Intra- and intermuscular shear wave speed distributions in the human hamstrings during isometric and eccentric contractions

Abstract

Purpose

Hamstring strain injuries frequently occur near the biceps femoris long head (BFlh) proximal myotendinous junction. Unequal intra- and/or intermuscular mechanical properties may increase injury susceptibility by generating localised gradients in tissue mechanics and strain concentration; however, this remains unexplored under eccentric loading. We examined local shear wave speed (SWS), a measure reflecting tissue stiffness and force transmission, along the BFlh and semitendinosus (ST) during isometric (ISO) and eccentric (ECC) contractions.

Methods

Fifteen healthy males performed ISO and ECC knee flexions at 10%, 30%, and 50% of maximal voluntary torque (MVT). BFlh and ST SWS was measured at 30% (prox) and 50% (mid) of the thigh length using ultrasound elastography. BFlh pennation angle (PA), fascicle length (FL), and muscle thickness (MT) were also assessed.

Results

Absolute SWS was lower proximally in BFlh across intensities and conditions, whilst ST showed lower proximal SWS only in ISO at 30% and 50%MVT. When normalised to passive values, intramuscular SWS differences disappeared during ECC but persisted in BFlh during ISO ≤ 30%MVT. Intermuscular SWS ratio (BFlh/ST) was 0.63–0.94 at rest and 10%MVT, increasing to 0.78–1.01 at higher intensities. Moderate correlations were found between BFlh architecture and proximal SWS during ECC only (PA: r = 0.62–0.65; FL: r = − 0.64 to − 0.66; MT: r = 0.57).

Conclusion

Hamstring muscles exhibit region-specific SWS profiles, largely influenced by passive mechanical properties. This heterogeneity may contribute to localised strain concentrations and injury risk, highlighting a potential target for prevention strategies.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00421-025-05968-y.

Introduction

Hamstring strain injuries (HSIs) are the most prevalent thigh injuries in sports involving sprinting (Edouard et al. 2016). Approximately 70% of all HSIs occur in the biceps femoris long head (BFlh) (Grange et al. 2023). Although a higher frequency of injuries near the proximal BFlh has been reported (Askling et al. 2007; Crema et al. 2016; De Smet & Best 2000; Koulouris & Connell 2003), a recent systematic review suggests a more even distribution along the length of the muscle (Grange et al. 2023). Nevertheless, studies on BFlh mechanics (Fiorentino et al. 2012, 2014; Silder et al. 2010) and aponeurosis morphology (Evangelidis et al. 2015; Lazarczuk et al. 2024) point to the proximal region as an area of mechanical vulnerability. Overall, the available evidence indicates that a considerable proportion of BFlh injuries occur in the proximal and central regions, yet the precise reasons for this susceptibility remain unclear.

Computer simulations show that muscle fibre strains become increasingly non-uniform along the BFlh with faster running speeds, with the highest strains concentrating near the proximal MTJ (Fiorentino et al. 2014). In line with these findings, in vivo dynamic magnetic resonance imaging studies have reported increased strains near the BFlh proximal aponeurosis during low-intensity eccentric muscle actions (Fiorentino et al. 2012). In contrast, Bennett et al. (2014), using B-mode ultrasonography, found no differences in strain magnitude between proximal and distal BFlh fascicles in submaximal isometric contractions up to ~ 70%MVC, suggesting that any regional non-uniformity in the mechanical responses of the BFlh may depend on the type of contraction. Overall, the limited existing evidence indicates that local strain distribution varies along the BFlh during eccentric loading, exposing the proximal/central areas to higher, more hazardous strains.

Differences in intramuscular mechanical properties may contribute to the non-uniform strain distribution observed in the BFlh. For example, regions with steep mechanical gradients could act as loci of stress and strain concentration, potentially leading to tissue failure. To the best of our knowledge, the in vivo distribution of active mechanical properties (i.e., during muscle activation) within the BFlh has only been examined during isometric contractions. Studies using ultrasound shear wave elastography (SWE) have shown that BFlh local shear wave speed [SWS, a measure of mechanical response that has been used to infer tissue elasticity (Bercoff et al. 2004)] or shear modulus (estimated from SWS) is lower at the proximal (Vaz et al. 2021) and central (Miyamoto & Hirata 2021) regions compared to the distal region during submaximal isometric contractions. Interestingly, this inhomogeneity diminishes with increasing contraction intensity despite persisting differences in regional EMG activity (Miyamoto & Hirata 2021). This suggests that other factors besides muscle activation may also contribute to differences in active mechanical properties. For example, passive SWS has been shown to vary within the hamstring muscles depending on hip and knee joint angles (Bouvier et al. 2022; Miyamoto et al. 2020) but how such passive mechanical properties influence spatial variation in active tissue mechanics within this muscle group has not been examined directly. Importantly, the local mechanical behaviour of the BFlh during eccentric contractions, which are more relevant to HSIs, has yet to be investigated.

