Best methods and data to reconstruct paediatric lower limb bones for musculoskeletal modelling

Full-length radiograph based automatic musculoskeletal modeling using convolutional neural network

Full-length radiographs contain information from which many anatomical parameters of the pelvis, femur, and tibia may be derived, but only a few anatomical parameters are used for musculoskeletal modeling. This study aimed to develop a fully automatic algorithm to extract anatomical parameters from full-length radiograph to generate a musculoskeletal model that is more accurate than linear scaled one. A U-Net convolutional neural network was trained to segment the pelvis, femur, and tibia from the full-length radiograph. Eight anatomic parameters (six for length and width, two for angles) were automatically extracted from the bone segmentation masks and used to generate the musculoskeletal model. Sørensen-Dice coefficient was used to quantify the consistency of automatic bone segmentation masks with manually segmented labels. Maximum distance error, root mean square (RMS) distance error and Jaccard index (JI) were used to evaluate the geometric accuracy of the automatically generated pelvis, femur and tibia models versus CT bone models. Mean Sørensen-Dice coefficients for the pelvis, femur and tibia 2D segmentation masks were 0.9898, 0.9822 and 0.9786, respectively. The algorithm-driven bone models were closer to the 3D CT bone models than the scaled generic models in geometry, with significantly lower maximum distance error (28.3 % average decrease from 24.35 mm) and RMS distance error (28.9 % average decrease from 9.55 mm) and higher JI (17.2 % average increase from 0.46) (P < 0.001). The algorithm-driven musculoskeletal modeling (107.15 ± 10.24 s) was faster than the manual process (870.07 ± 44.79 s) for the same full-length radiograph. This algorithm provides a fully automatic way to generate a musculoskeletal model from full-length radiograph that achieves an approximately 30 % reduction in distance errors, which could enable personalized musculoskeletal simulation based on full-length radiograph for large scale OA populations.

Comparative validation of two patient-specific modelling pipelines for predicting knee joint forces during level walking

Over the past few years, the use of computer models and simulations tailored to the patient's physiology to assist clinical decision-making has increased enormously.While several pipelines to develop personalized models exist, their adoption on a large scale is still limited due to the required niche computational skillset and the lengthy operations required. Novel toolboxes, such as STAPLE, promise to streamline and expedite the development of image-based skeletal lower limb models. STAPLE-generated models can be rapidly generated, with minimal user input, and present similar joint kinematics and kinetics compared to models developed employing the established INSIGNEO pipeline. Yet, it is unclear how much the observed discrepancies scale up and affect joint contact force predictions. In this study, we compared image-based musculoskeletal models developed (i) with the INSIGNEO pipeline and (ii) with a semi-automated pipeline that combines STAPLE and nmsBuilder, and assessed their accuracy against experimental implant data.Our results showed that both pipelines predicted similar total knee joint contact forces between one another in terms of profiles and average values, characterized by a moderately high level of agreement with the experimental data. Nonetheless, the Student t-test revealed statistically significant differences between both pipelines. Of note, the STAPLE-based pipeline required considerably less time than the INSIGNEO pipeline to generate a musculoskeletal model (i.e., 60 vs 160 min). This is likely to open up opportunities for the use of personalized musculoskeletal models in clinical practice, where time is of the essence.

A Narrative Review of Personalized Musculoskeletal Modeling Using the Physiome and Musculoskeletal Atlas Projects

In this narrative review, we explore developments in the field of computational musculoskeletal model personalization using the Physiome and Musculoskeletal Atlas Projects. Model geometry personalization; statistical shape modeling; and its impact on segmentation, classification, and model creation are explored. Examples include the trapeziometacarpal and tibiofemoral joints, Achilles tendon, gastrocnemius muscle, and pediatric lower limb bones. Finally, a more general approach to model personalization is discussed based on the idea of multiscale personalization called scaffolds.

