Measuring Dynamic Tendon Torsion Using Ultrasound Speckle Tracking: Validation with Silicone Phantom and In Vivo Application on Human Tibialis Posterior Tendon

Abstract

The torsional characteristics of human tendons are recognized to have functional and clinical relevance, but are underexplored due to the limited in vivo assessment methods available to measure the dynamic torsion characteristics of a tendon during movement. This study aimed to validate the use of transverse plane ultrasound speckle tracking (ST) for measuring dynamic torsion on silicone phantoms, and to evaluate the capability and reliability of ST in measuring dynamic torsion of the human tibialis posterior tendon (TPT) in vivo. Of the ten silicone phantoms tested in the validation study, ST measurement results strongly correlated with the referencing marker tracking method (R² = 0.81-0.95) and had measurement error similar to or smaller than the hypothesized accuracy of 3° (p > 0.045). Subsequently, when ST was applied to nineteen healthy participants' TPT in vivo, it was capable of characterizing the dynamic external torsion of the TPT during 0-20° passive foot pronation. Strong correlations were found between the ST-measured angle and the foot pronation angle (R² = 0.98-0.99), and the test-retest reliability was moderate to good (ICC = 0.73-0.87). These findings suggested that ST is a valid and reliable method for measuring dynamic tendon torsion characteristics.

Keywords: reliability; silicone phantom; speckle tracking; tendon torsion; tibialis posterior tendon; ultrasonography; validity.

Figure 1

Validation of speckle tracking in measuring dynamic torsion angles. (A) Ten cylindrical-shaped silicone phantoms were prepared by mixing mold-making silicone with glass bead powder with a standardized size of 6.0 mm in radius and 60.0 mm in length; (B) The silicone phantoms were mounted on a torsional mechanical testing system that rotates and pulls the phantom simultaneously. Three testing conditions (conditions 1, 2, and 3) were used to induce different magnitudes of dynamic torsion in the phantom; (C) Speckle tracking and the reference marker tracking methods were conducted separately on the same set of phantoms to assess the mean absolute error of speckle-tracking measurements.

Figure 2

Tracking block configurations used in the speckle-tracking algorithm. Five combinations of tracking block size and position were used: large (L1 and L2) and small (S1, S2, and S3). The sizes of large and small tracking blocks were 1/4 and 1/5 the diameter of the phantom’s cross-section, respectively. The locations of the tracking blocks were 2×, 1.5×, or 1× their size away from the center of the phantom.

Figure 3

In vivo dynamic torsion assessment of the tibialis posterior tendon (TPT). (A) A passive foot pronation test was conducted using an isokinetic dynamometer to induce dynamic torsion of the TPT; (B) During the dynamic task, a flat-shaped ultrasound probe was fixated with a silicone probe holder at 1-cm above the medial malleolar to record transverse plane radiofrequency data of the TPT; (C) The dynamic tendon torsion angle was calculated offline in the custom MATLAB ST algorithm. The red area in the ultrasound image represents the region of interest (ROI), and the white box represents the tracking block.

Figure 4

Results of dynamic torsion angle measurements from speckle tracking (ST) (orange lines) and marker tracking (MT) results (blue lines) during the torsional mechanical tests under three T-MTS testing conditions and five tracking block configurations. In each graph, the blue and orange lines denote the average dynamic torsion angle trajectory over time of the 10 silicone phantom samples, measured by the MT and ST methods, respectively. The blue- and orange-shaded areas depict the standard errors of the measurements analyzed at every frame of the ultrasound image recording. The coefficient of determination (R²) and p-value for comparisons between ST and MT measurements are provided for each combination of testing conditions and tracking block configurations.

Figure 5

Correlation between speckle-tracking results and foot pronation angle under five tracking block configurations. In the graph, the line denotes the mean dynamic torsion angle of the tibialis posterior tendon (TPT) in the 19 participants during passive foot pronation. The blue-shaded area depicts the standard deviations of the measurements at each frame during the ultrasound radiofrequency data recording. The coefficient of determination (R²) and p-values for comparisons between the TPT dynamic torsion angle and the foot pronation angle are provided for each tracking block configuration.

Figure 6

Bland–Altman plots evaluating in vivo speckle-tracking test–retest reliability across five tracking block configurations. The scatter dots denote the means and differences in each pair of intra-session measurements, the solid horizontal lines denote the mean differences between the test and retest measurements, the dashed horizontal lines denote the upper and lower limits of agreement (±1.96 times the standard deviation of the measurement differences), and the black lines denote the regression fit of the differences on the means. The systemic biases of the measurements were 2.2–3.6° across all tracking block configurations. No proportional biases were shown by correlation analyses between mean and differences in measurement values (p > 0.05). The correlation coefficients and p-values for the relationship between the mean difference in test and retest measurements (vertical axis) and the mean magnitude of measurements (horizontal axis) are provided for each tracking block configuration.

Figure 7

Illustration of the suggested tibialis posterior tendon (TPT, highlighted in green) external dynamic torsion (white arrow) during foot pronation (black arrow). (Source of 3D models: BodyParts3D (Data version 4.3), © The Database Center for Life Science licensed under CC Attribution-Share Alike 2.1 Japan.).

Figure 8

Illustration of the potential mechanical loadings applied to the tendon during joint movement. During normal joint movement, tendons may be subjected to (A) tensile, (B) torsional, (C) shearing, and (D) bending loads, resulting in stretching, twisting, shearing, or curving deformations, respectively. Examples of the deformed tendons are presented in yellow. (A) Under tensile load, lateral displacement or strain can be observed from the longitudinal plane. (B) Under torsional load, rotation of the tendon’s cross-section can be observed from the transverse plane. (C) Under shearing load, intratendinous sliding or non-uniform displacements can be observed from the longitudinal plane. (D) Under bending load, stretching and compressive strain can be observed from the longitudinal plane, and a translation motion (potentially combined with an out-of-plane motion) can be observed from the transverse plane.