Direct visualization and measurement of the plantar aponeurosis behavior in foot arch deformation via the windlass mechanism

Abstract

The plantar aponeurosis (PA) is an elastic longitudinal band that contributes to the generation of a propulsive force in the push‐off phase during walking and running through the windlass mechanism. However, the dynamic behavior of the PA remains unclear owing to the lack of direct measurement of the strain it generates. Therefore, this study aimed to visualize and quantify the PA behavior during two distinct foot postures: (i) neutral posture and (ii) windlass posture with midtarsal joint plantarflexion and metatarsophalangeal joint dorsiflexion, using computed tomography scans. Six healthy adult males participated in the experiment, and three‐dimensional reconstruction of the PA was conducted to calculate its path length, width, thickness, and cross‐sectional area. This study successfully visualized and quantified the morphological changes in the PA induced by the windlass mechanism, providing a precise reference for biomechanical modeling. This study also highlighted the interindividual variability in the PA morphology and stretching patterns. Although the windlass posture was not identical to that observed in the push‐off phase during walking, the observed PA behavior provides valuable insights into its mechanics and potential implications for foot disorders.

1. INTRODUCTION

The plantar aponeurosis (PA), a longitudinal band covering the sole of the foot that is primarily composed of type I collagen fibers (Stecco et al., 2013), exhibits a complex structure comprising the medial, central, and lateral parts (Abrahams et al., 2013; Kelikian & Sarrafian, 2011; Standring, 2016). The central segment, the thickest of the three (Ehrmann et al., 2014; Schepsis et al., 1991; Stecco et al., 2013), extends from the calcaneal tuberosity to the proximal phalanges. The PA engages in the windlass mechanism, that is, during dorsiflexion of the metatarsophalangeal (MP) joint, the PA is pulled and wrapped around the heads of the metatarsal bones, leading to elevation of the longitudinal arch of the foot (Hicks, 1954). This winding action creates tension and enhances foot stiffness, thereby facilitating effective generation of a propulsive force during walking and running [but see Welte et al., 2018 and Behling et al., 2023 for alternative arguments suggesting that the windlass mechanism may not actually stiffen the foot]. Additionally, the PA contributes to the storage and release of mechanical energy and plays a pivotal role in achieving efficient locomotion in humans (Ker et al., 1987; Kim & Voloshin, 1995; Stearne et al., 2016). However, the PA is susceptible to excessive mechanical stress during dynamic movements, potentially leading to foot disorders such as plantar fasciitis (Lemont et al., 2003; Wearing et al., 2006). Therefore, understanding the PA behavior during dynamic movements is crucial for unraveling the fundamental biomechanics and motor control of the human foot and clarifying the pathogenic mechanisms underlying foot disorders.

Efforts have been made to construct biomechanical foot models to estimate the strain and tension forces exerted by the PA during activities, for example, walking and running (Caravaggi et al., 2009; Chen et al., 2019; Matsumoto et al., 2022). These models conceptualize the PA as a fan‐shaped bundle of linear springs connecting the calcaneal tuberosity to the proximal phalanges according to the anatomy and mechanical properties obtained from cadaver dissections (Guo et al., 2018; Kitaoka et al., 1994; Sichting et al., 2020; Wright & Rennels, 1964), and medical imaging techniques, such as ultrasonography (Bisi‐Balogun & Rector, 2017; Boussouar et al., 2017; Crofts et al., 2014) and shear wave elastography (Chino et al., 2019; Nozaki et al., 2022; Wang et al., 2019). These models estimate the strain and tension of the PA on the basis of the motion‐captured data of the foot and the stiffness of the PA assumed according to the previous studies (Caravaggi et al., 2009; Caravaggi et al., 2010; Gefen, 2003; Kitaoka et al., 1994). The strain of the PA at toe‐off during walking was estimated to be in the range of 3%–6% (Caravaggi et al., 2009; Caravaggi et al., 2010; Matsumoto et al., 2022), and that during the drop jump was approximately 6%–7% (Matsumoto et al., 2023). The tension forces were successively calculated using the published Young's modulus and cross‐sectional area of the PA (Caravaggi et al., 2009; Caravaggi et al., 2010; Chen et al., 2019). However, the accuracy and precision of path and strain estimations using these models have not been evaluated, primarily because of the absence of direct in vivo measurements of the true path and length of the PA during foot movements.

