Stroke remains one of the leading causes of long-term disability worldwide, and many survivors continue to experience impairments in motor control, balance, and gait despite structured rehabilitation programs (1–5). Although the primary neurological injury explains much of this functional limitation, accumulating evidence indicates that peripheral musculoskeletal structures also undergo secondary alterations after stroke (2, 6–10). Reduced loading of the paretic limb, spasticity, asymmetrical gait mechanics, and prolonged immobilization may contribute to adaptive or maladaptive changes in tendon and muscle morphology (3, 8, 10–16). Despite these observations, tendon-specific structural changes in stroke survivors—particularly at weight-bearing lower-limb sites—remain only partially characterized, with available data scattered across small, site-specific studies (3, 8, 10–16).

Musculoskeletal ultrasonography has become an accessible and reliable tool for evaluating tendon and muscle architecture in neurological and orthopedic conditions (15, 17–21). Prior research in stroke rehabilitation has largely focused on changes in muscle thickness, fascicle length, muscle stiffness, or spasticity-related adaptations (2, 7–9, 12, 17). In contrast, the number of studies directly assessing tendon morphology in post-stroke patients is limited, with most data focusing on the Achilles tendon or describing entheseal alterations such as changes in the plantar fascia (3, 11–15, 22). A systematic review of ultrasonographic muscle and tendon properties in the spastic lower leg after stroke highlighted major gaps in longitudinal data and called for prospective studies starting early after stroke (8). A subsequent Achilles tendon study likewise emphasized the need for longitudinal designs that link tendon morphology to functional outcomes and targeted rehabilitation interventions (3) furthermore, recent tendon-focused work suggests that chronic hemiparesis may lead to increased Achilles tendon thickness (3, 11) and altered tendon mechanical properties compared with healthy or contralateral limbs (8, 22), yet evidence for similar adaptations in other tendons remains sparse and is mostly limited to cross-sectional fascia studies in the foot and lower leg (8, 13–16, 22). Little is known about how these structural changes evolve over the course of rehabilitation (8, 22).

Understanding how tendon properties respond during rehabilitation is clinically relevant, because tendon stiffness, thickness, and mechanical behavior influence muscle–tendon unit function and ankle joint mechanics, which in turn may contribute to gait capacity, balance, and ultimately functional independence after stroke (2, 10, 12, 23–25). Tendon-level adaptations may also help explain inter-individual variability in motor recovery and treatment responsiveness. Therefore, the primary aim of this study was to examine the within-patient change in quadriceps tendon thickness over a 4-week inpatient rehabilitation period. Secondary aims were to (i) compare quadriceps tendon thickness with healthy controls at baseline and post-treatment as cross-sectional reference comparisons, (ii) describe changes in other lower-limb tendons/fascia (patellar and Achilles tendon thicknesses and plantar fascia thickness), and (iii) assess prespecified associations between changes in quadriceps tendon thickness and changes in balance and functional independence.

Forty-five patients and fifteen controls were included. Median age was 64 (60–67) years in patients and 60 (55.5–64.5) years in controls (p = 0.185); sex distribution and anthropometrics were comparable between groups (p > 0.05 for all) (Table 1). Baseline stroke-related characteristics (including time since stroke, stroke type/side, and baseline Brunnstrom stages, FAC, BBS, Barthel Index, and MAS distributions) are presented in Table 1.

Ordinal outcomes were examined using shift analyses (Table 2; Supplementary Table S1). Brunnstrom stages showed predominantly favorable transitions with no worsening: Brunnstrom UE improved by ≥1 stage in 19/45 (42.2%), Brunnstrom LE in 15/45 (33.3%), and Brunnstrom Hand in 6/45 (13.3%), with the remainder unchanged (Table 2). FAC improved by ≥1 level in 27/45 (60.0%), while 18/45 (40.0%) remained unchanged (Table 2). In contrast, MAS demonstrated a marked floor effect—particularly at the hip and knee—where most participants were already graded 0 at baseline and remained unchanged (43/45, 95.6% for both); similarly, 35 (77.7%) participants were graded 0 at baseline for foot MAS, and overall 38/45 (84.4%) remained unchanged, with no participants worsening in any MAS domain. The full transition patterns by baseline category and post-treatment category are provided in Supplementary Table S1.

