Test-Retest Reliability and Agreement of an Ultrasound Guide System

Brightness-mode (B-mode) ultrasound is a popular non-invasive imaging modality used to assess skeletal muscle size. While magnetic resonance imaging (MRI) remains the gold standard for measuring muscle size and volume. However, the reliability and agreement of different B-mode ultrasound imaging techniques in conjunction with the various muscle volume estimation equations has not yet been studied. This study aimed to assess the test-retest reliability of a custom-built probe guide (PGT) and traditional (TRAD) ultrasound methods to assess muscle volume using four volume estimation equations. We also aimed to evaluate the slice-by-slice agreement and the agreement within the most accurate (i.e., “best version”) of four different muscle volume estimation equations. This study consisted of forty recreationally active young adults (55% female; 21  3 years, 24.54  4.21 kg/m2) completing one visit to the laboratory. Participants underwent non-invasive ultrasound imaging on their vastus lateralis (VL) across four separate image acquisition sessions, each separated by 10 minutes. Two trials utilized the PGT while the other two used TRAD methods. The cross-sectional area (CSA) from the scans were quantified using an open-source imaging software. Muscle volume was estimated using the truncated cone, Cavalieri, cubic spline interpolation, and DPSO equations. Test-retest reliability statistics were calculated between trials for PGT and TRAD for all four muscle volume estimation equations. Excellent reliability was displayed across trials 1 and 2 for PGT and TRAD for all four muscle volume estimation equations (ICC2,1 = 0.956 - 0.987, SEM% = 2.02 – 5.34%, MD = 3.96 – 10.50 cm3). For the 7-site muscle volume estimations, the slice-by-slice agreement within the four muscle volume estimation equations displayed that as slice number increases error and bias decrease. When using 6 slices, agreement was the best for the truncated cone (SEE = 5.03, TE = 13.59, r = 0.997), Cavalieri (SEE = 4.85, TE = 12.57, r = 0.997), and cubic spline interpolation (SEE = 4.57, TE = 5.01, r = 0.999). The DPSO method displayed excellent agreement (SEE = 4.59, TE = 4.70, r = 0.999) at slice 5 with the lowest amount of bias (-0.96 cm3). The agreement for the “best version” within the 7-site muscle volume estimation equations, the cubic spline interpolation and DPSO displayed good agreement (SEE = 9.30, TE = 27.02, r = 0.997) with a reduced amount of bias (-25.37 cm3, LOA: -43.60 - -7.14). While the Cavalieri and truncated cone also displayed relatively good agreement (SEE = 4.16, TE = 17.97, r = 0.999) with a low amount of bias (17.48 cm3, LOA: 9.30 - 25.63). Overall, the PGT and TRAD methods exhibit excellent reliability for whole muscle volume estimation across all four equations. Equations that incorporate adjustments for missing CSA values (i.e., cubic spline interpolation and DPSO) displayed the best overall agreement for 7-site muscle volume and slice-by-slice analyses.