The PA plays a crucial role in the biomechanics of walking, running, and jumping, particularly through the windlass mechanism. This study therefore aimed to visualize and quantify the PA behavior in two foot postures, neutral and windlass posture during push‐off, by employing computed tomography (CT) scans. The path length, width, thickness, and cross‐sectional area of the PA were measured. We formulated the hypothesis that the path length, width, thickness, and cross‐sectional area of the PA vary between the two postures. The obtained anatomical data may serve as a precise reference for biomechanical modeling, contributing to a deeper understanding of the PA function, mechanical effects, and pathogenetic mechanisms underlying foot disorders such as plantar fasciitis.

2. MATERIALS AND METHODS

2.1. Participants

Six healthy adult males (age, 34.5 ± 2.2 years; height, 173.6 ± 6.8 cm; weight, 68.6 ± 2.9 kg; mean ± standard deviation) participated in the experiment. None of the participants had obvious musculoskeletal disorders or foot deformities. In accordance with the Declaration of Helsinki, all participants provided written informed consent following a detailed explanation of the purpose and risks of the study. This study's experimental procedure was approved by the Office for Life Science Research Ethics and Safety at the University of Tokyo (number: 22‐178).

2.2. CT scan

CT (SOMATOM go.Top 64; Siemens, Munich, Bavaria, Germany) was performed at a tube voltage and current of 120 kVp and 70–110 mA, respectively. CT scans of the right foot were obtained with participants in two different foot postures in the supine position: (1) non‐weight‐bearing neutral posture (Figures 1A and 2) simulated weight‐bearing posture with MP joint dorsiflexion and midtarsal joint plantarflexion (windlass posture, Figure 1B). In the neutral posture, participants were positioned with the lower extremity on a Styrofoam object in a knee‐flexed position to prevent the generation of tensile forces by the gastrocnemius muscle, thereby minimizing stretching of the PA (Figure 1A). The windlass posture was induced by applying a load to the foot while maintaining ankle plantarflexion and slight knee flexion using a calf stretcher (ProStretch Plus; Medi‐Dyne, Texas, USA) (Figure 1B) to replicate the alignment of the foot and lower leg during push‐off as closely as possible in the supine position. Participants were instructed to exert maximum force, pressing their ball of the foot against a calf stretcher with the attached a 45‐degree wedge while maintaining still during the CT scan (Figure 1B). To quantify the arch deformation (the angle of medial longitudinal arch [MLA]) and dorsiflexion of the MP joint due to windlass posture, five infrared‐reflecting markers (diameter: 4 mm) were attached to the following structures before performing CT, according to Leardini et al. (2007): the (1) sustentaculum tali, (2) Achilles tendon attachment, (3) first metatarsal head, (4) first metatarsal base, and (5) first proximal phalangeal head (Figure S1). The MLA angle was determined by the sagittally projected angles formed by the first three markers, while the MP joint angle was determined by the sagittally projected angles formed by the last three markers. Reconstruction was performed using a soft‐tissue kernel (Br36). The slice interval and pixel size of the obtained images were 0.5 mm and 0.4–0.6 mm, respectively. The axis formed by connecting the lowest points of the first metatarsal head and the calcaneal tuberosity was established for each volumetric foot data point. Subsequently, the data were resliced perpendicular to this axis to ensure alignment with the PA path.

FIGURE 1.

CT‐scanned foot postures. Neutral (A) and windlass (B) postures. The windlass posture is a simulated weight‐bearing posture with metatarsophalangeal joint DF and midtarsal joint PF using a calf stretcher. DF, dorsiflexion; PF, plantarflexion.

FIGURE 2.

Segmentation of the proximal (i), middle (ii), and distal (iii) portions of the plantar aponeurosis (PA) on the CT images resliced perpendicular to the PA path.

2.3. Morphological analysis of the PA

To quantify the morphological characteristics of the PA from the CT scan data, semi‐automatic segmentation (Figure 2) was performed using the image analysis software Mimics 24.0 (Materialize, Leuven, Belgium). The CT intensity in Hounsfield units (HU) was sampled from the PA regions in several images. Segmentation was performed using 0–100‐HU thresholds and manually corrected where necessary. The segmented PA region was reconstructed three‐dimensionally using the marching cube algorithm.

