Ultrasound-Based assessment of gastrocnemius architecture during locomotion: Analysis of fascicle tracking accuracy along the gait cycle

: Ultrasonography is a widely used technique for assessing muscle morphology by tracking fascicle length and pennation angle during contraction. In the last two decades, numerous automatic methods for fascicle tracking have been developed, but the majority of them were designed for static or well-controlled dynamic contractions, with limited applicability to unconstrained tasks, such as locomotion. The automatic fascicle tracking during these movements poses significant challenges, as out-of-plane movements of the fascicles can compromise their visibility in the bidimensional ultrasound image. In this study, we propose a semi-automatic tracking algorithm specifically designed to track fascicles during unconstrained cyclic movements. The approach integrates a frequency processing pipeline to enhance the visibility of the structures of interest and a supervision module specifically introduced to reduce tracking inaccuracies. We applied this method to ultrasound videos of the medial gastrocnemius in four subjects during walking. We then analyzed potential associations between gait phases and tracking errors relative to manual tracking. The results demonstrated the feasibility of the introduced tracking approach and a tendency of greater error occurrence during the stance-swing transition resulting from a diminished fascicle visibility.