Development and validation of a semi-automated algorithm to analyze shear wave elastography clips in muscle tissue

Abstract

Abstract Ultrasound shear wave elastography (SWE) is broadly used to quantify muscle stiffness. Currently, most stiffness measures are retrieved from manually placed small measurement zones, which is an operator-dependent and laborious procedure of questionable reliability. Automated time-series measurements over the full visible muscle are expected to improve measurement validity, robustness, and efficiency in larger studies. This study aimed to develop and validate a semi-automated algorithm for analyzing SWE clips of muscle tissue using the single-image, manufacturer-provided manual measurements in every image of the corresponding clips as reference. SWE clips of the relaxed and activated upper trapezius muscle of 52 healthy participants were analyzed manually and with the algorithm for the muscle’s Young’s modulus (kPa) and shear wave velocity (SWV). Results demonstrated excellent correlation between manual and algorithm measurements, Spearman’s ρ > 0.99, p < 0.001. Bland–Altman analyses indicated good method agreement with proportional biases of + 0.747 kPa and − 0.068 m/s for Young’s modulus and SWV, respectively, and widths of the limits of agreement of 8.653 kPa and 0.500 m/s, respectively. The proportional bias is within the minimal detectable change and therefore clinically negligible. These results support the algorithm as a tool enabling valid SWE time-series measurements in muscle tissue and an improved workflow.