Abstract
Background.
Several multi-segment foot models have been developed to evaluate foot and ankle motion using skin-marker motion analysis. However, few multi-segment models have been evaluated against a reference standard to establish kinematic accuracy.
Research Question.
How accurately do skin-markers estimate foot and ankle motion for the modified Shriners Hospitals for Children Greenville (mSHCG) multi-segment foot model when compared against the reference standard, dual fluoroscopy (DF), during gait, in asymptomatic participants?
Methods.
Five participants walked overground as full-body skin-marker trajectory data and DF images of the foot and shank were simultaneously acquired. Using the mSHCG model, ankle and midfoot angles were calculated throughout stance for both motion analysis techniques. Statistical parametric mapping assessed differences in joint angles and marker positions between skin-marker and DF motion analysis techniques. Paired t tests, and linear regression models were used to compare joint angles and range of motion (ROM) calculated from the two techniques.
Results.
In the coronal plane, the skin-marker model significantly overestimated ROM (p = 0.028). Further, the DF model midfoot ROM was significantly positively related to differences between DF and skin-marker midfoot angles (p = 0.035, adjusted R2 = 0.76). In the sagittal plane, skin-markers underestimated ankle angles by as much as 7.26°, while midfoot angles were overestimated by as much as 9.01°. However, DF and skin-marker joint angles were not significantly different over stance. Skin-markers on the tibia, calcaneus, and fifth metatarsal had significantly different positions than the DF markers along the direction of walking for isolated portions that were less than 10% of stance. Euclidean distances between DF and skin-markers positions were less than 9.36 mm.
Significance.
As the accuracy of the mSHCG model was formerly unknown, the results of this study provide ranges of confidence for key angles calculated by this model.
Keywords: multi-segment foot model, gait, dual fluoroscopy, soft tissue artifact
1. Introduction
Skin-marker motion capture is an integral technology to study pathological gait, develop surgical plans, and quantify the biomechanical efficacy of surgery and physical therapy. The foot was initially modeled as a single segment [1, 2], primarily due to hardware limitations. However, this simplification may omit important joint functions or yield inaccurate kinematic measurements. Accordingly, multi-segment foot models were developed to independently measure the motion of various segments in the foot, such as the hindfoot, forefoot, and hallux [3–7].
While skin-marker motion capture provides relatively efficient estimates of human movement, markers move relative to the underlying skeleton that they are meant to track, leading to a phenomenon known as soft tissue artifact (STA). To date, the contribution of STA on multi-segment foot model kinematic calculations has been limited to evaluations of repeatability [8, 9], reliability [10–12], and/or variability [3, 9, 13], primarily in the context of marker placement within and across data acquisition sessions or sites. Studies have used cadavers to evaluate the effect of STA on multi-segment foot model accuracy by comparing skin-marker results to predetermined positions of the bones [14] or bone-mounted marker clusters [15]. However, cadaveric assessments do not account for in vivo forces or motion. In vivo evaluations have been conducted using bone pins [16]; however, in addition to being invasive, bone pins may interfere with normative motion. Single plane fluoroscopy has been used to evaluate the accuracy of multi-segment foot models [17], but this method can yield large errors for out-of-plane motions [18].
High-speed dual fluoroscopy (DF) has been shown to measure in vivo bone motion in the foot and ankle with sub-millimeter and sub-degree errors [19]. A recent study by Kessler et al. employed DF to evaluate a multi-segment foot model and reported good agreement between DF and skin-marker for the first metatarsal and sagittal plane measurements of the longitudinal arch [20]. However, evaluation of overall forefoot skin-marker motion compared to DF is lacking, specifically in the coronal and transverse planes.
The goal of this study was to use DF as a reference standard during gait analysis to evaluate three-dimensional (3D) ankle and midfoot kinematic accuracy and STA of the modified Shriners Hospitals for Children Greenville (mSHCG). The mSHCG multi-segment model was developed to evaluate 3D ankle and midfoot kinematic and kinetic differences between healthy and pathologic feet [3, 13, 21]. We chose to evaluate the accuracy of this model because it is often used for clinical and research applications. Previous research has demonstrated that the mSHCG model has less than 8° of variability within and between clinicians across the various foot segments of children with healthy and planovalgus feet [3, 13]. Despite this evaluation of variability, the accuracy of this model has not been evaluated against a reference standard.