The BBS increased from 36 (27–44) at baseline to 42 (33–51) post-treatment (Δ: 6 [4–9], p < 0.001). Using a published MCID threshold of ≥5 points for the BBS, 29/45 (64.4%) participants achieved a change exceeding the MCID (34). The Barthel Index increased from 70 (50–85) at baseline to 75 (60–90) post-treatment (Δ: 0 [0–5], p < 0.001). Based on published MCID approaches for the Barthel Index in the stroke literature (including an estimate of 1.85 points on the 20-point BI, corresponding to approximately 9.25 points on a 0–100 scale), 10/45 (22.2%) participants exceeded the MCID criterion (35).

Measurement reliability and error estimates (ICC with 95% CIs, SEM, and MDC₉₅) for each tendon/fascia and side are presented in Supplementary Table S2. For the quadriceps tendon, MDC₉₅ was 0.50 mm on the paretic side and 0.45 mm on the non-paretic side.

For the primary outcome, paretic-side quadriceps tendon thickness increased from 5.94 ± 0.96 mm at baseline to 6.48 ± 0.95 mm post-treatment (p < 0.001), with a median paired change of 0.4 (0.3–0.8) mm (Table 3). The effect size was large (Cohen’s dz. = 1.358, 95% CI: 0.947–1.761), and 21/45 participants (46.7%) exceeded the MDC₉₅ threshold on the paretic side. Quadriceps tendon thickness on the non-paretic side also increased (6.09 ± 0.83 to 6.54 ± 0.77 mm; p < 0.001; median Δ 0.3 mm [0.2–0.6]; dz. = 1.148, 95% CI: 0.766–1.521), with 17/45 (37.8%) exceeding MDC95 (Table 3). Secondary analyses showed statistically significant increases across the other sites on both sides (all p < 0.001). In sensitivity analyses, all pre–post comparisons remained statistically significant when re-checked using the Wilcoxon signed-rank test (all p < 0.001), and the direction of effects was unchanged.

Cross-sectional comparisons with healthy controls at baseline and post-treatment are summarized in Table 4 (Holm-adjusted p-values reported separately within each tendon/fascia for the two side-specific comparisons, and separately for baseline and post-treatment). At baseline, quadriceps tendon thickness on the paretic side was lower in patients than controls (5.94 ± 0.96 vs. 6.58 ± 0.87 mm; p = 0.024; p_adj = 0.048; Cohen’s d = −0.689, 95% CI: −1.284 to −0.089), whereas the non-paretic side did not remain significant after adjustment (p = 0.054; p_adj = 0.108). At post-treatment, quadriceps thickness did not differ from controls on either side (both p_adj > 0.999). Post-treatment patellar and Achilles tendon thicknesses were higher in patients than controls on both sides (all p_adj ≤ 0.010). All other baseline and post-treatment patient–control comparisons were not statistically significant after Holm adjustment (Table 4).

In prespecified correlation analyses (four tests; FDR-adjusted q-values reported), changes in quadriceps tendon thickness were modestly associated with improvements in balance and functional independence (Table 5). ΔQuadriceps was positively correlated with ΔBBS on both sides (paretic: rho = 0.346, 95% CI: 0.037 to 0.622, p = 0.020, q = 0.037; non-paretic: rho = 0.365, 95% CI: 0.100 to 0.585, p = 0.014, q = 0.037). ΔQuadriceps_NP also showed a modest positive association with ΔBarthel (rho = 0.329, 95% CI: 0.015 to 0.580, p = 0.028, q = 0.037), whereas ΔQuadriceps_P was not associated with ΔBarthel (rho = 0.131, 95% CI: −0.205 to 0.422, p = 0.392, q = 0.392). The full correlation heatmap was considered exploratory and is provided in the Supplementary Figure S1.

