ABSTRACT
Ultrasound devices are common in muscle physiology laboratories due to their ease of use and validity to assess skeletal muscle characteristics. The current study assessed the reliability of ultrasound skeletal muscle image analysis across multiple raters with limited experience. Vastus lateralis (VL), rectus femoris (RF), and first dorsal interosseus (FDI) images were separately analyzed by three novice raters to determine muscle thickness (MT), cross-sectional area (CSA), and echo-intensity (EI). Separate analyses of variance (ANOVA) assessed statistical differences between and within raters. Intra-class correlation coefficients (ICC) between (inter-rater) and within (intra-rater) raters, the standard error of the measurement (SEM) and minimal difference needed to be considered real were calculated. Inter-rater reliability was high for the VL and RF (ICC: 0.984-0.999), while the FDI was lower (0.614-0.962). Further, intra-rater reliability was greater than 0.961 for each rater. SEM values calculated for inter-rater reliability expressed as a percentage of the mean ranged from 0.4-5.8% across variables. Similarly, SEM values for intra-rater reliability were between 0.8-5.8%, 0.6-3.6%, and 0.4-3.2% for Raters 1, 2 and 3, respectively. Despite this, significant differences (p<0.05) between raters were observed for RF MT and EI, VL CSA and EI, and FDI MT, suggesting that potentially more measurement trials or greater practice time may be necessary to reduce systematic error among multiple raters. Post-image acquisition processing is reliable among and within raters as determined through ICCs and SEMs. This study provided consistent results among three separate novice raters given the same training, a unique yet realistic setting in muscle physiology laboratories.