2. Materials and Methods
2.1. Participant Recruitment
Following Institutional Review Board approval (University of Utah IRB # 65620) and informed consent, healthy, young adult participants were screened for abnormalities and osteoarthritis in the foot and ankle, as well as for a history of back or lower limb pain or surgery. An experienced fellowship-trained foot and ankle surgeon screened weight-bearing radiographs. The Kellgren-Lawrence (KL) scale was used to evaluate degenerative changes in the foot and ankle joints of each participant [22, 23]. Participants were excluded from this study if a foot or ankle joint had a KL score > 1. Based on these requirements, five healthy control participants were enrolled in this study (3 female, 2 male; age: 27.8 ± 3.1 yo; BMI: 22.9 ± 2.6 kg/m2).
2.2. Experimental Paradigm and Activities
A custom high-speed DF system was positioned adjacent to two in-ground force plates (AMTI, Watertown, MA, USA) and within the capture volume of a 10-camera near-infrared motion analysis system (Vicon Motion Systems, Oxford, UK) (Figure 1). The DF system was aligned temporally and spatially to the motion analysis system using an external trigger and a calibration cube, respectively [24]. DF images and skin-marker trajectories were acquired at 200 Hz. Ground reaction forces were acquired at 1000 Hz. Participants walked barefoot at a self-selected speed to promote natural movement. Skin-marker trajectories were acquired throughout the entire activity. However, DF images were only acquired during stance since this was when the foot was in the combined field-of-view of both fluoroscopes.
Figure 1:
Skin-markers employed for the modified Shriners Hospitals for Children Greenville (mSHCG) model (top) and the combined dual fluoroscopy and skin-marker motion capture experimental setup (bottom). Top: The markers labeled in cyan were used for the static calibration trial to define (i.e., establish a local coordinate system for) the tibia, hindfoot, and forefoot segments of the mSHCG model. Following the static trial, the markers labeled in cyan were removed. The three-dimensional positions of the markers labeled in black were used to determine the position and orientation of the tibia, hindfoot, and forefoot segments during walking. Bottom: Overground experimental setup of a participant during data acquisition. E1 = emitter 1; E2 = emitter 2; II1 = image intensifier 1; II2 = image intensifier 2.
2.3. Skin-marker Trajectory Acquisition and Processing
The mSHCG marker set was employed to quantify ankle and foot kinematics as described previously [3, 13] (Figure 1). Briefly, this marker set defines and enables the 3D positional tracking of the thorax/trunk, pelvis, and bilateral thigh, shank, hindfoot, forefoot, and hallux segments (Supplemental Table 1). Additional markers were added to the mSHCG marker set to provide redundancies in tracking the position of the trunk, thigh, and shank segments, since placement of the DF system could obstruct the view of the motion capture cameras. Bilateral shoulder markers and thigh marker clusters were added to the mSHCG marker set. The upper and lower shank markers of the mSHCG marker set were incorporated into marker shank clusters (Figure 1).
Skin-marker trajectories were acquired, reconstructed, labeled, and gap filled using Vicon Nexus (v. 2.5). A previously described BodyLanguage pipeline [3] was used to filter marker trajectories with a fourth-order Butterworth filter with a cutoff frequency of 10 Hz (Vicon Nexus). Gait events (foot-strike and foot-off) were automatically detected from the force plate data (Vicon Nexus).
2.4. Computed Tomography and Dual Fluoroscopy Data Processing
A computed tomography (CT) scan was acquired (SOMATOM Definition AS; Siemens Medical Solutions, Malvern, PA) of each participant to generate data necessary to determine bone position from the DF images. CT images of the mid-tibia through the toe tips were acquired at 1.0-mm slice thickness, 341 ± 35.1 mm field of view, 512 × 512 acquisition matrix, 100 kV, 20–185 mAs, with CareDose™ to modulate tube current. The DF emitter beam energy settings were determined for each participant at the beginning of data acquisition to qualitatively optimize the fluoroscopic images. Energy settings ranged from 66–77 kVP and 1.4–2.2 mAs. CT image processing and model-based tracking were performed as described previously [19, 25]. Briefly, the tibia, talus, calcaneus, first metatarsal, and fifth metatarsal were segmented from the CT images (Amira 5.5, Visage Imaging, San Diego, CA) and used to create ray-traced projections of each bone through the CT volume. Semi-automatic model-based tracking determined the position and orientation (i.e., trajectory) of each bone through the optimization of pixel intensity agreement between the bone projection and the fluoroscopic images [26].