Intra-operator Repeatability of Manual Segmentations of the Hip Muscles on Clinical Magnetic Resonance Images

The manual segmentation of muscles on magnetic resonance images is the gold standard procedure to reconstruct muscle volumes from medical imaging data and extract critical information for clinical and research purposes. (Semi)automatic methods have been proposed to expedite the otherwise lengthy process. These, however, rely on manual segmentations. Nonetheless, the repeatability of manual muscle volume segmentations performed on clinical MRI data has not been thoroughly assessed. When conducted, volumetric assessments often disregard the hip muscles. Therefore, one trained operator performed repeated manual segmentations (n = 3) of the iliopsoas (n = 34) and gluteus medius (n = 40) muscles on coronal T1-weighted MRI scans, acquired on 1.5 T scanners on a clinical population of patients elected for hip replacement surgery. Reconstructed muscle volumes were divided in sub-volumes and compared in terms of volume variance (normalized variance of volumes - nVV), shape (Jaccard Index-JI) and surface similarity (maximal Hausdorff distance-HD), to quantify intra-operator repeatability. One-way repeated measures ANOVA (or equivalent) tests with Bonferroni corrections for multiple comparisons were conducted to assess statistical significance. For both muscles, repeated manual segmentations were highly similar to one another (nVV: 2-6%, JI > 0.78, HD < 15 mm). However, shape and surface similarity were significantly lower when muscle extremities were included in the segmentations (e.g., iliopsoas: HD -12.06 to 14.42 mm, P < 0.05). Our findings show that the manual segmentation of hip muscle volumes on clinical MRI scans provides repeatable results over time. Nonetheless, extreme care should be taken in the segmentation of muscle extremities.

Multi-level personalization of neuromusculoskeletal models to estimate physiologically plausible knee joint contact forces in children

Neuromusculoskeletal models are a powerful tool to investigate the internal biomechanics of an individual. However, commonly used neuromusculoskeletal models are generated via linear scaling of generic templates derived from elderly adult anatomies and poorly represent a child, let alone children with a neuromuscular disorder whose musculoskeletal structures and muscle activation patterns are profoundly altered. Model personalization can capture abnormalities and appropriately describe the underlying (altered) biomechanics of an individual. In this work, we explored the effect of six different levels of neuromusculoskeletal model personalization on estimates of muscle forces and knee joint contact forces to tease out the importance of model personalization for normal and abnormal musculoskeletal structures and muscle activation patterns. For six children, with and without cerebral palsy, generic scaled models were developed and progressively personalized by (1) tuning and calibrating musculotendon units' parameters, (2) implementing an electromyogram-assisted approach to synthesize muscle activations, and (3) replacing generic anatomies with image-based bony geometries, and physiologically and physically plausible muscle kinematics. Biomechanical simulations of gait were performed in the OpenSim and CEINMS software on ten overground walking trials per participant. A mixed-ANOVA test, with Bonferroni corrections, was conducted to compare all models' estimates. The model with the highest level of personalization produced the most physiologically plausible estimates. Model personalization is crucial to produce physiologically plausible estimates of internal biomechanical quantities. In particular, personalization of musculoskeletal anatomy and muscle activation patterns had the largest effect overall. Increased research efforts are needed to ease the creation of personalized neuromusculoskeletal models.

Development of predictive statistical shape models for paediatric lower limb bones

BACKGROUND AND OBJECTIVE

Accurate representation of bone shape is important for subject-specific musculoskeletal models as it may influence modelling of joint kinematics, kinetics, and muscle dynamics. Statistical shape modelling is a method to estimate bone shape from minimal information, such as anatomical landmarks, and to avoid the time and cost associated with reconstructing bone shapes from comprehensive medical imaging. Statistical shape models (SSM) of lower limb bones have been developed and validated for adult populations but are not applicable to paediatric populations. This study aimed to develop SSM for paediatric lower limb bones and evaluate their reconstruction accuracy using sparse anatomical landmarks.

METHODS

We created three-dimensional models of 56 femurs, 29 pelves, 56 tibias, 56 fibulas, and 56 patellae through segmentation of magnetic resonance images taken from 29 typically developing children (15 females; 13 ± 3.5 years). The SSM for femur, pelvis, tibia, fibula, patella, haunch (i.e., combined femur and pelvis), and shank (i.e., combined tibia and fibula) were generated from manual segmentation of comprehensive magnetic resonance images to describe the shape variance of the cohort. We implemented a leave-one-out cross-validation method wherein SSM were used to reconstruct novel bones (i.e., those not included in SSM generation) using full- (i.e., full segmentation) and sparse- (i.e., anatomical landmarks) input, and then compared these reconstructions against bones segmented from magnetic resonance imaging. Reconstruction performance was evaluated using root mean squared errors (RMSE, mm), Jaccard index (0-1), Dice similarity coefficient (DSC) (0-1), and Hausdorff distance (mm). All results reported in this abstract are mean ± standard deviation.

