Tendon length estimates are influenced by tracking location

A convenience sample of 22 participants (4 females, n = 22, 42 ± 9 years, 177 ± 9 cm, 79 ± 10 kg) gave written consent to voluntarily participate in this study (NoARK, trial registration: NCT03704532). Each individual had experienced an acute Achilles tendon rupture in one of their legs (injured leg, treated non-operatively according to Reito et al. 2018) approximately 1 year (13 ± 2 months) before data collection, while they had no known musculoskeletal problems in the non-injured lower limb. The study protocol was approved by the ethics committee of the Central Finland Hospital District (Approval number: 2U/2018).

Data were collected within one session from both legs. In all cases, the non-injured leg was tested first. While participants lay prone with the foot hanging relaxed over the edge of a plinth, MG tendon resting length was measured using a validated ultrasound-assisted method (Barfod et al. 2015; Intziegianni et al. 2015). Then, in an ankle dynamometer participants performed a warm-up consisting of several submaximal isometric plantar flexion contractions (⁓ 5 min), followed by assessment of maximal voluntary isometric torque (MVC). Participants were instructed to reach the maximum torque in ⁓ 2–3 s, and then to return to the zero level within 2–3 s, resulting in a pyramid-shaped torque curve. Verbal encouragement was given to ensure maximal effort. Slow ramp contractions were then performed to 10%, 20% and 30% of MVC, whereby participants reached the desired torque level in ⁓ 2 s and then returned to the zero level in ⁓ 2 s. For both legs, the target percentages were calculated from the MVC torque of the non-injured leg. During contractions, MG MTJ displacement was recorded with US. For each condition, two successful trials (i.e. target torque level was reached within ± 1%) were recorded and analysed, and the trial showing the largest MTJ displacement was chosen for the final analysis.

Torque signals were sampled via a 16-bit A/D-board (Power 1401, Cambridge Electronic Design, Cambridge, UK) connected to a personal computer. Digital signals were visualised and recorded in Spike2 Software (Cambridge Electronic Design, Cambridge, UK). To synchronise data acquisition, at the start of each recording, a signal was sent from Spike2 Software to start the video recording in the US device.

MG tendon resting length The tendon insertion on the calcaneus and the most distal point of the MG MTJ were located with US and marked on the skin. The most distal site of MTJ was identified by scanning the area using both longitudinal and transverse views. Both the tendon insertion and origin locations were similarly visible in the non-injured and injured limbs. The distance between these marks, following the curvature of the leg, was then measured with a measuring tape (Barfod et al. 2015). Day-to-day reliability of the tendon length estimates of two assessors was excellent in terms of intraclass-correlation (0.99, 90% CI 0.95–0.99) and standard error of mean (0.17 cm). These reliability tests were done for 20 individuals 2–4 times, with 2–6 days between measurements.

MG MTJ displacement tracking The US probe was attached over the identified MTJ so that during contraction, the area remained visible within the field of view. Correct placement and orientation of the probe was confirmed by observing the image visually during contractions. The position of the MG MTJ was tracked using Tracker 5.1.3 Software (https://​physlets.​org/​tracker/​) from two locations: (1) the exact MTJ (MTJ condition); and (2) from an insertion point of a clearly visible fascicle on the aponeurosis 1–3 cm proximal to the MTJ (FAS condition). We used a 4 cm calibration bar and an oval-shaped feature template (0.3–1 cm in length, 0.25–0.50 cm in height; Fig. 1). The FAS was tracked 1.3 ± 0.6 cm proximal to MTJ. This value was added to the MG tendon resting length when calculating tendon strain in FAS condition. Tracking accuracy was visually confirmed for each trial. In case visually perfect accuracy was not reached, the tracking was redone. An example of tracking process is shown in the Supplementary video. The person performing the tracking was blinded to the limb and condition. Reliability of tracking relies on the anatomical location and the template features chosen. We checked that the differences in maximum displacement between two analysers was < 0.1 cm when tracking FAS with the above instruction. Within-analyser reliability had a typical error of 0.06 cm and intraclass correlation of 0.967 [90% CI 0.705–0.997]. Displacement data were smoothed using a 0.14 s moving average.

Maximum proximo-distal MTJ and FAS displacements were extracted from each condition and participant. MG tendon strain was then calculated as \(\varepsilon = \frac{\Delta L}{{L0}} \times 100\), where ΔL is the measured displacement and L0 is the resting length. Different resting lengths were used for the MTJ and FAS conditions. Some videos could not be analysed due to video quality. The final sample size for each condition is shown in Table 1.

Statistical analysis was performed using IBM SPSS Software (IBM New York, NY, US). Two-way repeated measures analysis of variance was performed for each leg to assess the interaction between tracking location (MTJ, FAS) and contraction intensity (10, 20, 30, and 100% MVC). Huynh–Feldt correction was applied in case of violated sphericity. In case of an interaction, one-way repeated measures ANOVAs were performed for each variable. Pairwise comparisons were performed after Bonferroni correction. Pairwise t-tests were used to characterize differences in selected variables between limbs, as well as between FAS and MTJ (Table 1). Statistical significance level was set at α = 0.05 in all cases. Values are reported as mean ± standard deviation in the text. Magnitudes of the differences are also expressed as mean difference ± 95% confidence intervals (95% CI).