Browsing by Author "Ao, Di"

Now showing 1 - 6 of 6
  • Thumbnail Image
    Item
    A computational method for estimating trunk muscle activations during gait using lower extremity muscle synergies
    (Frontiers Media S.A., 2022) Li, Geng; Ao, Di; Vega, Marleny M.; Shourijeh, Mohammad S.; Zandiyeh, Payam; Chang, Shuo-Hsiu; Lewis, Valerae O.; Dunbar, Nicholas J.; Babazadeh-Naseri, Ata; Baines, Andrew J.; Fregly, Benjamin J.; Rice Computational Neuromechanics Laboratory
    One of the surgical treatments for pelvic sarcoma is the restoration of hip function with a custom pelvic prosthesis after cancerous tumor removal. The orthopedic oncologist and orthopedic implant company must make numerous often subjective decisions regarding the design of the pelvic surgery and custom pelvic prosthesis. Using personalized musculoskeletal computer models to predict post-surgery walking function and custom pelvic prosthesis loading is an emerging method for making surgical and custom prosthesis design decisions in a more objective manner. Such predictions would necessitate the estimation of forces generated by muscles spanning the lower trunk and all joints of the lower extremities. However, estimating trunk and leg muscle forces simultaneously during walking based on electromyography (EMG) data remains challenging due to the limited number of EMG channels typically used for measurement of leg muscle activity. This study developed a computational method for estimating unmeasured trunk muscle activations during walking using lower extremity muscle synergies. To facilitate the calibration of an EMG-driven model and the estimation of leg muscle activations, EMG data were collected from each leg. Using non-negative matrix factorization, muscle synergies were extracted from activations of leg muscles. On the basis of previous studies, it was hypothesized that the time-varying synergy activations were shared between the trunk and leg muscles. The synergy weights required to reconstruct the trunk muscle activations were determined through optimization. The accuracy of the synergy-based method was dependent on the number of synergies and optimization formulation. With seven synergies and an increased level of activation minimization, the estimated activations of the erector spinae were strongly correlated with their measured activity. This study created a custom full-body model by combining two existing musculoskeletal models. The model was further modified and heavily personalized to represent various aspects of the pelvic sarcoma patient, all of which contributed to the estimation of trunk muscle activations. This proposed method can facilitate the prediction of post-surgery walking function and pelvic prosthesis loading, as well as provide objective evaluations for surgical and prosthesis design decisions.
  • Thumbnail Image
    Item
    Changes in walking function and neural control following pelvic cancer surgery with reconstruction
    (Frontiers, 2024) Li, Geng; Ao, Di; Vega, Marleny M.; Zandiyeh, Payam; Chang, Shuo-Hsiu; Penny, Alexander N.; Lewis, Valerae O.; Fregly, Benjamin J.; Rice Computational Neuromechanics Laboratory
    Introduction: Surgical planning and custom prosthesis design for pelvic cancer patients are challenging due to the unique clinical characteristics of each patient and the significant amount of pelvic bone and hip musculature often removed. Limb-sparing internal hemipelvectomy surgery with custom prosthesis reconstruction has become a viable option for this patient population. However, little is known about how post-surgery walking function and neural control change from pre-surgery conditions. Methods: This case study combined comprehensive walking data (video motion capture, ground reaction, and electromyography) with personalized neuromusculoskeletal computer models to provide a thorough assessment of pre- to post-surgery changes in walking function (ground reactions, joint motions, and joint moments) and neural control (muscle synergies) for a single pelvic sarcoma patient who received internal hemipelvectomy surgery with custom prosthesis reconstruction. Pre- and post-surgery walking function and neural control were quantified using pre- and post-surgery neuromusculoskeletal models, respectively, whose pelvic anatomy, joint functional axes, muscle-tendon properties, and muscle synergy controls were personalized using the participant’s pre-and post-surgery walking and imaging data. For the post-surgery model, virtual surgery was performed to emulate the implemented surgical decisions, including removal of hip muscles and implantation of a custom prosthesis with total hip replacement. Results: The participant’s post-surgery walking function was marked by a slower self-selected walking speed coupled with several compensatory mechanisms necessitated by lost or impaired hip muscle function, while the participant’s post-surgery neural control demonstrated a dramatic change in coordination strategy (as evidenced by modified time-invariant synergy vectors) with little change in recruitment timing (as evidenced by conserved time-varying synergy activations). Furthermore, the participant’s post-surgery muscle activations were fitted accurately using his pre-surgery synergy activations but fitted poorly using his pre-surgery synergy vectors. Discussion: These results provide valuable information about which aspects of post-surgery walking function could potentially be improved through modifications to surgical decisions, custom prosthesis design, or rehabilitation protocol, as well as how computational simulations could be formulated to predict post-surgery walking function reliably given a patient’s pre-surgery walking data and the planned surgical decisions and custom prosthesis design.