This study investigated ultrasound-based thickness changes in major lower-limb tendons and plantar fascia structures in individuals with post-stroke hemiplegia and explored how these changes relate to functional recovery over a four-week inpatient rehabilitation period. The primary finding was a within-patient increase in quadriceps tendon thickness over the 4-week inpatient period. Baseline patient–control differences were observed only on the paretic side, and post-treatment cross-sectional reference comparisons suggested no clear difference from controls; however, controls were not followed longitudinally. Over the same period, the patellar tendon, Achilles tendon, and plantar fascia thicknesses also demonstrated bilateral increases, although of smaller magnitude. In addition, quadriceps tendon thickness change showed modest associations with improvements in balance and functional independence, whereas other correlations were weak. Importantly, MDC₉₅-based responder analyses indicated that a substantial subset of participants demonstrated Quadriceps tendon thickness changes exceeding measurement error, and MCID-based summaries suggested that many participants achieved clinically meaningful improvement in balance (BBS), whereas fewer met MCID-based criteria for functional independence (Barthel). Because we lacked a longitudinal control group or non-intervention comparator, the observed changes should be interpreted as time-related change over the inpatient period (potentially reflecting natural recovery, hospitalization-related mobilization/activity, and standard rehabilitation exposure), rather than as definitive rehabilitation effects. Taken together, these findings suggest that peripheral tendon thickness change may accompany functional gains during neurorehabilitation and may reflect adaptations related to changes in mechanical loading of the lower extremity.

Reduced baseline quadriceps tendon thickness is consistent with known mechanisms of post-stroke musculoskeletal deconditioning, including reduced paretic-extremity loading, asymmetrical gait patterns, diminished voluntary activation, and spasticity-driven disuse (1, 8, 21, 23). Previous work has described a range of post-stroke muscle architectural changes—including reduced muscle thickness, shorter fascicles, increased stiffness and impaired mechanical properties—in the quadriceps, tibialis anterior and gastrocnemius muscles (2, 3, 6, 7, 9). However, tendon-specific data remain limited (8). A systematic review of ultrasonographic muscle and tendon properties in the spastic lower leg reported evidence of altered tendon characteristics in stroke survivors, but emphasized that longitudinal studies starting early after stroke are still lacking (8). Our findings add to this limited literature by suggesting that quadriceps tendon morphology is already altered at baseline and may change over a short, structured inpatient rehabilitation period.

Over the 4-week inpatient period, we observed increases in thickness across all examined sites, with the quadriceps tendon demonstrating the largest changes. Tendons are biologically responsive to repetitive loading, and increases in cross-sectional thickness are generally interpreted as reflecting fibrillar reorganization, altered collagen turnover, and/or changes in water content (24, 25). Prior work focusing primarily on the Achilles tendon has reported post-stroke alterations in tendon morphology and mechanics, including increased thickness, reduced stiffness and Young’s modulus, and indications of disrupted collagen organization and abnormal mechanical behavior (3, 8, 11, 12, 22). The current results suggest that such thickness changes over the rehabilitation period is not restricted to posterior-chain tendons; the patellar tendon and plantar fascia also appear sensitive to the cumulative loading associated with strengthening, gait training and balance exercises. This pattern is consistent with studies demonstrating widespread entheseal and fascial abnormalities at the quadriceps, patellar, Achilles, and plantar fascia insertions in metabolic disease and in stroke cohorts (13–16, 18), and may reflect a similar tendency toward multi-site tendon and fascia involvement in stroke-related deconditioning. A key interpretive issue is whether the observed thickness changes exceed measurement error. Using repeated acquisitions with probe removal and repositioning, MDC₉₅-based responder analyses showed that 46.7% of participants on the paretic side and 37.8% on the non-paretic side exceeded MDC₉₅ for quadriceps thickness. Thus, while group-level changes were robust, individual-level changes were close to the limits of detectability in a subset, underscoring the importance of interpreting small absolute differences alongside SEM/MDC rather than relying on statistical significance alone.

The observed associations between quadriceps tendon thickness change and improvements in both balance (BBS) and activities of daily living (Barthel) may have functional relevance. The quadriceps plays a key role in knee control, weight acceptance, postural stability and sit-to-stand transitions-domains that are strongly represented in these clinical scales (1, 2, 10, 23). Increased tendon thickness might reflect improved neuromuscular engagement and more symmetrical loading patterns during rehabilitation, in line with evidence that muscle–tendon unit properties influence gait mechanics and functional performance (10, 12, 24, 25). Although effect sizes were modest, they are consistent with the notion that peripheral tendon adaptations may accompany central motor recovery and contribute to functional gains (3, 8, 10). Other site–outcome correlations are reported as exploratory in the Supplementary material. In addition, we contextualized functional improvements using published MCID thresholds. In our cohort, a substantial proportion of participants achieved clinically meaningful improvement in balance based on the published BBS MCID threshold (34). In contrast, fewer participants exceeded MCID-based criteria for the Barthel Index (35), which is consistent with the relatively smaller paired changes observed in functional independence over a short inpatient period.

