Automated Characterization of Muscle Architectural Variation in Ultrasound Images

Abstract

Fascicle (fiber bundle) orientation and length are useful parameters to infer the force potential of skeletal muscles. Ultrasound imaging is commonly used to determine fascicle characteristics. In most of the literature, automated methods simplify analysis by assimilating fascicles and aponeuroses (connective tissue on which fascicles are anchored) to straight lines that are homogeneously arranged along the muscle length. In practice, manual adjustments are often needed to the realizations of the proposed automated methods due to the low signal-to-noise ratio of the images. We propose a fully automated and robust method that determines non-linear aponeuroses and fascicles over entire panoramic images, reflecting the real structure of the muscle and its fibers.