Efforts to understand the mechanisms behind HSIs have primarily focussed on BFlh, which is most affected, whilst the neighbouring semitendinosus (ST) has received relatively little attention. The two muscles are closely anatomically connected, sharing a proximal (conjoint) tendon, whilst ST proximal muscle fibres directly connect to the BFlh epimysium and proximal aponeurosis (Farfan et al. 2021). This anatomical connection may influence the load distribution between the BFlh and ST (Azzopardi et al. 2021). For example, we have found that ST active muscle SWS during eccentric contractions is higher than that of BFlh at the same intensity (Evangelidis et al. 2021), whilst fatigue induced by submaximal eccentric contractions appears to selectively reduce active SWS in the BFlh compared to the ST (Evangelidis et al. 2023). Taken together these findings suggest the presence of an intermuscular gradient in load bearing capacity between the BFlh and ST, which may be exacerbated by fatigue [a risk factor for HSIs (Opar et al. 2012)], potentially leading to excessive shearing stresses and strains within the more compliant BFlh. Such biomechanical conditions could help explain the higher incidence of strain injuries in the BFlh compared to the ST. However, no study has yet directly examined the intra- and intermuscular differences in BFlh and ST muscle mechanical behaviour at the proximal and middle regions (where injuries often occur) under eccentric loading.

The aim of this study was to examine the SWS distribution along the BFlh and ST muscles during isometric and eccentric contractions. Based on current evidence, we hypothesised that SWS would be lower at the proximal compared to the middle site, and that this inhomogeneity would be resolved with increasing contraction intensity. We also hypothesised that ST would exhibit higher local SWS than the BFlh, especially during eccentric contractions, leading to an intermuscular gradient in mechanical response.

Materials and methods

Participants

Fifteen healthy, recreationally active males (age 23.9 ± 3.8 years, height 1.73 ± 0.05 m, body mass 64.5 ± 7.9 kg, iPAQ score 1912 ± 1030 mean ± SD) participated in this study. Participants provided their written informed consent and completed a health screen and physical activity (iPAQ short version) questionnaires before their first visit to the lab. All participants were healthy, with no history of hamstring strain or knee joint injuries. The study adhered to the principles of the Declaration of Helsinki and was approved by the University Ethics Committee on Human Research [2019–268].

Overview

Participants visited the lab on three separate occasions, seven days apart. During the first visit, their anthropometric characteristics were measured, and they were familiarised with the procedures for the two main sessions (visits 2 and 3). In the main sessions, participants performed an identical protocol of either isometric (ISO) or eccentric (ECC) submaximal knee flexions at 10%, 30%, and 50% maximal voluntary torque (MVT) on an isokinetic dynamometer (Fig. 1). Knee flexor torque and muscle shear wave speed (SWS) measurements at proximal and middle sites of the biceps femoris long head (BFlh) and semitendinosus (ST) were collected (Fig. 2). Participants were instructed to avoid any strenuous physical activity for at least two days before each session.

Fig. 1.

A Overview of the study design. B Schematic representation of a typical main session. The two main sessions were identical except for the contraction mode (isometric or eccentric) and were conducted 7 days apart

Fig. 2.

Representative shear wave speed (SWS) measurements recorded at the proximal (prox) and middle (mid) regions of the biceps femoris long head (BFlh) and semitendinosus (ST) during eccentric contractions. Panels A-D correspond to passive conditions and to 10%, 30%, and 50% of maximal voluntary torque, respectively

The order of the main sessions and contraction intensity levels was randomised and counterbalanced across participants. At each contraction intensity level, both muscles (BFlh, ST) and sites (proximal, middle) were examined before proceeding to the next intensity level. Within each intensity level, the order of muscles and sites was randomised and counterbalanced, with each participant following the same order in both sessions.

Dynamometry

Participants lay prone on a custom-made minimally padded wooden board which was fitted to the isokinetic dynamometer (Con-Trex MJ, CMV, AG, Dübendordf, Switzerland) to allow for a 30° hip flexion (0° corresponds to full extension). The lateral epicondyle of the examined leg was aligned with the dynamometer’s axis of rotation, and participants were secured in place with inelastic straps around the torso, pelvis, and above the knee joint of the examined leg. The non-involved leg was slightly flexed at a comfortable position, and a strap was placed around the ankle joint to minimise extraneous movements during contractions. The dynamometer’s shin brace was secured just above the medial malleolus.

For the ISO condition, the dynamometer’s crank was fixed at 30° knee flexion (0° full extension) that is near the angle of peak torque (Kellis & Blazevich 2022). In the ECC condition, the range of motion was set from 40° of flexion to full extension, to include the angle of peak torque with an angular velocity of 5°·s⁻¹ to ensure sufficient elastography data would be collected, considering the low temporal resolution (1 Hz) of the SWE scanner (detailed below). Passive torque was recorded at the crank angle (for the ISO condition) or ROM (for the ECC condition). Participants then completed a standardised warm-up of isometric or eccentric knee flexions with progressively increasing intensity (three repetitions at 50%, 2 at 75%, and 1 at 90% of their perceived maximal effort), followed by two maximal (isometric or eccentric depending on the condition) voluntary contractions. The highest recorded torque (MVT) was used to calculate the target torque for the subsequent submaximal contractions.