A custom code was written in MatLab (v. 2017b; The MathWorks, Inc., Natick, MA, USA) to create virtual markers with trajectories derived from the DF-tracked bones (Supplemental Figure 1). The skin-markers on the calibration cube were embedded with steel beads at the center, which made the markers visible to the DF and Vicon systems. This combined visibility enabled co-localization of each marker in the DF and Vicon global coordinate systems, which was used to determine transformation matrices between the two coordinate systems [24]. For each fluoroscopy image, model-based tracking defined a transformation matrix between the bone CT coordinate system and the DF global coordinate system to determine the position and orientation of the bone. The positions of the shank, hindfoot, and forefoot skin-markers during the static trial were transformed from the Vicon global coordinate system to the DF global coordinate system using the calibration cube. The marker positions were transformed from the DF global coordinate system to their respective CT bone coordinate systems on the tibia, calcaneus, and first or fifth metatarsals using the model-based tracking transformation matrices from the static trial to define virtual markers on each bone in the CT coordinate system. Next, the respective DF-tracked bone trajectories from the walking trial were applied to these virtual makers to transform the virtual markers from the CT coordinate system to the DF global coordinate system. Finally, the trajectories of the virtual markers were transformed from the local DF global coordinate system to the Vicon global coordinate system [24] to create DF trajectories.
2.5. Dual Fluoroscopy and Skin-marker mSHCG Model Implementation
The mSHCG segments were defined for the static trial using a BodyLanguage script (Vicon Nexus) [3]. The static standing trial with calibration markers established anatomical segment definitions for the walking trials.
To compare DF and skin-marker kinematics and quantify skin-marker STA during gait, the mSHCG model was separately applied to DF and skin-marker trajectories. As less temporal trajectory data were available for DF, the skin-marker trials were truncated to the last tracked DF frame. For skin-marker kinematics, skin-marker trajectories were applied to the mSHCG model (SMmodel) to determine segment motion (Vicon Nexus) [3, 13]. For DF kinematics, the shank, hindfoot, and forefoot skin-marker trajectories of the walk trial were replaced with the DF virtual marker trajectories (Visual3D v. 6.2.1, C-Motion, Germantown, MD, USA). Then, the virtual marker trajectories, along with the original trunk, pelvis, thigh, and hallux skin-markers were applied to the mSHCG model (DFmodel) to determine segment motion (Vicon Nexus).
2.6. Data and Statistical Analyses
Dorsi/plantarflexion, inversion/eversion, and internal/external rotation angles for the SMmodel and DFmodel were calculated for the ankle, between the calcaneus and tibia, and the midfoot, between the first and fifth metatarsals, which comprised the forefoot, and calcaneus (Vicon Nexus). All statistical analyses were performed using a custom MatLab script (v. 2017b). Foot-strike and toe-off events were used to normalize the kinematic data to the stance phase of gait.
The mean (± 1 standard deviation) ankle and midfoot angles were calculated for the DFmodel and SMmodel for the portions of stance in which DF data were available for all participants and visualized to evaluate kinematic differences between DF and skin-marker kinematics. Previous work [3] recommended differences in assessed pathological conditions should not be considered clinically significant if less than 4°, therefore SMmodel ankle and midfoot mean angles were considered clinically similar unless differences exceeded 4°. To determine how DFmodel and SMmodel angles differed throughout stance, one-dimensional statistical parametric mapping (SPM) was applied (v. 0.4, MATLAB-based open source software: www.spm1d.org). Specifically, SPM employed a two-sided paired t test to evaluate statistical differences between DFmodel and SMmodel ankle and midfoot kinematics for each point of normalized stance.