Beyond the quadriceps tendon, our data also showed measurable but smaller changes in the patellar tendon, Achilles tendon and plantar fascia. At baseline, these structures did not exhibit the same degree of between-group difference as the quadriceps tendon, yet rehabilitation was associated with bilateral increases rather than a strictly paretic-extremity–specific response. In MDC₉₅-based responder analyses, the proportion exceeding MDC₉₅ was lower for these sites than for the quadriceps (e.g., ~24–29% for patellar, ~29–38% for Achilles, and ~13–22% for plantar fascia), supporting that individual-level change beyond measurement error was present but less frequent and more heterogeneous. This pattern may reflect a more global adjustment of the lower-extremity tendon–fascia complex to increased ambulatory loading and task practice, consistent with reports of structural alterations in the Achilles tendon and plantar fascia in stroke and other chronic loading conditions (3, 11, 12, 15, 18, 22). The absence of stronger correlations between changes in these structures and BBS or Barthel scores may indicate that our functional scales are less sensitive to plantar flexor- and foot-related contributions to gait and balance, or that remodeling in these regions contributes more to other domains, such as propulsion, endurance or pain, which were not specifically captured by the outcome measures used in this study (10, 12, 24, 25).

Several methodological strengths enhance the relevance of this work, including bilateral assessment of four major lower-extremity structures, the use of standardized ultrasonographic protocols, and the prospective design allowing short-term structural tracking. Integrating tendon thickness with clinical outcomes provides a more comprehensive view of potential peripheral contributions to stroke recovery. To reduce the risk of chance findings, we prespecified Quadriceps tendon thickness as the primary outcome, applied multiplicity control for side-specific secondary thickness comparisons, and limited correlation testing to a small prespecified set. Nevertheless, some limitations should be acknowledged. First, the observational pre–post design without a longitudinal control group or a non-intervention comparator limits causal inference; therefore, the observed changes should be interpreted as time-related change over the inpatient period, potentially reflecting natural recovery, hospitalization-related mobilization, and standard rehabilitation exposure. Second, we provide quantitative context for sample size using observed effects from this cohort. For the primary outcome, the paired effect size was large (Cohen’s dz. = 1.36, 95% CI: 0.95–1.76); under two-sided α = 0.05, a paired design would require approximately n ≈ 9 participants to detect dz. = 0.95 with 80% power, whereas with n = 45 the minimum detectable paired effect is approximately dz. ≈ 0.42 at 80% power, indicating that the primary comparison was well powered for moderate-to-large effects. In contrast, association analyses are more sample-demanding; with n = 45, power is ~80% only for correlations of approximately |ρ| ≥ 0.41, so smaller-to-moderate correlations (e.g., ρ ≈ 0.30–0.35) should be interpreted as preliminary. Third, while B-mode ultrasonography provides accessible information on gross morphology, it cannot characterize collagen organization or mechanical properties, which could be explored using elastography or MRI (17, 22, 36). Fourth, the rehabilitation program was standardized but tendon-specific loading dose was not quantified, limiting mechanistic interpretation. Fifth, time since stroke was heterogeneous, which may modify the relative contribution of spontaneous recovery versus therapy exposure. Sixth, some clinical outcomes are ordinal and may show floor/ceiling effects; in our cohort, MAS demonstrated a marked floor effect, particularly at the hip and knee, limiting sensitivity to detect change over a short interval. Finally, long-term follow-up was not available, and the durability of thickness changes remains unknown.

In conclusion, our findings suggest that tendon morphology in post-stroke hemiplegia is not static but can undergo measurable thickness change over a 4-week inpatient rehabilitation period. Within this cohort, quadriceps tendon thickness appeared particularly prominent, and paralleled modest gains in balance and functional independence. These observations raise the possibility that monitoring tendon adaptations could provide additional insight into peripheral contributions to motor recovery and, in larger and longer-term studies, might help refine individualized rehabilitation strategies.