After the warm-up, participants performed two sets of four isometric or eccentric knee flexions at 10%, 30%, and 50%MVT with 1-min rest between contractions and 5-min rest between sets (24 contractions in total). A monitor displayed their individual target torque and real-time torque trace, and they were instructed to match the two lines as quickly as possible. An examiner provided verbal cues and standardised verbal encouragement during each contraction. To ensure consistency in contraction duration between the ISO and ECC conditions, the isometric contractions were held for 8 s.

For the eccentric condition, participants were instructed to begin contracting isometrically just prior to the start of the crank movement. This allowed for parallel development of torque and muscle SWS before the onset of the eccentric phase and enabled the SWE signal to stabilise as torque reached the target level (see below).

The dynamometer signals (torque, crank angle, and velocity) were sampled at 2000 Hz using an A/D converter (PowerLab, ADInstruments, Australia). During offline analysis, a 4th order Butterworth-type filter with a 14 Hz (torque) and 4 Hz (angle/velocity) cut-off frequency (determined via residual analysis (Robertson et al. 2014)) was applied. All torque measurements were gravity-corrected using the passive torque recordings. For the eccentric contractions, the crank acceleration and deceleration phases were excluded and data within the isovelocity phase (5°·s⁻¹ ± 10%) were averaged for further analysis. In isometric contractions, torque data were averaged over the duration of the contraction before further analysis. All analyses were conducted by the same investigator.

Shear wave elastography

All measurements were conducted by the same investigator using an Aixplorer ultrasound scanner (SuperSonic Imagine, Aix-en-Provence, France) coupled with a linear transducer (SL15-4, 4–15 Hz) and the following settings: musculoskeletal preset, smoothing 9, and high persistence. SWS of the BFlh and ST was measured at 30% (proximal, prox) and 50% (middle, mid) of the thigh length (from the greater trochanter to the knee joint space). These locations were selected to facilitate the examination of potential intermuscular mechanical interactions between the BFlh and ST in neighbouring regions. The transducer was placed perpendicularly to the skin with minimal pressure, aligned to ensure the aponeuroses and several muscle fascicles were clearly visualised both at rest and during contraction. Once the optimal position was identified, the skin was marked with surgical markers to ensure accurate replication within and between sessions. Participants were also instructed to maintain their marks between sessions using markers provided to them.

During the SWS data acquisition, the region of interest (ROI) was set to its maximal dimensions and placed centrally over the muscle belly, avoiding any visible non-contractile tissues. For passive SWS measurements, participants were instructed to remain as relaxed as possible and duplicate measurements were taken. If any fascicle movement was visible on the ultrasound monitor, the measurement was repeated. For active SWS measurements, recording began approximately 3 s before each contraction to allow the signal to settle, minimising artefacts and/or rejection areas on the elastogram.

All recordings were exported from the scanner in video format and analysed using a custom MATLAB script (MathWorks Inc, Natick, USA). For analysis, the maximal possible rectangular ROI was drawn within the elastogram, avoiding any artefacts and rejection areas. SWS was calculated for every second and averaged across the duration of the passive measurement or contraction, prior to any further analysis. Torque and SWE signals were visually aligned based on the concurrent rise in torque and SWS. As SWE values were averaged across multiple elastograms during the contraction phase, exact temporal synchronisation with the torque data was not necessary.

SWS was normalised to passive values by calculating the ratio of SWS during active muscle contraction to that measured under passive conditions. The intermuscular (BFlh/ST) SWS ratio was calculated by dividing the SWS of the BFlh by that of the ST at the corresponding region (either 30% or 50% of thigh length) and contraction level for each muscle.

Muscle architecture

Duplicate B-mode ultrasound images recorded from the BFlh proximal and middle sites at rest were used to assess muscle architecture. Images from one participant were excluded from further analysis due to inadequate image quality for reliable measurement of muscle architecture parameters. ST muscle architecture was not assessed due to the parallel orientation of its fascicles at both measurement sites. Additionally, the ST deep aponeurosis could not be visualised in several participants, precluding the measurement of its muscle thickness (MT).

The BFlh images were imported into ImageJ (v.1.54c, National Institutes of Health, USA) for analysis. Local BFlh pennation angle (PA) was measured as the angle between the fascicles and the intermediate/deep aponeurosis at the proximal and middle sites. For each site, two images were analysed, and two clearly visible fascicles in each image were selected for analysis. The representative local PA was calculated as the average PA across fascicles and images. BFlh fascicle length (FL) was estimated using trigonometry (FL = MT/sin(PA)) (Kawakami et al. 1995).

MT at the proximal and middle sites was measured as the perpendicular distance between the superficial and intermediate/deep aponeuroses of the BFlh at three different locations within each image. The representative MT measurement for each site was the average across the locations and images.

Statistical analysis

Data are presented as mean ± SD. The Shapiro–Wilk test was used to assess the normality of the data.