The DFmodel and SMmodel mean range of motion (ROM) and 95% confidence intervals (CIs) were calculated for the ankle and midfoot angles in each plane. A paired t test was used to compare DFmodel and SMmodel ROM. A linear regression model was used to determine if greater DF ROM was associated with greater angular differences between the DFmodel and SMmodel and reported as (p value; adjusted R2 value). Significance was set at p < 0.05.
Positions of the virtual markers and skin-markers were compared to characterize STA. SPM with a two-sided paired t test evaluated statistical differences between virtual marker and skin-marker positions in each direction (X = direction of walking, Y = medial-lateral direction, Z = vertical direction). The Euclidean distance was calculated between the skin-marker and virtual marker positions.
3. Results
3.1. Differences between DFmodel and SMmodel Kinematics
Comparison of DFmodel and SMmodel ankle and midfoot angles via SPM paired t test revealed no statistically significant differences. Differences between the DFmodel and SMmodel for mean ankle angles and coronal and transverse midfoot angles were less than 4° throughout stance (Figure 2). SMmodel mean sagittal midfoot angles were overestimated, but were only greater than 4° between 26–28% and 81–85% of stance (Figure 3).
Figure 2:
Mean ankle (left) and midfoot (right) dorsi/plantarflexion (top), inversion/eversion (middle) and internal/external rotation (bottom) angles during walking. Gait was normalized to stance phase. Extension, eversion, and external rotation were negative. Black lines and shading = mean angles determined using dual fluoroscopy ± 1 standard deviation (SD); orange lines and shading = mean angles determined using skin-markers ± 1 SD.
Figure 3:
Differences between SMmodel and DFmodel calculations of ankle (left) and midfoot (right) dorsi/plantarflexion (top), inversion/eversion (middle), and internal/external rotation (bottom) angles during walking. Gait was normalized to stance phase. Extension, eversion, and external rotation were negative. Black lines = mean; shaded regions = ± 1 standard deviation.
On a per-participant basis (Figure 4), maximum differences between the DFmodel and SMmodel angles were largest in the sagittal plane. In the sagittal plane, skin-markers underestimated ankle angles by as much as 7.26°, while midfoot angles were overestimated by as much as 9.01°.
Figure 4:
Ankle (left) and midfoot (right) dorsi/plantarflexion (top), inversion/eversion (middle) and internal/external rotation (bottom) for each participant. Gait was normalized to stance phase. Solid lines = angles determined using dual fluoroscopy; dashed lines = angles determined using skin-markers. Each color represents a different participant.
3.2. DFmodel and SMmodel Range of Motion
Qualitatively, mean ROM and 95% CIs were similar between the SMmodel and DFmodel (Figure 5). However, SMmodel coronal plane midfoot ROM was significantly larger than DFmodel ROM (p = 0.028). Additionally, DFmodel midfoot ROM in the coronal plane was significantly related to mean angular differences between the DFmodel and SMmodel, with a strong positive relationship (p = 0.035, adjusted R2 = 0.76).
Figure 5:
Mean DFmodel (black) and SMmodel (orange) ankle and midfoot dorsi/plantarflexion (left), inversion/eversion (center) and internal/external rotation (right) range of motion (ROM) during walking. Error bars indicate ± 95% confidence intervals; * indicates statistically significant difference (p = 0.028).
3.3. Soft Tissue Artifact
Statistically significant differences between DFmodel and SMmodel marker positions were found for the MT5H, MCAL, TIBU, and TIBUL markers (see marker labels in Figure 1) along the X axis (direction of walking) (Figure 6). DFmodel and SMmodel TIBU positions were significantly different in the X direction for 21–22% (p < 0.001), 25–26% (p = 0.006), and 43–44% (p = 0.004) of stance. DFmodel and SMmodel TIBL positions were significantly different in the X direction for 62–64% (p < 0.001) of stance. DFmodel and SMmodel MCAL positions were significantly different in the X direction for 40–47% (p < 0.001) and 56–61% (p = 0.001) of stance. DFmodel and SMmodel MT5H positions were significantly different in the X direction for 1–8% (p = 0.01) and 83–84% (p = 0.046) of stance.
