic treadmill training

Thesis (M.S.) California State University, Los Angeles, 2012Committee members: Deborah S Won, Frank Ashrafi, Juan Castaneda, Fred DaneshgaranAutomated Detection, TrackingAnalysis of step kinematics in spinally injured rodents while performing robotic treadmill training is essential for developing a new therapy for rehabilitation of walking in spinal cord injured patients. Major challenges to realizing such an automated video-based system include the limitations of image quality from conventional video cameras not specialized for motion capture and the increased complexity of the computational algorithm required to make it automated. Our system can analyze data much faster and with much more ease for the user than the manual tracking system. The processing time was reduced from 5-9 hours to ~1 hour for the analysis of a 30-second video file captured at 30 frames per second while achieving the same accuracy of detection. The ease of use and speed of processing of our system has allowed us to perform joint angle analysis on a novel rehabilitation therapy for spinal cord injury at low cost