The intraclass correlation coefficient (ICC) was calculated to assess relative reliability using a two-way random-effects model with absolute agreement (ICC 2,k), where k = number of repeated measurements averaged per participant. The ICC coefficient was classified as < 0.50 (poor), 0.50–0.75 (moderate), 0.75–0.90 (good), and > 0.90 (excellent) (Koo & Li 2016). The within-subject coefficient of variation (CVw) was calculated as

$$CVw\left(\%\right)=100\times \sqrt{\frac{1}{n}\sum _{i=1}^n\frac{{\mathit{SD}}_{w,i}^2}{{\mathit{Mean}}_i^2},}$$

where SDw is the within-subject standard deviation of repeated measurements.

The total coefficient of variation, reflecting variability across all repeated measurements and participants, was calculated as

$$CVt\left(\%\right)=100\times \frac{\mathit{SD}}{\mathit{Mean}},$$

where SD is the standard deviation of all measurements pooled across participants. CV was interpreted as very good (< 5%), good (5–10%), and acceptable (< 15%).

Minimal detectable differences at a single site (within-region, MDDw) and between the proximal and middle sites (MDDpm) were calculated post hoc, using the observed standard deviation of the differences and assuming 80% power. The results should be interpreted cautiously and not taken as an indication of the study’s a priori sensitivity.

The within-region minimal detectable difference was calculated as

$$MDDw={(t}_{1-\frac{\alpha }{2},df}+t_{1-\beta })\times \frac{{\mathit{SDdiff}}_w}{\sqrt{n}}\times \sqrt{2},$$

where t is the critical value from the two-tailed t-distribution with df = n-1 and significance level a = 0.05, t_1-β is the critical value from the t-distribution corresponding to 80% power (i.e., β = 0.20), SDdiff_w is the within-subject standard deviation of the paired differences between repeated measurements at the same site, n is the number of participants, and the factor √2 accounts for the variance introduced by the two repeated measurements.

The between-region MDD was calculated as

$$MDDpm={(t}_{1-\frac{\alpha }{2},df}+t_{1-\beta })\times \frac{{\mathit{SDdiff}}_{\mathit{pm}}}{\sqrt{n}}\times \sqrt{2,}$$

where t is the critical value from the two-tailed t-distribution with df = n-1 and significance level a = 0.05, t_1-β is the critical value from the t-distribution corresponding to 80% power (i.e., β = 0.20), SDdiff_pm is the within-subject standard deviation of the paired differences between proximal and middle sites, n is the number of participants, and the factor √2 accounts for the variance introduced by the two repeated measurements.

Paired t tests were conducted to compare knee flexor MVT between conditions, as well as torque values recorded during the corresponding elastography measurements at proximal and middle sites. Separate two-way repeated-measures analyses of variance (ANOVA) were used to examine SWS (contraction intensity [passive, 10%, 30%, 50%MVT] x measurement site [proximal, middle]). The intermuscular (BFlh/ST) SWS ratio was analysed using a three-way repeated-measures ANOVA (contraction type [isometric, eccentric] x contraction intensity [passive, 10%, 30%, 50%MVT] x site [proximal, middle]). Mauchly’s test was used to check for sphericity, and the Greenhouse–Geisser correction was applied when sphericity was violated. Post hoc comparisons, where applicable, were performed using the Holm–Bonferroni correction. For all t tests, normality was assessed with the Shapiro–Wilk test. Effect sizes were calculated using Cohen’s d for paired t tests and omega squared (ω²) for ANOVAs. Cohen’s d effect sizes were interpreted as trivial (< 0.2), small (0.2), medium (0.5), and large (0.8), whilst ω² effect sizes were classified as trivial (< 0.01), small (0.01), medium (0.06), and large (0.14). Bivariate correlations between BFlh SWS and architecture (PA, FL, MT) were assessed using Pearson’s product moment coefficient (r) and classified as weak (< 0.4), moderate (0.4–0.7), and strong (> 0.7). Statistical significance was set at p ≤ 0.05. All statistical analyses were conducted with JASP 0.17.1 (JASP Team 2023).

Results

Repeatability

The full repeatability statistics are provided in the Supplementary Information (SI). Briefly, torque measurements demonstrated good-to-very good absolute reliability across all conditions, intensity levels, muscles, and measurement sites (CVw, ISO: 1.1–4.0%; ECC: 0.8–7.6%) and moderate-to-excellent relative reliability [ICC(2,2): ISO: 0.683–0.949; ECC: 0.766–0.971]. The corresponding SWS measurements exhibited acceptable to very good absolute reliability (CVw, ISO: 2.6–12.3%; ECC: 2.7–15.4%) and poor-to-excellent relative reliability [ICC(2,2): ISO: 0.298–0.975; ECC: 0.461–0.961]. For muscle architecture measurements, absolute reliability was good to very good [CVw (BFlh_prox, BFlh_mid); PA: 3.3%, 2.0%; FL: 3.9%, 2.8%; MT: 7.1%, 2.6%] and the respective relative reliability was good to excellent [ICC(2, k) (BFlh_prox, BFlh_mid); PA: 0.872, 0.910; FL: 0.973, 0.982; MT: 0.966, 0.996), with k = 4 for PA, k = 2 for FL, and k = 6 for MT].

