Critical evaluation of commonly used methods to determine the concordance between sonography and magnetic resonance imaging: A comparative study

Introduction: An increasing number of studies investigate the influence of training interventions on muscle thickness (MT) by using ultrasonography. Ultrasonography is stated as a reliable and valid tool to examine muscle morphology. Researches investigating the effects of a training intervention lasting a few weeks need a very precise measurement since increases in MT can be assumed as small. Therefore, the aim of the present work was to investigate the concordance between MT via sonography and muscle cross-sectional area (MCSA) determined via MRI imaging (gold standard) in the calf muscle. Methods: Reliability of sonography measurement and the concordance correlation coefficient, the mean error (ME), mean absolute error (MAE) and the mean absolute percentage error (MAPE) between sonography and MRI were determined. Results: Results show intraclass correlation coefficients (ICC) of 0.88–0.95 and MAPE of 4.63–8.57%. Concordance between MT and MCSA was examined showing ρ = 0.69–0.75 for the medial head and 0.39–0.51 c for the lateral head of the gastrocnemius. A MAPE of 15.88–19.94% between measurements were determined. Based on this, assuming small increases in MT due to training interventions, even with an ICC of 0.95, MAPE shows a high error between two investigators and therefore limited objectivity. Discussion: The high MAPE of 15.88–19.94% as well as CCC of ρc = 0.39–0.75 exhibit that there are significant differences between MRI and sonography. Therefore, data from short term interventions using sonography to detect changes in the MT should be handled with caution.