RESULTS

Femurs, pelves, tibias, fibulas, and patellae reconstructed via SSM using full-input had RMSE between 0.89 ± 0.10 mm (patella) and 1.98 ± 0.38 mm (pelvis), Jaccard indices between 0.77 ± 0.03 (pelvis) and 0.90 ± 0.02 (tibia), DSC between 0.87 ± 0.02 (pelvis) and 0.95 ± 0.01 (tibia), and Hausdorff distances between 2.45 ± 0.57 mm (patella) and 9.01 ± 2.36 mm (pelvis). Reconstruction using sparse-input had RMSE ranging from 1.33 ± 0.61 mm (patella) to 3.60 ± 1.05 mm (pelvis), Jaccard indices ranging from 0.59 ± 0.10 (pelvis) to 0.83 ± 0.03 (tibia), DSC ranging from 0.74 ± 0.08 (pelvis) to 0.90 ± 0.02 (tibia), and Hausdorff distances ranging from 3.21 ± 1.19 mm (patella) to 12.85 ± 3.24 mm (pelvis).

CONCLUSIONS

The SSM of paediatric lower limb bones showed reconstruction accuracy consistent with previously developed SSM and outperformed adult-based SSM when used to reconstruct paediatric bones.

Predicting the hip joint centre in children: New regression equations, linear scaling, and statistical shape modelling

Determination of the hip joint centre (HJC) is important to accurately estimate hip joint motion, moments and muscle forces. The most accurate method for HJC estimation without medical imaging is an area of interest in the biomechanics community, especially in a paediatric population, which has not been widely evaluated. HJC locations were calculated by sphere-fitting to the acetabulum of three-dimensional pelvises segmented from 333 CT scans of children aged 4 to 18 years old. Three methods for determining the HJC were compared: regression equations, linear scaling, and shape model prediction. The new regression equations developed in this study produced Euclidean distance errors of 6.23 mm ± 2.90 mm. Linear scaling of paediatric bone produced errors of 3.90 mm ± 2.52 mm and adult bone scaling of 5.45 mm ± 3.26 mm. Prediction of the HJC using a paediatric statistical shape model produced mean Euclidian distance errors of 2.95 mm ± 1.65 mm. Overall, shape model prediction of the HJC produced the lowest errors, with linear scaling of a mean paediatric pelvis providing better estimates than regression equations.

Morphological variation in paediatric lower limb bones

Available methods for generating paediatric musculoskeletal geometry are to scale generic adult geometry, which is widely accessible but can be inaccurate, or to obtain geometry from medical imaging, which is accurate but time-consuming and costly. A population-based shape model is required to generate accurate and accessible musculoskeletal geometry in a paediatric population. The pelvis, femur, and tibia/fibula were segmented from 333 CT scans of children aged 4–18 years. Bone morphology variation was captured using principal component analysis (PCA). Subsequently, a shape model was developed to predict bone geometry from demographic and linear bone measurements and validated using a leave one out analysis. The shape model was compared to linear scaling of adult and paediatric bone geometry. The PCA captured growth-related changes in bone geometry. The shape model predicted bone geometry with root mean squared error (RMSE) of 2.91 ± 0.99 mm in the pelvis, 2.01 ± 0.62 mm in the femur, and 1.85 ± 0.54 mm in the tibia/fibula. Linear scaling of an adult mesh produced RMSE of 4.79 ± 1.39 mm in the pelvis, 4.38 ± 0.72 mm in the femur, and 4.39 ± 0.86 mm in the tibia/fibula. We have developed a method for capturing and predicting lower limb bone shape variation in a paediatric population more accurately than linear scaling without using medical imaging.