Torque

Maximal voluntary torque (MVT) of the knee flexors did not differ between conditions (ISO, 120.0 ± 14.4 Nm; ECC, 122.9 ± 21.0 Nm, p = 0.413). Similarly, submaximal torque normalised to MVT (%MVT) showed no significant differences between contractions for the corresponding elastography measurements at proximal and middle sites for each muscle and condition (ISO, p = 0.084–0.692; ECC, p = 0.134–0.716).

Intramuscular SWS distribution

Isometric contractions

For the BFlh, significant main effects of contraction intensity (F_3,42 = 199.17, p < 0.001, ω² = 0.86) and site (F_1,14 = 52.20, p < 0.001, ω² = 0.22) were found on absolute SWS along with a significant contraction intensity x site interaction (F_2.1,29.4 = 11.87, p < 0.001, ω² = 0.09) (Fig. 3A, Fig. SI 1). Post hoc comparisons showed lower BFlh SWS at the proximal vs. middle site at 10%MVT (M = –0.82 m/s, 95% CI [–1.46, –0.17], p < 0.001, d = –0.74), 30%MVT (M = –1.63 m/s, 95% CI [–2.28, –0.99], p < 0.001, d = –1.49), and 50%MVT (M = –0.85 m/s, 95% CI [–1.50, –0.21], p < 0.001, d = –0.78).

Fig. 3.

Absolute A and normalised B shear wave speed (SWS) in the biceps femoris long head (BFlh; green) and semitendinosus (ST; orange) proximal (prox; lighter tone) and middle (mid; darker tone) sites during isometric contractions under passive conditions and at 10%, 30%, and 50% of maximal voluntary torque (MVT). *p < 0.05, **p < 0.01, ***p < 0.001

For the ST, there were significant main effects of contraction intensity (F_3,42 = 296.86, p < 0.001, ω² = 0.91) and site (F_1,14 = 17.29, p < 0.001, ω² = 0.16) and a significant contraction intensity x site interaction (F_3,42 = 3.38, p = 0.027, ω² = 0.04). Post hoc comparisons showed lower ST SWS at the proximal site vs. middle site at 30%MVT (M = –1.00 m/s, 95% CI [–1.71, –0.28], p < 0.001, d = –1.08) and 50%MVT (M = –0.70 m/s, 95% CI [–1.14, 0.02], p = 0.012, d = –0.75) (Fig. 3A).

The ANOVA analysis for the normalised (to passive values) BFlh SWS in ISO revealed significant main effects of intensity (F_3,42 = 165.92, p < 0.001, ω² = 0.82) and site (F_1,14 = 16.17, p = 0.001, ω² = 0.08), along with a significant intensity x site interaction (F_1.8,25.6 = 6.96, p = 0.005, ω² = 0.05). Post hoc comparisons showed lower BFlh SWS at the proximal vs. the middle site at 10%MVT (M = –0.25 m/s, 95% CI [–0.54, 0.05], p = 0.045, d = –0.49) and 30%MVT (M = –0.51 m/s, 95% CI [–0.80, 0.21], p < 0.001, d = –1.02) (Fig. 3B).

In contrast, for the ST, relative SWS in ISO showed only a main effect of contraction intensity (F_3,42 = 266.65, p < 0.001, ω² = 0.89) but no main effect of site (p = 0.488) or intensity x site interaction (p = 0.445).

Eccentric contractions

For BFlh, there were significant main effects of contraction intensity (F_2.2,30.5 = 275.99, p < 0.001, ω² = 0.89) and site (F_1,14 = 50.98, p < 0.001, ω² = 0.24) and a significant contraction intensity x site interaction (F_3,42 = 6.23, p = 0.001, ω² = 0.06). Post hoc comparisons showed lower BFlh absolute SWS at the proximal vs. middle site at 10%MVT (M = –0.88 m/s, 95% CI [–1.54, –0.23], p < 0.001, d = –0.89), 30%MVT (M = –1.37 m/s, 95% CI [–2.03, –0.72], p < 0.001, d = –1.38), and 50%MVT (M = –0.79 m/s, 95% CI [–1.44, –0.14], p < 0.001, d = –0.79) (Fig. 4A, Fig SI 2).

Fig. 4.

Absolute A and normalised B shear wave speed (SWS) in the biceps femoris long head (BFlh; green) and semitendinosus (ST; orange) proximal (prox; lighter tone) and middle (mid; darker tone) sites during eccentric contractions at 10%, 30%, and 50% of maximal voluntary torque (MVT). **p < 0.01, ***p < 0.001

For the ST, there were significant main effects of contraction intensity (F_1.9,26.5 = 268.73, p < 0.001, ω² = 0.91) and site (F_1,14 = 9.78, p = 0.007, ω² = 0.12), but the contraction intensity x site interaction was not significant (F_3,42 = 2.46, p = 0.076, ω² = 0.02) (Fig. 4A).