On the resliced images, the origin, branch point, via point, point below the metatarsal head, and insertion were visually identified and manually digitized using image analysis software ImageJ (National Institutes of Health [NIH], Bethesda, MD, USA). The PA was divided into two parts: the proximal band from the origin to the branch point and the five distal bands from the branch points to the insertions (Figure 3). The length of the first to fifth PAs (PA1–5, from medial to lateral) was calculated based on the digitized points. The length of the proximal part of the ith PA was calculated as follows:

$$L_{\mathit{\text{prox}}\_i}\mathbf{=}\left|{\mathbf{p}}_{\mathit{br}\_i}\mathbf{-}{\mathbf{p}}_{\mathit{\text{or}}}\right|,$$ (1)

where ${\mathbf{p}}_{\mathit{\text{or}}}$ and ${\mathbf{p}}_{\mathit{br}\_i}$ denote the position vectors of the origin and branch points of the ith PA, respectively. The length of the distal part of the ith PA was calculated as follows:

$$L_{\mathit{\text{dist}}\_i}\mathbf{=}\left|{\mathbf{p}}_{\mathit{ins}\_i}\mathbf{-}{\mathbf{p}}_{\mathit{MH}\_i}\right|\mathbf{+}\left|{\mathbf{p}}_{\mathit{MH}\_i}\mathbf{-}{\mathbf{p}}_{\mathit{via}\_i}\right|\mathbf{+}\left|{\mathbf{p}}_{\mathit{via}\_i}\mathbf{-}{\mathbf{p}}_{\mathit{br}\_i}\right|,$$ (2)

where ${\mathbf{p}}_{\mathit{via}\_i},{\mathbf{p}}_{\mathit{MH}\_i},$ and ${\mathbf{p}}_{\mathit{ins}\_i}$ are the position vectors of the via point, point below the metatarsal head, and insertion of the ith PA, respectively. The total length of the ith PA was calculated as the sum of the lengths of its proximal and distal parts.

$$L_{\mathit{\text{total}}\_i}=L_{\mathit{\text{prox}}\_i}+{\mathrm{L}}_{\mathit{\text{dist}}\_i}.$$ (3)

FIGURE 3.

Definition of the origin (${\mathbf{p}}_{\mathit{\text{or}}}$), branch point (${\mathbf{p}}_{\mathit{br}\_i}$), via point (${\mathbf{p}}_{\mathit{via}\_i}$), point below the metatarsal head (${\mathbf{p}}_{\mathit{MH}\_i}$), and insertion (${\mathbf{p}}_{\mathit{ins}\_i}$) to calculate the lengths of the proximal ($L_{\mathit{\text{prox}}\_i}$) and distal ($L_{\mathit{\text{dist}}\_i}$) parts of the ith PA. PA, plantar aponeurosis.

The length of the proximal part was normalized to the total length in the neutral posture.

Additionally, the thickness, width, and cross‐sectional area of the PA in the neutral posture were measured on the resliced images corresponding to the proximal 10%, middle 50%, and distal 90% of the proximal band using ImageJ software.

2.4. Kinematic analysis of the foot

To quantify foot posture and motion due to changes in foot posture, the angles of the MLA and MP joint were calculated on the basis of the markers. The three‐dimensional (3D) surface models of the markers were generated from the CT data and the positions of the marker centers were calculated by spherical approximation. Moreover, the 3D bone surface models were generated and the 3D rotation angles of the talocalcaneal (TC), talonavicular (TN), calcaneocuboid (CC), and Lisfranc joints (the first to third metatarsals relative to the navicular, and the fourth and fifth metatarsals relative to the cuboid) were calculated. The local anatomical coordinate systems of the talus, calcaneus, navicular, cuboid, and five metatarsals were defined as described by Gutekunst et al. (2013) and Negishi et al. (2021). The definitions of the bone coordinate systems and the anatomical landmarks used to define the coordinate systems are provided in Table S1 and Figure S2, respectively. The 3D rotation angles of all the joints are described by the yxz Euler angles. The rotational angles around the y‐, x‐, and z‐axes represented plantarflexion–dorsiflexion, inversion–eversion, and adduction–abduction, respectively. The bony landmarks of each bone in the neutral posture were digitized manually using Geomagic Design X 2020 (3D Systems, Rock Hill, SC, USA), whereas those in the windlass posture were calculated using the homogeneous transformation matrix obtained from an iterative closest point registration.