Free Achilles tendon strain during selected rehabilitation, locomotor, jumping, and landing tasks

A better understanding of the strains experienced by the Achilles tendon during commonly prescribed exercises and locomotor tasks is needed to improve efficacy of Achilles tendon training and rehabilitation programs. The aim of this study was to estimate in vivo free Achilles tendon strain during selected rehabilitation, locomotor, jumping, and landing tasks. Sixteen trained runners with no symptoms of Achilles tendinopathy participated in this study. Personalised free Achilles tendon moment arm and force-strain curve were obtained from imaging data and used in conjunction with motion capture and surface electromyography to estimate free Achilles tendon strain using electromyogram-informed neuromusculoskeletal modelling. There was a strong correspondence between Achilles tendon force estimates from the present study and experimental data reported in the literature (R² > 0.85). The average tendon strain was highest for maximal hop landing (8.8±1.6%), lowest for walking at 1.4 m/s (3.1±0.8%) and increased with locomotor speed during running (run 3.0 m/s: 6.5±1.6%; run 5.0 m/s: 7.9±1.7%) and during heel rise exercise with added mass (BW: 5.8±1.3%; 1.2 BW: 6.9±1.7%). The peak tendon strain was highest during running (5 m/s: 13.7±2.5%) and lowest during walking (1.4 m/s: 7±1.8%). Overall findings provide a preliminary evidence base for exercise selection to maximise anabolic tendon remodelling during training and rehabilitation of the Achilles tendon.

The future of in-field sports biomechanics: wearables plus modelling compute real-time in vivo tissue loading to prevent and repair musculoskeletal injuries

This paper explores the use of biomechanics in identifying the mechanistic causes of musculoskeletal tissue injury and degeneration. It appraises how biomechanics has been used to develop training programmes aiming to maintain or recover tissue health. Tissue health depends on the functional mechanical environment experienced by tissues during daily and rehabilitation activities. These environments are the result of the interactions between tissue motion, loading, biology, and morphology. Maintaining health of and/or repairing musculoskeletal tissues requires targeting the "ideal" in vivo tissue mechanics (i.e., loading and deformation), which may be enabled by appropriate real-time biofeedback. Recent research shows that biofeedback technologies may increase their quality and effectiveness by integrating a personalised neuromusculoskeletal modelling driven by real-time motion capture and medical imaging. Model personalisation is crucial in obtaining physically and physiologically valid predictions of tissue biomechanics. Model real-time execution is crucial and achieved by code optimisation and artificial intelligence methods. Furthermore, recent work has also shown that laboratory-based motion capture biomechanical measurements and modelling can be performed outside the laboratory with wearable sensors and artificial intelligence. The next stage is to combine these technologies into well-designed easy to use products to guide training to maintain or recover tissue health in the real-world.

Torsion Tool: An automated tool for personalising femoral and tibial geometries in OpenSim musculoskeletal models

Common practice in musculoskeletal modelling is to use scaled musculoskeletal models based on a healthy adult, but this does not consider subject-specific geometry, such as tibial torsion and femoral neck-shaft and anteversion angles (NSA and AVA). The aims of this study were to (1) develop an automated tool for creating OpenSim models with subject-specific tibial torsion and femoral NSA and AVA, (2) evaluate the femoral component, and (3) release the tool open-source. The Torsion Tool (https://simtk.org/projects/torsiontool) is a MATLAB-based tool that requires an individual's tibial torsion, NSA and AVA estimates as input and rotates corresponding bones and associated muscle points of a generic musculoskeletal model. Performance of the Torsion Tool was evaluated comparing femur bones as personalised with the Torsion Tool and scaled generic femurs with manually segmented bones as golden standard for six typically developing children and thirteen children with cerebral palsy. The tool generated femur geometries closer to the segmentations, with lower maximum (-19%) and root mean square (-18%) errors and higher Jaccard indices (+9%) compared to generic femurs. Furthermore, the tool resulted in larger improvements for participants with higher NSA and AVA deviations. The Torsion Tool allows an automatic, fast, and user-friendly way of personalising femoral and tibial geometry in an OpenSim musculoskeletal model. Personalisation is expected to be particularly relevant in pathological populations, as will be further investigated by evaluating the effects on simulation outcomes.

Impact of scaling errors of the thigh and shank segments on musculoskeletal simulation results

BACKGROUND

Musculoskeletal simulations are widely used in the research community. The locations of surface markers are mostly used to scale a generic model to the participant's anthropometry. Marker-based scaling approaches include errors due to inaccuracies in marker placements.

RESEARCH QUESTION

How do scaling errors of the thigh and shank segments influence simulation results?