When normalised, ECC BFlh SWS exhibited only a main effect of contraction intensity (F_2,28.4 = 246.3, p < 0.001, ω² = 0.89) with no effect of site (p = 0.134) or interaction (p = 0.110) (Fig. 4B). Similarly, ST relative SWS in ECC showed a main effect of contraction intensity (F_2.1,29.3 = 239.6, p < 0.001, ω² = 0.87) but no effect of site (p = 0.364) or intensity x site interaction (p = 0.482).

Intermuscular SWS distribution

A three-way repeated-measures ANOVA on BFlh/ST SWS ratio revealed significant main effects of contraction type (F_1,14 = 8.71, p = 0.011, ω² = 0.09), intensity (F_3,42 = 16.3, p < 0.001, ω² = 0.32), and site (F_1,14 = 15.06, p = 0.002, ω² = 0.07) (Fig. 5). The contraction type x intensity interaction was also significant (F_3,42 = 4.54, p = 0.008, ω² = 0.07), but the contraction type x site (p = 0.781), contraction intensity x site (F_3,42 = 2.77, p = 0.054, ω² = 0.019), or contraction type x intensity x site (p = 0.543) interactions were not significant.

Fig. 5.

Intermuscular shear wave speed (SWS) ratio between biceps femoris long head (BFlh) and semitendinosus (ST) at the proximal and middle sites during isometric and eccentric contractions under passive conditions and at 10%, 30%, and 50% of maximal voluntary torque (MVT)

BFlh muscle architecture (n = 14)

BFlh MT was 2.09 ± 0.31 cm at the proximal site and 2.19 ± 0.34 cm at the middle site (p = 0.136). Local PA was 12.3° ± 2.7° in BFlh_prox and 13.8° ± 2.7° in BFlh_mid (p = 0.039, d = –0.68), whilst FL was 10.1 ± 1.9 cm for BFlh_prox and 9.3 ± 1.5 cm for BFlh_mid (p = 0.102).

There were no significant correlations between local BFlh MT and SWS, except for BFlh_prox at 50%MVT in the ECC condition (r = 0.57, 95% CI [0.05, 0.84], p = 0.034). Moderate correlations were found between BFlh_prox PA and SWS at 30%MVT (r = 0.67, 95% CI [0.21, 0.88], p = 0.010) and 50%MVT (r = 0.62, 95% CI [0.13, 0.86], p = 0.019) during ECC, but there were no significant relationships for the middle site in any condition.

BFlh FL at the proximal site also exhibited moderate correlations with SWS at 10%MVT (r = − 0.64, 95% CI [− 0.87, − 0.17], p = 0.013), and 30%MVT (r = − 0.66, 95% CI [− 0.88, − 0.19], p = 0.011) during ECC. However, there were no significant relationships between BFlh FL and SWS for the middle site.

Discussion

This is the first study that examined the BFlh and ST local stiffness distribution in vivo during both isometric and eccentric contractions using ultrasound SWE. We found that: (1) BFlh exhibited lower absolute SWS at the proximal compared to the middle site during both ISO and ECC, whilst ST showed this pattern only in ISO (partially in line with our hypothesis). Contrary to our hypothesis, this inhomogeneity persisted up to moderate contraction intensity, (2) the intramuscular differences in SWS distribution were largely resolved when passive SWS was accounted for, especially during eccentric contractions, and (3) the intermuscular (BFlh/ST) SWS ratio was below equality (< 1) at rest and during low-intensity isometric and eccentric contractions but became more balanced at higher torque levels (partially supporting our hypothesis).

Intramuscular local SWS distribution

Whilst BFlh injuries appear to be relatively evenly distributed along the muscle (Grange et al. 2023), the proximal region exhibits anatomical (Evangelidis et al. 2015; Lazarczuk et al. 2024) and mechanical (Fiorentino et al. 2012, 2014; Silder et al. 2010) characteristics that may make it more susceptible to strain concentrations during eccentric loading, potentially explaining why a substantial proportion of injuries occur there. These strain concentrations may result from inhomogeneity in intramuscular mechanical properties, facilitating the formation of steep mechanical gradients that increase localised tissue stress. Our findings support the presence of such gradients within the BFlh during both isometric and eccentric contractions, and within the ST during isometric contractions (Figs. 3, 4).

Whilst there is no other SWE study investigating regional mechanical differences during eccentric loading, limited data from previous SWE studies focussed on isometric contractions are partly in agreement. Vaz et al. (2021) found lower shear modulus (estimated from SWS) at the proximal vs. distal sites of the BFlh at contraction intensities comparable to those examined in our study. Similarly, Miyamoto & Hirata (2021) reported that BFlh SWS was lower at proximal compared to central and distal sites at 20%MVC during isometric contractions, although these differences were not observed at higher intensities up to 80%MVC.