Figure 6:
Mean and standard deviation (SD) of dual fluoroscopy virtual marker (black) and skin-marker (red) positions along the X axis (direction of walking) that were significantly different (p ≤ 0.046) for portions of stance, denoted by *.
Differences between DFmodel and SMmodel marker positions on the shank, hindfoot, and forefoot were as high as 9.36 mm for the MT5H marker. For markers on the shank and hindfoot, differences in marker positions were less than 6.63 mm and 4.84 mm, respectively.
4. Discussion
This study used segment motion derived from DF and skin-markers to calculate differences in joint angles and marker positions and evaluate the accuracy of the mSHCG model. The SMmodel joint angles were not significantly different from the DFmodel throughout stance. Additionally, ROM was often well-matched between the DFmodel and SMmodel; only the SMmodel coronal plane midfoot ROM was statistically significantly greater than the DFmodel. Skin-marker positions were fairly accurate; the only significant differences between the skin-marker and DF-tracked virtual marker positions were in the direction of walking for minimal portions of stance. This may be attributable to skin deformity as a result of deceleration forces. These results provide investigators with a range of anticipated errors for the mSHCG model, which should facilitate improved interpretation of model results.
There is an important distinction between statistical and clinical significance that should be considered when interpreting the results of this study. Based on the recommendation of Saraswat et al.[3], differences assessed in pathological conditions using the mSHCG model should not be considered clinically significant if less than 4°; this value was determined by the reproducibility of marker placement. By this definition, the current study demonstrated that the SMmodel produced ankle and midfoot angles that were clinically similar to the DFmodel, as SMmodel; notably, DFmodel mean angles were often within 4° and few angular differences exceeded 4°. Thus, our results provide confidence in the clinical utility of the mSHCG model for diagnosis and treatment planning.
As DFmodel midfoot ROM in the coronal plane increased, angular differences between the DFmodel and SMmodel also increased. This may indicate that skin-marker angular estimations are less accurate as coronal plane midfoot ROM increases. Increased error in the midfoot angles may have resulted from differences between skin-marker and DF forefoot segment positions. Skin-marker motion of the forefoot was determined by two markers on the first metatarsal and one marker on the fifth metatarsal, while DF motion was determined through direct model-based tracking of the first and fifth metatarsals. Skin-markers may have been subject to STA in relation to the DF marker locations, which may have been exacerbated by the rigid-body assumption applied to the forefoot segment for both models. In fact, significant differences between the DF-tracked virtual and skin-marker MT5H markers were noted in the X direction during foot loading and unloading. These significant differences between marker positions on the forefoot could be attributed to quicker foot speeds and/or higher forces during loading and unloading. For example, the markers on the forefoot travel a greater distance to reach the ground than those on the heel as foot-flat is achieved following heelstrike. This may result in increased forces on the forefoot and therefore more STA during loading.
A recent study evaluated the convergent validity of the Rizzoli model between skin-marker motion analysis and dual fluoroscopy, wherein ankle angles, medial longitudinal arch sagittal plane angles, and differences in marker positions were reported [20]. Our results are not directly comparable, as we evaluated the mSHCG model. However, Kessler et al. saw the greatest differences in marker positions for the tibia, which was supported by the significant differences we observed for tibia marker positions in the X direction.
Using bone pins, Nester et al. evaluated the rigid body assumption in the foot and reported maximum differences between segments encompassing multiple bones [27]. Based on the findings of Nester et al., as much as 7° of difference between the SMmodel and DFmodel could be attributed to the rigid body assumption. Another study by Nester et al. found differences between skin-marker and bone pin angles to be a maximum of 10° during walking [16], which is slightly higher than the maximum difference of 9.01° between the SMmodel and DFmodel found in the current study. However, Nester et al. did not acquire skin-marker and bone pin trajectory data simultaneously but were required to collect separate trials for each [16].
Our study had limitations. First, we evaluated only one multi-segment foot model, as it was impractical to place additional skin-markers on the foot to define additional models. Second, our sample size was low and only one trial of five participants was analyzed. Finally, due to the movement of the foot and the physical size of the combined field of view, only the foot-strike through push-off portion of stance could be analyzed.