METHODS

Motion capture data and magnetic resonance images from a child with cerebral palsy and a typically developing child were used to create a subject-specific reference model for each child. These reference models were modified to mimic scaling errors due to inaccurately placed lateral epicondyle markers, which are frequently used to scale the thigh and shank segments. The thigh length was altered in 1 % steps from the original length and the shank length was accordingly adjusted to keep the total leg length constant. Thirty additional models were created, which included models with an altered thigh length of ±15 %. Subsequently, musculoskeletal simulations with OpenSim were performed with all models. Joint kinematics, joint kinetics, muscle forces and joint contact forces (JCF) were compared between the reference and altered models.

RESULTS

The investigated scaling error influenced joint kinematics and joint kinetics by up to 9.4° (hip flexion angle) and 0.15 Nm/kg (knee flexion moment), respectively. Maximum muscle and JCF differences of 46 % (medial gastrocnemius) and 72 % (hip JCF) bodyweight, respectively, were observed between the reference and altered models. Scaling errors mainly changed the magnitude but not the shape of most analyzed parameters. The influence of scaling errors on simulation results were similar in both participants.

SIGNIFICANCE

Scaling errors of the thigh segment influence simulation results at all joints due to the global optimization approach used in musculoskeletal simulations. Our findings can be used to estimate potential errors due to marker-based scaling approaches in previous and future studies.

A Systematic Review of the Associations Between Inverse Dynamics and Musculoskeletal Modeling to Investigate Joint Loading in a Clinical Environment

The assessment of knee or hip joint loading by external joint moments is mainly used to draw conclusions on clinical decision making. However, the correlation between internal and external loads has not been systematically analyzed. This systematic review aims, therefore, to clarify the relationship between external and internal joint loading measures during gait. A systematic database search was performed to identify appropriate studies for inclusion. In total, 4,554 articles were identified, while 17 articles were finally included in data extraction. External joint loading parameters were calculated using the inverse dynamics approach and internal joint loading parameters by musculoskeletal modeling or instrumented prosthesis. It was found that the medial and total knee joint contact forces as well as hip joint contact forces in the first half of stance can be well predicted using external joint moments in the frontal plane, which is further improved by including the sagittal joint moment. Worse correlations were found for the peak in the second half of stance as well as for internal lateral knee joint contact forces. The estimation of external joint moments is useful for a general statement about the peak in the first half of stance or for the maximal loading. Nevertheless, when investigating diseases as valgus malalignment, the estimation of lateral knee joint contact forces is necessary for clinical decision making because external joint moments could not predict the lateral knee joint loading sufficient enough. Dependent on the clinical question, either estimating the external joint moments by inverse dynamics or internal joint contact forces by musculoskeletal modeling should be used.

Automated creation and tuning of personalised muscle paths for OpenSim musculoskeletal models of the knee joint

Computational modelling is an invaluable tool for investigating features of human locomotion and motor control which cannot be measured except through invasive techniques. Recent research has focussed on creating personalised musculoskeletal models using population-based morphing or directly from medical imaging. Although progress has been made, robust definition of two critical model parameters remains challenging: (1) complete tibiofemoral (TF) and patellofemoral (PF) joint motions, and (2) muscle tendon unit (MTU) pathways and kinematics (i.e. lengths and moment arms). The aim of this study was to develop an automated framework, using population-based morphing approaches to create personalised musculoskeletal models, consisting of personalised bone geometries, TF and PF joint mechanisms, and MTU pathways and kinematics. Informed from medical imaging, personalised rigid body TF and PF joint mechanisms were created. Using atlas- and optimisation-based methods, personalised MTU pathways and kinematics were created with the aim of preventing MTU penetration into bones and achieving smooth MTU kinematics that follow patterns from existing literature. This framework was integrated into the Musculoskeletal Atlas Project Client software package to create and optimise models for 6 participants with incrementally increasing levels of personalisation with the aim of improving MTU kinematics and pathways. Three comparisons were made: (1) non-optimised (Model 1) and optimised models (Model 3) with generic joint mechanisms; (2) non-optimised (Model 2) and optimised models (Model 4) with personalised joint mechanisms; and (3) both optimised models (Model 3 and 4). Following optimisation, improvements were consistently shown in pattern similarity to cadaveric data in comparison (1) and (2). For comparison (3), a number of comparisons showed no significant difference between the two compared models. Importantly, optimisation did not produce statistically significantly worse results in any case.

In Silico-Enhanced Treatment and Rehabilitation Planning for Patients with Musculoskeletal Disorders: Can Musculoskeletal Modelling and Dynamic Simulations Really Impact Current Clinical Practice?