2.5. Statistical analysis

To compare the statistical differences in the length, thickness, width, and cross‐sectional area of the PA and the angles of the MLA and foot joints between the two foot postures, we performed a paired t‐test if normality was confirmed using the Kolmogorov–Smirnov test. If normality was violated, the signed‐rank test was used for statistical comparison. To assess statistical differences in the thickness, width, and cross‐sectional area across different positions of the PA, a one‐way repeated‐measures analysis of variance was employed when normality was confirmed. When normality was violated, the Friedman test was used for statistical comparisons. Multiple tests were conducted using the Bonferroni method for parameters that displayed significant differences among the PA positions. All statistical tests were performed at a significance level of 5% using R software (version 4.3.0; R Foundation for Statistical Computing, Vienna, Austria). Furthermore, to elucidate the strength of the observed difference in the data with a small sample size, effect size was determined using Cohen's d statistics when a significant difference was confirmed. The effect size is generally considered large if d ≥ 0.8 (Sullivan & Feinn, 2012).

2.6. Assessment of measurement reproducibility of morphological parameters of the PA

The intra‐observer correlation coefficient for the 38 morphological measurements of the PA, that is, the proximal and distal lengths of five PAs in the neutral and windlass postures, and the thicknesses, widths, and cross‐sectional areas at 10%, 50%, and 90% of the proximal PA in the neutral and windlass postures, was assessed by having one examiner reassess the CT data for Subject A after an interval of more than 4 months. The inter‐observer reproducibility was also assessed by having a different experienced examiner perform independent segmentation of the CT data for Subject A, and the correlation coefficient between the two examiners was calculated. Calculation of the correlation coefficient was performed in R software.

3. RESULTS

The intraobserver and interobserver correlation coefficients for the measurements of the morphological measurements of the PA were 0.999 (95% coefficient interval, 0.999–1) and 0.999 (95% coefficient interval, 0.998–1), respectively. The absolute intraobserver differences of the length, thickness, width, and cross‐sectional area were 0.6 ± 0.6, 0.1 ± 0.1, 0.9 ± 0.5 mm, and 2.0 ± 1.6 mm², respectively. Similarly, the absolute interobserver differences were 1.5 ± 1.1, 0.2 ± 0.2, 1.6 ± 0.8 mm, and 1.1 ± 0.8 mm², respectively. These findings suggest that the measurement of the morphology of the PA was highly reproducible.

The 3D surface reconstruction of the PA with foot bones for a representative subject (Subject A) is shown in Figure 4. The 3D surface models of all six participants are presented in Figures S3–S5. The PA bundle was longitudinally oriented from the calcaneal tuberosity to the toes, passing beneath the intermediate and lateral cuneiform and branching into five bands around the Lisfranc joint (Figure 4A). In the windlass posture, the PA underwent longitudinal stretching, moved slightly inferiorly as the metatarsal heads pressed it down through the windlass mechanism, and was wrapped around the metatarsal head upward to attach it to the dorsiflexed proximal phalanges (Figure 4B,C).

FIGURE 4.

Three‐dimensional model of the plantar aponeurosis (PA). Sagittal and frontal views in the neutral (A) and windlass (B) postures. Comparisons of the PA paths between two postures when the position and orientation of the calcaneus were aligned between the neutral (red) and windlass (yellow) postures (C).

Table 1 shows the mean angles of the MLA and MP, TC, TN, CC, and Lisfranc joints during the two postures. The rotational angles were positive for dorsiflexion, eversion, and abduction. The joint angle of the Lisfranc joint around eversion–inversion and abduction–adduction was not represented, as the joint movements between the two postures were in the dorsiflexion direction. During the transition from the neutral to windlass posture, the MP joint dorsiflexed to approximately 50° (p < 0.01, d = 2.9), whereas the angle of the MLA decreased by approximately 5° and that of the Lisfranc joint plantarflexed by approximately 5° (N‐1MT, p = 0.024, d = 1.3; N‐2MT, p < 0.01, d = 1.7; N‐3MT, p < 0.01, d = 2.4; C‐4MT, p < 0.01, d = 2.4; and C‐5MT, p < 0.01, d = 3.0) (Table 1), indicating that the longitudinal arch of the foot increased as the MP joint dorsiflexed owing to the windlass mechanism. The CC and TN joints were adducted by 9° and 3° (p = 0.049, d = 1.1), respectively, and the TC joints were inverted by 4° (Table 1).

TABLE 1.

Comparisons of the angles of the MLA and foot joints between the neutral and windlass postures.