  • Thumbnail Image
    Item
    EMG-driven musculoskeletal model calibration with estimation of unmeasured muscle excitations via synergy extrapolation
    (Frontiers Media S.A., 2022) Ao, Di; Vega, Marleny M.; Shourijeh, Mohammad S.; Patten, Carolynn; Fregly, Benjamin J.; Rice Computational Neuromechanics Lab
    Subject-specific electromyography (EMG)-driven musculoskeletal models that predict muscle forces have the potential to enhance our knowledge of internal biomechanics and neural control of normal and pathological movements. However, technical gaps in experimental EMG measurement, such as inaccessibility of deep muscles using surface electrodes or an insufficient number of EMG channels, can cause difficulties in collecting EMG data from muscles that contribute substantially to joint moments, thereby hindering the ability of EMG-driven models to predict muscle forces and joint moments reliably. This study presents a novel computational approach to address the problem of a small number of missing EMG signals during EMG-driven model calibration. The approach (henceforth called “synergy extrapolation” or SynX) linearly combines time-varying synergy excitations extracted from measured muscle excitations to estimate 1) unmeasured muscle excitations and 2) residual muscle excitations added to measured muscle excitations. Time-invariant synergy vector weights defining the contribution of each measured synergy excitation to all unmeasured and residual muscle excitations were calibrated simultaneously with EMG-driven model parameters through a multi-objective optimization. The cost function was formulated as a trade-off between minimizing joint moment tracking errors and minimizing unmeasured and residual muscle activation magnitudes. We developed and evaluated the approach by treating a measured fine wire EMG signal (iliopsoas) as though it were “unmeasured” for walking datasets collected from two individuals post-stroke–one high functioning and one low functioning. How well unmeasured muscle excitations and activations could be predicted with SynX was assessed quantitatively for different combinations of SynX methodological choices, including the number of synergies and categories of variability in unmeasured and residual synergy vector weights across trials. The two best methodological combinations were identified, one for analyzing experimental walking trials used for calibration and another for analyzing experimental walking trials not used for calibration or for predicting new walking motions computationally. Both methodological combinations consistently provided reliable and efficient estimates of unmeasured muscle excitations and activations, muscle forces, and joint moments across both subjects. This approach broadens the possibilities for EMG-driven calibration of muscle-tendon properties in personalized neuromusculoskeletal models and may eventually contribute to the design of personalized treatments for mobility impairments.
  • Thumbnail Image
    Item
    Evaluation of Synergy Extrapolation for Predicting Unmeasured Muscle Excitations from Measured Muscle Synergies
    (Frontiers, 2020) Ao, Di; Shourijeh, Mohammad S.; Patten, Carolynn; Fregly, Benjamin J.; Rice Computational Neuromechanics Lab
    Electromyography (EMG)-driven musculoskeletal modeling relies on high-quality measurements of muscle electrical activity to estimate muscle forces. However, a critical challenge for practical deployment of this approach is missing EMG data from muscles that contribute substantially to joint moments. This situation may arise due to either the inability to measure deep muscles with surface electrodes or the lack of a sufficient number of EMG channels. Muscle synergy analysis (MSA) is a dimensionality reduction approach that decomposes a large number of muscle excitations into a small number of time-varying synergy excitations along with time-invariant synergy weights that define the contribution of each synergy excitation to all muscle excitations. This study evaluates how well missing muscle excitations can be predicted using synergy excitations extracted from muscles with available EMG data (henceforth called “synergy extrapolation” or SynX). The method was evaluated using a gait data set collected from a stroke survivor walking on an instrumented treadmill at self-selected and fastest-comfortable speeds. The evaluation process started with full calibration of a lower-body EMG-driven model using 16 measured EMG channels (collected using surface and fine wire electrodes) per leg. One fine wire EMG channel (either iliopsoas or adductor longus) was then treated as unmeasured. The synergy weights associated with the unmeasured muscle excitation were predicted by solving a nonlinear optimization problem where the errors between inverse dynamics and EMG-driven joint moments were minimized. The prediction process was performed for different synergy analysis algorithms (principal component analysis and non-negative matrix factorization), EMG normalization methods, and numbers of synergies. SynX performance was most influenced by the choice of synergy analysis algorithm and number of synergies. Principal component analysis with five or six synergies consistently predicted unmeasured muscle excitations the most accurately and with the greatest robustness to EMG normalization method. Furthermore, the associated joint moment matching accuracy was comparable to that produced by initial EMG-driven model calibration using all 16 EMG channels per leg. SynX may facilitate the assessment of human neuromuscular control and biomechanics when important EMG signals are missing.
  • Thumbnail Image
    Item
    How Well Do Commonly Used Co-contraction Indices Approximate Lower Limb Joint Stiffness Trends During Gait for Individuals Post-stroke?