Over the past decades, the use of computational physics-based models representative of the musculoskeletal (MSK) system has become increasingly popular in many fields of clinically driven research, locomotor rehabilitation in particular. These models have been applied to various functional impairments given their ability to estimate parameters which cannot be readily measured in vivo but are of interest to clinicians. The use of MSK modelling and simulations allows analysis of relevant MSK biomarkers such as muscle and joint contact loading at a number of different stages in the clinical treatment pathway in order to benefit patient functional outcome. Applications of these methods include optimisation of rehabilitation programs, patient stratification, disease characterisation, surgical pre-planning, and assistive device and exoskeleton design and optimisation. This review provides an overview of current approaches, the components of standard MSK models, applications, limitations, and assumptions of these modelling and simulation methods, and finally proposes a future direction.

Stimulus Parameters Underlying Sound-Symbolic Mapping of Auditory Pseudowords to Visual Shapes

Sound symbolism refers to non-arbitrary mappings between the sounds of words and their meanings and is often studied by pairing auditory pseudowords such as "maluma" and "takete" with rounded and pointed visual shapes, respectively. However, it is unclear what auditory properties of pseudowords contribute to their perception as rounded or pointed. Here, we compared perceptual ratings of the roundedness/pointedness of large sets of pseudowords and shapes to their acoustic and visual properties using a novel application of representational similarity analysis (RSA). Representational dissimilarity matrices (RDMs) of the auditory and visual ratings of roundedness/pointedness were significantly correlated crossmodally. The auditory perceptual RDM correlated significantly with RDMs of spectral tilt, the temporal fast Fourier transform (FFT), and the speech envelope. Conventional correlational analyses showed that ratings of pseudowords transitioned from rounded to pointed as vocal roughness (as measured by the harmonics-to-noise ratio, pulse number, fraction of unvoiced frames, mean autocorrelation, shimmer, and jitter) increased. The visual perceptual RDM correlated significantly with RDMs of global indices of visual shape (the simple matching coefficient, image silhouette, image outlines, and Jaccard distance). Crossmodally, the RDMs of the auditory spectral parameters correlated weakly but significantly with those of the global indices of visual shape. Our work establishes the utility of RSA for analysis of large stimulus sets and offers novel insights into the stimulus parameters underlying sound symbolism, showing that sound-to-shape mapping is driven by acoustic properties of pseudowords and suggesting audiovisual cross-modal correspondence as a basis for language users' sensitivity to this type of sound symbolism.

Targeted Achilles Tendon Training and Rehabilitation Using Personalized and Real-Time Multiscale Models of the Neuromusculoskeletal System

Musculoskeletal tissues, including tendons, are sensitive to their mechanical environment, with both excessive and insufficient loading resulting in reduced tissue strength. Tendons appear to be particularly sensitive to mechanical strain magnitude, and there appears to be an optimal range of tendon strain that results in the greatest positive tendon adaptation. At present, there are no tools that allow localized tendon strain to be measured or estimated in training or a clinical environment. In this paper, we first review the current literature regarding Achilles tendon adaptation, providing an overview of the individual technologies that so far have been used in isolation to understand in vivo Achilles tendon mechanics, including 3D tendon imaging, motion capture, personalized neuromusculoskeletal rigid body models, and finite element models. We then describe how these technologies can be integrated in a novel framework to provide real-time feedback of localized Achilles tendon strain during dynamic motor tasks. In a proof of concept application, Achilles tendon localized strains were calculated in real-time for a single subject during walking, single leg hopping, and eccentric heel drop. Data was processed at 250 Hz and streamed on a smartphone for visualization. Achilles tendon peak localized strains ranged from ∼3 to ∼11% for walking, ∼5 to ∼15% during single leg hop, and ∼2 to ∼9% during single eccentric leg heel drop, overall showing large strain variation within the tendon. Our integrated framework connects, across size scales, knowledge from isolated tendons and whole-body biomechanics, and offers a new approach to Achilles tendon rehabilitation and training. A key feature is personalization of model components, such as tendon geometry, material properties, muscle geometry, muscle-tendon paths, moment arms, muscle activation, and movement patterns, all of which have the potential to affect tendon strain estimates. Model personalization is important because tendon strain can differ substantially between individuals performing the same exercise due to inter-individual differences in these model components.