In conclusion, our results demonstrate that skin-marker motion of the ankle and midfoot closely resembles that of DF when used in conjunction with the mSHCG model. However, larger ROM was significantly related to larger differences between skin-marker and DF angles for midfoot coronal plane angles. In addition, the positions of the fifth metatarsal skin marker was significantly different than the DF-tracked virtual markers during loading and unloading, all of which may indicate that skin-marker motion is less accurate during quicker motions. The is a purely technical marker in the mSHCG model, and therefore may suggest that an alternative placement may improve accuracy. Future research could apply the methodology described herein to analyze the accuracy of other marker models and inform the development of new models that minimize error.
Supplementary Material
Highlights.
Skin marker mean angles and range of motion were clinically similar to dual fluoroscopy.
Midfoot inversion/eversion differed most between skin markers and dual fluoroscopy.
Skin markers on the shank and foot were within 10 mm of dual fluoroscopy positions.
Acknowledgements
The authors gratefully acknowledge the contributions of Dylan Blair and statistical assistance from Greg Stoddard. This research was supported by the National Institutes of Health with grant numbers: R21AR069773, R21-AR063844, F32-AR067075 and the LS Peery Discovery Program in Musculoskeletal Restoration.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflict of Interest
The authors have no conflicts of interest to disclose.
References
- [1].Moseley L, Smith R, Hunt A, Gant R. Three-dimensional kinematics of the rearfoot during the stance phase of walking in normal young adult males. Clinical Biomechanics 1996;11:39–45. [DOI] [PubMed] [Google Scholar]
- [2].Siegel KL, Kepple TM, O’Connell PG, Gerber LH, Stanhope SJ. A technique to evaluate foot function during the stance phase of gait. Foot & ankle international 1995;16:764–70. [DOI] [PubMed] [Google Scholar]
- [3].Saraswat P, MacWilliams BA, Davis RB. A multi-segment foot model based on anatomically registered technical coordinate systems: method repeatability in pediatric feet. Gait & posture 2012;35:547–55. [DOI] [PubMed] [Google Scholar]
- [4].Leardini A, Benedetti MG, Catani F, Simoncini L, Giannini S. An anatomically based protocol for the description of foot segment kinematics during gait. Clin Biomech (Bristol, Avon) 1999;14:528–36. [DOI] [PubMed] [Google Scholar]
- [5].MacWilliams BA, Cowley M, Nicholson DE. Foot kinematics and kinetics during adolescent gait. Gait & posture 2003;17:214–24. [DOI] [PubMed] [Google Scholar]
- [6].Stebbins J, Harrington M, Thompson N, Zavatsky A, Theologis T. Repeatability of a model for measuring multi-segment foot kinematics in children. Gait & posture 2006;23:401–10. [DOI] [PubMed] [Google Scholar]
- [7].Simon J, Doederlein L, McIntosh AS, Metaxiotis D, Bock HG, Wolf SI. The Heidelberg foot measurement method: development, description and assessment. Gait & posture 2006;23:411–24. [DOI] [PubMed] [Google Scholar]
- [8].Caravaggi P, Benedetti MG, Berti L, Leardini A. Repeatability of a multi-segment foot protocol in adult subjects. Gait & posture 2011;33:133–5. [DOI] [PubMed] [Google Scholar]
- [9].Long JT, Eastwood DC, Graf AR, Smith PA, Harris GF. Repeatability and sources of variability in multi-center assessment of segmental foot kinematics in normal adults. Gait & posture 2010;31:32–6. [DOI] [PubMed] [Google Scholar]
- [10].Carson MC, Harrington ME, Thompson N, O’Connor JJ, Theologis TN. Kinematic analysis of a multi-segment foot model for research and clinical applications: a repeatability analysis. Journal of biomechanics 2001;34:1299–307. [DOI] [PubMed] [Google Scholar]