	Neutral		Windlass		Difference		p‐value	Effect size
Angle [°]	Mean	S.D.	Mean	S.D.	Mean	S.D.
MLA	160.7	(7.6)	156.2	(9.4)	−4.5	(6.5)
MP**	9.7	(5.2)	60.8	(15.4)	51.1	(17.4)	p < 0.01	d = 2.9
Pf(−)/Df(+)
N‐1MT*	16.5	(7.5)	12.5	(9.7)	−4.0	(3.1)	p = 0.024	d = 1.3
N‐2MT*	12.8	(9.6)	9.7	(10.6)	−3.1	(1.9)	p < 0.01	d = 1.7
N‐3MT**	17.9	(9.2)	15.1	(9.2)	−2.8	(1.2)	p < 0.01	d = 2.4
C‐4MT**	−4.0	(7.3)	−9.4	(7.2)	−5.4	(2.3)	p < 0.01	d = 2.4
C‐5MT**	2.5	(6.8)	−4.2	(6.0)	−6.7	(2.3)	p < 0.01	d = 3.0
TN	−31.0	(7.8)	−30.9	(7.0)	0.1	(3.5)
CC	−36.7	(6.3)	−37.6	(6.6)	−0.9	(1.2)
TC	34.8	(4.5)	33.3	(4.5)	−1.5	(1.4)
Inv(−)/Ev(+)
TN	17.4	(10.2)	17.5	(10.7)	0.1	(1.6)
CC	15.1	(4.3)	14.1	(5.9)	−1.1	(2.1)
TC	1.6	(11.3)	−1.9	(11.5)	−3.5	(3.4)
Add(−)/Abd(+)
TN	21.2	(7.1)	12.5	(10.9)	−8.7	(8.7)
CC*	16.6	(6.7)	13.3	(7.2)	−3.3	(3.1)	p = 0.049	d = 1.1
TC	16.3	(4.0)	15.6	(3.8)	−0.7	(1.2)

Note: Joint angles were positive for Df, Ev, and Abd.

Abbreviations: Abd, abduction; Add, adduction; CC, calcaneocuboid; C‐iMT, the ith metatarsals relative to the cuboid; Df, dorsiflexion; Ev, eversion; Inv, inversion; MLA, medial longitudinal arch; MP, metatarsophalangeal; N‐iMT, the ith metatarsals relative to the navicular; Pf, plantarflexion; TC, talocalcaneal; TN, talonavicular.

^* p < .05;

^** p < .01.

As shown in Figure 5A, the natural length of the PA was the largest for the second digit and decreased as it approached the medial and lateral sides. In the windlass posture, all the PAs were significantly more stretched than in the neutral posture (Figure 5A, PA1, p < 0.001, d = 3.4; PA2, p < 0.001, d = 2.8; PA3, p = 0.003, d = 2.2; PA4, p = 0.008, d = 1.7; and PA5, p = 0.047, d = 1.1), but the strain of the PAs tended to be larger on the medial side (approximately 5%) than on the lateral side (approximately 2%) (Figure 5B). While transitioning from a neutral stance to a windlass posture, the present study observed that the proximal band became thinner particularly in the middle and distal region (middle, p = 0.044, d = 1.1; distal, p = 0.013, d = 1.5) and narrower particularly in the (proximal, p = 0.012, d = 1.6; middle, p = 0.047, d = 1.1) (Figure 6A). Therefore, the cross‐sectional area of the proximal band was generally smaller in the windlass posture than the neutral posture (distal, p = 0.023, d = 1.3; proximal, p = 0.041, d = 1.1).

FIGURE 5.

Comparisons of the lengths of the PA between the neutral and windlass postures (A) and the strain of the PA when in the windlass posture (B). Mean (solid line with circle plots = neutral, dashed line with triangle plots = windlass) ± standard deviation (red and yellow bands, respectively) * p < 0.05. ** p < 0.01. PA, plantar aponeurosis.

FIGURE 6.

Comparisons of the thickness, width, and cross‐sectional area of the PA measured at the proximal 10%, middle 50%, and distal 90% of the length of the proximal band ($L_{\mathit{\text{prox}}}$), between the neutral and windlass postures (A), and among different locations when in the neutral posture (B). Mean (solid line with circle plots) ± standard deviation (red band). * p < 0.05. ** p < 0.01. ⁺ p < 0.05. ⁺⁺ p < 0.01.