    (Frontiers, 2021) Li, Geng; Shourijeh, Mohammad S.; Ao, Di; Patten, Carolynn; Fregly, Benjamin J.; Rice Computational Neuromechanics Laboratory
    Muscle co-contraction generates joint stiffness to improve stability and accuracy during limb movement but at the expense of higher energetic cost. However, quantification of joint stiffness is difficult using either experimental or computational means. In contrast, quantification of muscle co-contraction using an EMG-based Co-Contraction Index (CCI) is easier and may offer an alternative for estimating joint stiffness. This study investigated the feasibility of using two common CCI’s to approximate lower limb joint stiffness trends during gait. Calibrated EMG-driven lower extremity musculoskeletal models constructed for two individuals post-stroke were used to generate the quantities required for CCI calculations and model-based estimation of joint stiffness. CCIs were calculated for various combinations of antagonist muscle pairs based on two common CCI formulations: Rudolph et al. (2000) (CCI1) and Falconer and Winter (1985) (CCI2). CCI1 measures antagonist muscle activation relative to not only total activation of agonist plus antagonist muscles but also agonist muscle activation, while CCI2 measures antagonist muscle activation relative to only total muscle activation. We computed the correlation between these two CCIs and model-based estimates of sagittal plane joint stiffness for the hip, knee, and ankle of both legs. Although we observed moderate to strong correlations between some CCI formulations and corresponding joint stiffness, these associations were highly dependent on the methodological choices made for CCI computation. Specifically, we found that: (1) CCI1 was generally more correlated with joint stiffness than was CCI2, (2) CCI calculation using EMG signals with calibrated electromechanical delay generally yielded the best correlations with joint stiffness, and (3) choice of antagonist muscle pairs significantly influenced CCI correlation with joint stiffness. By providing guidance on how methodological choices influence CCI correlation with joint stiffness trends, this study may facilitate a simpler alternate approach for studying joint stiffness during human movement.
  • Thumbnail Image
    Item
    Improving Personalization of Neuromusculoskeletal Models via EMG-driven Model Calibration
    (2022-04-20) Ao, Di; Fregly, Benjamin
    Neuromusculoskeletal disorders, which have high rates of prevalence and incidence, cause long-term disability. These individuals require substantial assistance in self-care, mobility, and house-hold activities over a long period of time, which places heavy healthcare and economic burdens on individuals, families and society. Surgical and rehabilitation treatments for neuromusculoskeletal disorders are typically subjective and consist of a standard menu of treatment options established using clinicians’ collective experience and judgment. Personalized treatment planning of neuromusculoskeletal fullest level of personalization, such models could facilitate objectivity and agreement between clinicians on the most effective treatment plan for a given individual. Electromyography (EMG)-driven musculoskeletal modeling is an emerging approach to identify subject-specific muscle force generating properties. Once identified, this modeling approach uses these properties to predict EMG-consistent muscle forces despite the muscle redundancy problem underlying musculoskeletal systems. Two critical issues with the practical deployment of EMG-driven modeling approach have received limited attention, both of which generally arise from low-quality model input variables. The first issue is missing EMG data from muscles that contribute substantially to joint moments, primarily due to inherent limitations of surface EMG recording and constraints arising from experimental conditions. The second issue is lack of subject-specific anatomical details in muscle-tendon geometrics, especially in the sizing and placement of simplified surface geometries that are employed to estimate muscle wrapping behavior over joint capsules, bony prominences, muscles of deeper layers, and neighboring soft tissues. This dissertation aims to improve EMG-driven musculoskeletal modeling methods by addressing two important challenges. First, a novel framework is presented to address the problem with EMG-driven model calibration that arises from missing EMG data of important muscles. An approach to estimate unmeasured muscle excitations using information extracted from measured muscle excitations (termed “synergy extrapolation” or “SynX”) was developed within the context of EMG-driven modeling. The development process started with the evaluation of SynX performance for different methodological combinations when EMG-driven model parameters were well-calibrated. Then, a multi-objective optimization problem was formulated that allowed SynX- predicted missing muscle excitations and EMG-driven model parameter values to be identified simultaneously. Two SynX methodological combinations, one targeted at analyzing experimentally measured motions and the other at generating computationally predicted motions, were identified. Both combinations were able to consistently provide accurate unmeasured muscle excitations and reliable muscle force estimates. Second, a novel workflow is presented to increase subject-specificity of muscle wrapping surface geometries via EMG-driven model calibration. The workflow started with the development of a two-level surrogate model of musculoskeletal geometry that could accurately approximate muscle-tendon geometries as functions of joint kinematics and muscle wrapping surface parameters for each muscle in the model. Then, these surrogate musculoskeletal geometric models were incorporated into the EMG-driven modeling process, which allowed muscle wrapping surface parameter to be adjusted through non-linear optimization. The capacity of the EMG-driven model to predict joint moments was significantly enhanced by the personalization of muscle wrapping surfaces, most notably by lowering the magnitude of joint contact force estimates. This dissertation presents two augmented EMG-driven modeling methods that show extensive potential in assessment of human neuromuscular control and biomechanics for optimal treatment design. Disorders is a new approach where an individuals’ specific characteristics are taken into account to restore lost function. Computational neuromusculoskeletal modeling is a powerful tool for providing personalized treatment planning for neuromusculoskeletal disorders. They can not only generate accurate and reliable estimates of important internal biomechanical variables during experimentally measured motions, but also predict objective functional outcomes for a variety of possible interventions. Every individual has unique anatomical and physiological characteristics of their neuromusculoskeletal system which can affect the accuracy of computational predictions. Therefore, personalized neuromusculoskeletal models that take into account these differences would provide more accurate outcome predictions. When employed to their