The observed heterogeneity in intramuscular active SWS distribution may reflect local differences in passive mechanical properties (Bouvier et al. 2022; Miyamoto et al. 2020). Indeed, when active SWS values were normalised to passive SWS, regional differences were eliminated in BFlh during ECC and in ST during ISO. However, in BFlh, regional differences persisted during isometric contractions up to 30%MVT (Fig. 4). These findings suggest that passive tissue properties likely contribute to regional variations in active muscle mechanics, especially for eccentric contractions. Notably, at the group level, passive SWS was not significantly different between regions in either the BFlh or ST (Fig. 3, Fig. SI 1–2); however, inter-individual variability was 2.5-fold larger at the middle compared to the proximal site of the BFlh during ECC (CV, 10.7% vs. 4.2%, respectively; see SI dataset) and 1.3-fold larger during ISO (CV, 8.6% and 6.6%, for middle and proximal regions, respectively). This increased variability indicates substantial differences in passive tissue characteristics between individuals, potentially reflecting factors such as muscle architecture, which may have concealed significant differences at the group level (discussed below).

Importantly, passive SWS itself is influenced by hip and knee joint configuration. Resting SWS varies with joint position, with higher values generally observed at longer muscle lengths (Bouvier et al. 2022; Miyamoto et al. 2020), highlighting that some of the regional differences in active SWS observed in our study should be interpreted in the context of the limb position and range of motion used during measurement (i.e., corresponding to the late swing phase of running and the angle at knee flexor peak torque).

Regarding the ST, there are currently no other elastography data on intramuscular differences in active SWS. Nevertheless, our findings on passive mechanical properties are consistent with those of Miyamoto et al. (2020) who reported higher shear modulus distally compared to the proximal and middle regions of the ST. In contrast, Kositsky et al. (2022) found no significant differences in shear modulus, as measured by SWE, between the proximal and distal compartments of the ST. This discrepancy may be partly due to methodological differences in the hip joint angles during measurements and the specific locations examined along the ST. Kositsky et al. (2022) assessed the muscle with the hip in neutral position and examined the areas just proximal and distal of the tendinous inscription. In contrast, Miyamoto et al (2020) used more flexed hip positions up to 90°, corresponding to longer muscle lengths, and examined regions at 25%, 50%, and 75% of the muscle belly.

Variability in local muscle architecture may contribute to regional mechanical property differences (Azizi and Deslauriers 2014; Koo and Hug 2015). In our study, BFlh PA was lower in the proximal region than in the middle by − 1.52º ± 2.49º; however, this difference did not exceed the post hoc-calculated minimal detectable difference (see SI). Nevertheless, this pattern is consistent with the previous findings (Tosovic et al. 2016), though not all studies have reported similar results (Kellis et al. 2010). Whilst FL and PA showed moderate correlations with ECC SWS at proximal BFlh at 30% and at 50%MVT, overall, BFlh architecture exhibited no consistent relationship with SWS in either passive or active conditions. It is important to note that these correlations were exploratory and should be interpreted with caution. Taken together, our findings suggest that any influence of intramuscular BFlh architecture on local tissue mechanics is likely limited; however, further research is needed to clarify this relationship.

Even after normalising to passive SWS, differences in BFlh local active SWS persisted at low-intensity isometric contractions, suggesting that additional factors may contribute to intramuscular heterogeneity. One possible explanation is that intramuscular SWS differences reflect selective neural activation along the BFlh muscle (Hegyi et al. 2019). Although EMG activity was not measured in our study, this interpretation is in line with Miyamoto and Hirata (2021) showing lower SWS and EMG activity in the proximal region compared to the middle and distal regions of the BFlh at 20%MVC. Interestingly, at higher intensities (up to 80%MVC), whilst BFlh EMG activity remained lower proximally, SWS became uniform across all sites indicating a decoupling between neuromuscular activation and local mechanical response (Miyamoto & Hirata 2021).

Despite an apparent equalisation of the proximo-distal mechanical responses at high contraction intensities, localised intramuscular gradients at lower activation levels may still increase the risk of injury. Since increases in SWS reflect both increases in strain and the number of active sarcomeres in parallel, regions exhibiting intramuscular mechanical gradients likely experience inhomogeneous local forces. A BFlh injury may occur when the muscle is experiencing rapid changes in activation (from high to low and vice versa) as seen during the late swing phase and early stance phase—both implicated in the timing of strain injury occurrence (Kenneally-Dabrowski et al. 2019; Liu et al. 2017).

Notably, a similar pattern of regional vulnerability has been reported in electrically evoked isometric contractions, where injury occurred in less active or inactive regions due to local elongation of passive structures (Fouré and Gondin 2021). Despite the different experimental conditions (voluntary vs. evoked contractions), these findings collectively highlight regional mechanical inhomogeneity as a key factor.

Overall, these findings suggest that inhomogeneity in regional muscle activation may contribute to mechanical property gradients up to a certain threshold of contraction intensity, beyond which other factors likely influence further increases in mechanical responses. Further research is needed to clarify these factors and their role in active muscle mechanical behaviour.