The proximal band was significantly wider distally (proximal vs. middle, p < 0.001, d = 4.1; proximal vs. distal, p < 0.001, d = 4.9) (Figure 6B). It was significantly thinner distally (proximal vs. distal, p = 0.042, d = 1.2), but the difference was very small. Therefore, the cross‐sectional area was significantly larger in the distal portion than in the middle and proximal portions (middle, p = 0.002, d = 7.0; proximal, p = 0.002, d = 1.8).

The proximal band constituted approximately 65% of the total PA length (Figure 7). However, there were large interindividual variations in the morphology of the PA. The percentage was as low as 55% (Subject B, PA1), but as high as 75% (Subject B and E, PA5) (Figure 7). There were also large interindividual variations in the stretching patterns of the PA. The proximal band of the PA was largely stretched in Subjects A, C, and D, but not in the other three subjects, indicating that the distal PAs were more stretched in these subjects than in the others (Figure 7).

FIGURE 7.

Length of the proximal part of the PA in the neutral (solid line with circle plots) and windlass (dashed line with triangle plots) postures. PA, plantar aponeurosis.

4. DISCUSSION

This study successfully visualized and quantified the PA behavior in two foot postures, neutral and windlass, by employing CT scans. To our best knowledge, this is the first study to visualize changes in the morphology of the PA due to the windlass mechanism in vivo. The obtained anatomical data serve as an accurate reference for biomechanical modeling. Previous foot models represented the PA as a fan‐shaped bundle of linear springs connecting the calcaneal tuberosity to the proximal phalanges (Caravaggi et al., 2009; Chen et al., 2019; Matsumoto et al., 2022), but the present study demonstrated that the PA takes on a broom‐shaped configuration, with the proximal PA comprising a single band and the remaining distal PA diverging into five bands stemming from the proximal band. Such differences in the path between the true PA and conventional fan‐shaped PA models may introduce errors in estimating the strain and tensile force generated by the PA using such models. Developing a biomechanical foot model that considers the actual path of the PA could improve the accuracy of estimating the kinematics and kinetics of the foot in motion and contribute to our understanding of the functions and mechanical effects of the PA and the pathogenic mechanisms of foot disorders, including plantar fasciitis.

The present study found that the path length, width, thickness, and cross‐sectional area of the PA significantly increased, narrowed, thinned, and shrunk during the transition from the neutral to windlass posture. The stretch of the PA in the windlass posture was consistent with previous reports (Caravaggi et al., 2009; Caravaggi et al., 2010; Matsumoto et al., 2022), but the present study for the first time observed the changes in the PA cross‐sectionally between the neutral and windlass postures. When a material is stretched longitudinally, it tends to contract laterally. Therefore, the changes observed in the dimensions of the PA during the transition from a neutral to windlass posture can be explained by the Poisson's effect, where longitudinal stretching induces lateral contraction and changes in width, thickness, and cross‐sectional area.

The present study highlights the presence of interindividual variability in the morphology of the PA in both neutral and windlass postures. Our study revealed that on average, the proximal band constituted approximately 65% of the total PA length. However, this ratio can vary considerably, ranging from approximately 55% to 75%, depending on the individual subjects and rays. In the windlass posture, while all subjects exhibited PA stretching, the degree varied, and some subjects showed more stretching in the proximal band than in the distal bands, whereas others displayed the opposite pattern. This indicates that constructing a biomechanical foot model with a generic PA geometry may not be sufficient to reliably estimate the strain and tensile force generated by the PA during movements. Using CT or magnetic resonance imaging for subject‐specific modeling of PA geometry or adapting generic PA models through ultrasound imaging may be essential for accurately estimating the dynamic behavior of the PA.

Measurements of the proximal band of the PA revealed a thickness of 2.5 mm and a width of 10–20 mm (Figure 6B). These values align well with the established figures in published literature: 2–4 mm in thickness (Berkowitz et al., 1991; Cardinal et al., 1996; Chen et al., 2014; Ehrmann et al., 2014; Stecco et al., 2013; Tsai et al., 2000; Vohra et al., 2002) and 15–20 mm in width (Bisi‐Balogun & Rector, 2017; Chen et al., 2014; Kelikian & Sarrafian, 2011; Stecco et al., 2013). The cross‐sectional areas of the proximal PA were approximately 30 mm² for the proximal and middle portions and 50 mm² for the distal part of the proximal band. In comparison, a single published comparative value was 69.2 mm² (Bisi‐Balogun & Rector, 2017). Discrepancies may arise from variations in study subjects and measurement techniques. Although Bisi‐Balogun and Rector (2017) used ultrasound imaging, we used high‐resolution CT scanning. Additionally, our approach involved reslicing the volumetric image of the foot to align images with the PA path, potentially resulting in a more accurate perpendicular representation of the cross‐sectional area than that on possibly oblique images from ultrasound imaging in a previous study.