- [11].Seo SG, Lee DY, Moon HJ, Kim SJ, Kim J, Lee KM, et al. Repeatability of a multi-segment foot model with a 15-marker set in healthy adults. Journal of foot and ankle research 2014;7:24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [12].Wright CJ, Arnold BL, Coffey TG, Pidcoe PE. Repeatability of the modified Oxford foot model during gait in healthy adults. Gait & posture 2011;33:108–12. [DOI] [PubMed] [Google Scholar]
- [13].Saraswat P, MacWilliams BA, Davis RB, D’Astous JL. A multi-segment foot model based on anatomically registered technical coordinate systems: method repeatability and sensitivity in pediatric planovalgus feet. Gait & posture 2013;37:121–5. [DOI] [PubMed] [Google Scholar]
- [14].Nester CJ, Liu AM, Ward E, Howard D, Cocheba J, Derrick T. Error in the description of foot kinematics due to violation of rigid body assumptions. Journal of biomechanics 2010;43:666–72. [DOI] [PubMed] [Google Scholar]
- [15].Okita N, Meyers SA, Challis JH, Sharkey NA. An objective evaluation of a segmented foot model. Gait & posture 2009;30:27–34. [DOI] [PubMed] [Google Scholar]
- [16].Nester C, Jones RK, Liu A, Howard D, Lundberg A, Arndt A, et al. Foot kinematics during walking measured using bone and surface mounted markers. Journal of biomechanics 2007;40:3412–23. [DOI] [PubMed] [Google Scholar]
- [17].Tranberg R, Karlsson D. The relative skin movement of the foot: a 2-D roentgen photogrammetry study. Clin Biomech (Bristol, Avon) 1998;13:71–6. [DOI] [PubMed] [Google Scholar]
- [18].Fregly BJ, Rahman HA, Banks SA. Theoretical accuracy of model-based shape matching for measuring natural knee kinematics with single-plane fluoroscopy. J Biomech Eng 2005;127:692–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [19].Roach KE, Wang B, Kapron AL, Fiorentino NM, Saltzman CL, Bo Foreman K, et al. In Vivo Kinematics of the Tibiotalar and Subtalar Joints in Asymptomatic Subjects: A High-Speed Dual Fluoroscopy Study. J Biomech Eng 2016;138. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Kessler SE, Rainbow MJ, Lichtwark GA, Cresswell AG, D’Andrea SE, Konow N, et al. A Direct Comparison of Biplanar Videoradiography and Optical Motion Capture for Foot and Ankle Kinematics. Frontiers in Bioengineering and Biotechnology 2019;7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21].Saraswat P, Andersen MS, Macwilliams BA. A musculoskeletal foot model for clinical gait analysis. Journal of biomechanics 2010;43:1645–52. [DOI] [PubMed] [Google Scholar]
- [22].Holzer N, Salvo D, Marijnissen AC, Vincken KL, Ahmad AC, Serra E, et al. Radiographic evaluation of posttraumatic osteoarthritis of the ankle: the Kellgren-Lawrence scale is reliable and correlates with clinical symptoms. Osteoarthritis and cartilage 2015;23:363–9. [DOI] [PubMed] [Google Scholar]
- [23].Kellgren JH, Lawrence JS. Radiological assessment of osteo-arthrosis. Annals of the rheumatic diseases 1957;16:494–502. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [24].Fiorentino NM, Kutschke MJ, Atkins PR, Foreman KB, Kapron AL, Anderson AE. Accuracy of Functional and Predictive Methods to Calculate the Hip Joint Center in Young Non-pathologic Asymptomatic Adults with Dual Fluoroscopy as a Reference Standard. Annals of biomedical engineering 2016;44:2168–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [25].Wang B, Roach KE, Kapron AL, Fiorentino NM, Saltzman CL, Singer M, et al. Accuracy and feasibility of high-speed dual fluoroscopy and model-based tracking to measure in vivo ankle arthrokinematics. Gait & posture 2015;41:888–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].Bey MJ, Zauel R, Brock SK, Tashman S. Validation of a new model-based tracking technique for measuring three-dimensional, in vivo glenohumeral joint kinematics. J Biomech Eng 2006;128:604–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [27].Nester CJ, Liu AM, Ward E, Howard D, Cocheba J, Derrick T, et al. In vitro study of foot kinematics using a dynamic walking cadaver model. Journal of biomechanics 2007;40:1927–37. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.