Intermuscular local SWS distribution

The proximal hamstrings exhibit complex anatomical relationships that may influence load distribution and injury susceptibility, yet these factors remain underexplored. The BFlh and ST muscles share a common tendon, with fibres of the ST originating directly from the BFlh epimysium and proximal aponeurosis (Farfan et al. 2021)—a region commonly involved in strain injuries. This anatomical arrangement may increase injury susceptibility (Farfan et al. 2021). We previously found that ST exhibits a greater mechanical response (measured with SWE at 50% of the thigh length) compared to the BFlh during eccentric contractions at the same intensity (Evangelidis et al. 2021), suggesting the presence of an intermuscular gradient in mechanical behaviour that may be further exacerbated by fatigue (Evangelidis et al. 2023). Such a progressively greater gradient could increase the risk of localised strain and injury, particularly during repetitive or high-load eccentric loading, e.g. during sprinting.

In this study, we found that the intermuscular (BFlh/ST) SWS ratio was lower at the proximal compared to the middle site, falling below equality in the passive condition and up to 30%MVT but became more balanced at the highest contraction intensity. BFlh often shows lower SWS than ST at the same intensity and contraction type, as observed in this study and our previous work (Evangelidis et al. 2021), although the physiological basis of this difference remains unclear. Differences in muscle–tendon unit anatomy, including tendon length (Kellis 2018), muscle architecture (Kellis et al. 2010), fibre-type distribution (Evangelidis et al. 2017; Fournier et al. 2022), and connective tissue content (Kim et al. 2024) may partially contribute to the observed SWS responses. Our results support the existence of intermuscular regions where stress (and strain) may concentrate. However, as the gradient in mechanical response diminishes with increasing contraction intensity, any intermuscular differences may be less relevant under maximal or near-maximal loading conditions.

Limitations

Our findings should be interpreted considering certain limitations. First, the sample size was determined by constraints in resources and time rather than a priori power calculations. However, this sample size is comparable to previous studies utilising shear wave elastography and exceeds the median sample size of those cited here. Additionally, some of our statistically significant differences did not exceed the post hoc-calculated minimal detectable differences and should be interpreted with caution. Future studies with larger cohorts are warranted to confirm our results. Second, sEMG activity was not measured due to spatial constraints at the measurement sites, particularly in the proximal regions. Consequently, we cannot exclude potential changes in muscle activation or coordination amongst the hamstrings, particularly during eccentric contractions, which may have influenced torque maintenance across the range of motion. Third, muscle architecture in ST was not assessed due to the parallel fibre orientation at the measurement levels. Additionally, in several participants, the deep aponeurosis of the ST was not visible preventing measurement of muscle thickness. Fourth, SWS measurement sites were determined relative to thigh length. Due to inter-individual anatomical differences, a given location along the thigh may not correspond to the same relative anatomical position within each muscle. Nevertheless, these sites were selected to enable the examination of potential intermuscular mechanical interactions between the BFlh and ST in neighbouring regions.

In conclusion, our findings demonstrate the presence of inhomogeneous local mechanical responses both within and between the BFlh and ST muscles. These inhomogeneities may contribute to strain injury susceptibility by generating regions of elevated mechanical gradients and strain concentration, particularly during eccentric contractions. These imbalances appear to arise primarily from differences in passive mechanical properties rather than active muscle function, whilst the role of muscle architecture remains unclear. Notably, intermuscular differences in mechanical behaviour diminish rapidly with increasing contraction intensity, suggesting that they are unlikely to contribute to the higher susceptibility of the BFlh compared to the ST. Future research should further investigate the role of muscle architecture and explore whether targeted strength training interventions can promote more uniform mechanical responses, thereby potentially reducing injury risk.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ANOVA

Analysis of variance

BFlh

Biceps femoris long head

ECC

Eccentric condition

FL

Fascicle length

ISO

Isometric condition

HSI

Hamstring strain injuries

MT

Muscle thickness

MVC

Maximal voluntary contraction

MVT

Maximal voluntary torque

PA

Pennation angle

sEMG

Surface electromyography

ST

Semitendinosus

SWE

Shear wave elastography

SWS

Shear wave speed

Author contributions

PE and YK conceived and designed the study. PE, CY, and HI conducted the experiments. PE analysed the data and drafted the manuscript. PN contributed to statistical analysis and data visualisation. All authors critically revised the manuscript and approved the final version.

Funding

PE role in this study was supported by a Grant-in-Aid for Japan Society for the Promotion of Science Research Fellows [JP 17F17412]. YK was supported by a Grant-in-Aid for Scientific Research (A) [No. 16H01870]. Both were supported by a Grant-in-Aid for Scientific Research (Fostering Joint International Research B; 21KK0175). PE and PN are currently supported by the National Institute for Health Research (NIHR) Exeter Biomedical Research Centre (BRC). The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care.

Data availability

All data generated or analysed during this study are included in this published article and its supplementary information files.

Declarations

Conflict of interest

The authors have no relevant financial or non-financial interests to disclose.

Ethical approval

The study was approved by the University Ethics Committee on Human Research [2019–268].

Consent to participate

Informed consent was obtained from all individual participants included in the study.