Tracking the movement of the muscle-tendon junction is important for quantifying how muscles and ... more Tracking the movement of the muscle-tendon junction is important for quantifying how muscles and tendons drive human movement. However, this remains an arduous task. Deep learning methods have recently shown promise for tracking, but there are limits to how well these networks generalize to novels subjects and tasks. Muscle-tendon junction movement is dependent upon muscle activation and joint-level motion. Therefore, we sought to determine if supplementing current networks with ankle angle and electromyographical (EMG) data would improve tracking. We observed a 20% increase in tracking accuracy for novel subjects when we incorporated EMG and ankle angle data into a current deep-learning network. Our results indicate that extending current methods to include data beyond the ultrasound image can significantly improve their accuracy.
Tracking the movement of the muscle-tendon junction is important for quantifying how muscles and ... more Tracking the movement of the muscle-tendon junction is important for quantifying how muscles and tendons drive human movement. However, this remains an arduous task. Deep learning methods have recently shown promise for tracking, but there are limits to how well these networks generalize to novels subjects and tasks. Muscle-tendon junction movement is dependent upon muscle activation and joint-level motion. Therefore, we sought to determine if supplementing current networks with ankle angle and electromyographical (EMG) data would improve tracking. We observed a 20% increase in tracking accuracy for novel subjects when we incorporated EMG and ankle angle data into a current deep-learning network. Our results indicate that extending current methods to include data beyond the ultrasound image can significantly improve their accuracy.
Uploads
Papers by Hassam Uddin