This study noted that the strain induced in the PA during the windlass posture was larger on the medial side than on the lateral side, and the generated strain fell within the range of 2%–5%, which was also consistent with that reported in previous studies (3%–6%) (Caravaggi et al., 2009; Caravaggi et al., 2010; Matsumoto et al., 2022). Furthermore, our observations suggest that the stress developed in the proximal band of the PA during the windlass posture likely presented a larger magnitude than during the neutral posture and also that in the proximal half than in the distal half because of the significant difference in the cross‐sectional area between the two postures and between the proximal and distal bands (Figure 6). This stress distribution corresponds to the clinical observation that plantar fasciitis caused by excessive and repetitive mechanical stress tends to develop around the heel rather than around the middle or distal bands of the PA (Thompson et al., 2014; Tu, 2018; Tu & Bytomski, 2011).

It must be noted, however, that the windlass posture in the present study was not identical to that observed in the push‐off phase of walking. The angles of the MLA and MP joints in the windlass posture in the present study (156°and 51°, respectively, on average) were confirmed to be similar to those measured in the push‐off phase during walking (160°and 35°, respectively, on average) (Leardini et al., 2007). However, in the present study, the movements of the TC, TN, and CC joints between the neutral and windlass postures were 4° of inversion, 9° of adduction, and 3° of adduction in the present study, whereas those observed in actual human walking based on bone pins or biplane fluoroscopy between the flatfoot and push‐off phases were 5°–10° of inversion, 15° of abduction, and 10° of abduction (Arndt et al., 2004; Koo et al., 2015; Lundgren et al., 2008). Although direct comparisons were not feasible, the observed differences in hindfoot bone movements suggested that the windlass posture in this study was similar, but not precisely identical, to the foot posture during push‐off during walking. This distinction may be attributed to the application of a reaction force on the heel that was in contact with the stretcher during the windlass posture in the present study. Therefore, it is essential to note that the windlass posture in this study is not identical to that observed during actual human walking. Investigations into the full‐weight‐bearing windlass posture should be conducted using an upright CT scanner (Jinzaki et al., 2020) in future studies when opportunities arise. Nevertheless, the kinematics of the foot observed here were found to be consistent with those of a previous study that performed a CT scan of the foot under simulated weight‐bearing with 30° of hallucal MP dorsiflexion (Kihara et al., 2023) using a loading device (Kimura et al., 2015).

One limitation of the current study was the inability to recruit female participants for the experiment, despite being open to both sexes. This limitation may be attributed to concerns related to x‐ray exposure, a component of the present study. However, it is recognized that sex‐related differences in foot morphology, including that of PA, can be substantial (Huerta & García, 2007; Krauss et al., 2008; Nozaki et al., 2020; Nozaki et al., 2021; Stankovic et al., 2018; Taş, 2018; Tümer et al., 2019). The findings obtained solely from male participants in this study should be validated in females whenever opportunities arise in future research.

5. CONCLUSION

This study successfully visualized and quantified the behavior of the PA in neutral and windlass postures using CT scans, marking the first in vivo visualization of morphological changes induced by the windlass mechanism. The obtained anatomical data provide a precise reference for biomechanical modeling. The study also highlighted interindividual variabilities in both the PA morphology and stretching patterns, emphasizing the necessity for subject‐specific modeling or ingenious adaptation of generic models to ensure accurate estimation of the dynamic PA behavior. Although the windlass posture in this study was not identical to that of the push‐off phase during walking, the observed foot kinematics remained consistent with those of previous studies, contributing valuable insights into understanding the PA mechanics and potential implications for foot disorders.

Supporting information

Data S1. Supporting Information.

Matsumoto, Y. , & Ogihara, N. (2025). Direct visualization and measurement of the plantar aponeurosis behavior in foot arch deformation via the windlass mechanism. Clinical Anatomy, 38(2), 116–126. 10.1002/ca.24171

DATA AVAILABILITY STATEMENT

The datasets generated and/or analyzed during the current study and the custom‐made software are available from the corresponding author upon reasonable request.