Machine learning methods to support personalized neuromusculoskeletal modelling.

Many biomedical, orthopaedic, and industrial applications are emerging that will benefit from personalized neuromusculoskeletal models. Applications include refined diagnostics, prediction of treatment trajectories for neuromusculoskeletal diseases, in silico design, development, and testing of medical implants, and human-machine interfaces to support assistive technologies. This review proposes how physics-based simulation, combined with machine learning approaches from big data, can be used to develop high-fidelity personalized representations of the human neuromusculoskeletal system. The core neuromusculoskeletal model features requiring personalization are identified, and big data/machine learning approaches for implementation are presented together with recommendations for further research.

Validation of a model-based inverse kinematics approach based on wearable inertial sensors.

Wearable inertial measurement units (IMUs) are a promising solution to human motion estimation. Using IMUs 3D orientations, a model-driven inverse kinematics methodology to estimate joint angles is presented. Estimated joint angles were validated against encoder-measured kinematics (robot) and against marker-based kinematics (passive mechanism). Results are promising, with RMS angular errors respectively lower than 3 and 6 deg over a minimum range of motion of 50 deg (robot) and 160 deg (passive mechanism). Moreover, a noise robustness analysis revealed that the model-driven approach reduces the effects of experimental noises, making the proposed technique particularly suitable for application in human motion analysis.

Alterations of musculoskeletal models for a more accurate estimation of lower limb joint contact forces during normal gait: A systematic review.

Musculoskeletal modelling is a methodology used to investigate joint contact forces during a movement. High accuracy in the estimation of the hip or knee joint contact forces can be obtained with subject-specific models. However, construction of subject-specific models remains time consuming and expensive. The purpose of this systematic review of the literature was to identify what alterations can be made on generic (i.e. literature-based, without any subject-specific measurement other than body size and weight) musculoskeletal models to obtain a better estimation of the joint contact forces. The impact of these alterations on the accuracy of the estimated joint contact forces were appraised. The systematic search yielded to 141 articles and 24 papers were included in the review. Different strategies of alterations were found: skeletal and joint model (e.g. number of degrees of freedom, knee alignment), muscle model (e.g. Hill-type muscle parameters, level of muscular redundancy), and optimisation problem (e.g. objective function, design variables, constraints). All these alterations had an impact on joint contact force accuracy, so demonstrating the potential for improving the model predictions without necessarily involving costly and time consuming medical images. However, due to discrepancies in the reported evidence about this impact and despite a high quality of the reviewed studies, it was not possible to highlight any trend defining which alteration had the largest impact.

A Compton scattering setup for pulse shape discrimination studies in germanium detectors.

Pulse shape discrimination is an important handle to improve sensitivity in low background experiments. A dedicated setup was built to investigate the response of high-purity germanium detectors to single Compton scattered events. Using properly collimated γ-ray sources, it is possible to select events with known interaction location. The aim is to correlate the position dependent signal shape with geometrical and electrical properties of the detector. We report on design and performance of the setup with a first look on data.

Real-time inverse kinematics and inverse dynamics for lower limb applications using OpenSim.

Real-time estimation of joint angles and moments can be used for rapid evaluation in clinical, sport, and rehabilitation contexts. However, real-time calculation of kinematics and kinetics is currently based on approximate solutions or generic anatomical models. We present a real-time system based on OpenSim solving inverse kinematics and dynamics without simplifications at 2000 frame per seconds with less than 31.5 ms of delay. We describe the software architecture, sensitivity analyses to minimise delays and errors, and compare offline and real-time results. This system has the potential to strongly impact current rehabilitation practices enabling the use of personalised musculoskeletal models in real-time.

Toward modeling locomotion using electromyography-informed 3D models: application to cerebral palsy.

This position paper proposes a modeling pipeline to develop clinically relevant neuromusculoskeletal models to understand and treat complex neurological disorders. Although applicable to a variety of neurological conditions, we provide direct pipeline applicative examples in the context of cerebral palsy (CP). This paper highlights technologies in: (1) patient-specific segmental rigid body models developed from magnetic resonance imaging for use in inverse kinematics and inverse dynamics pipelines; (2) efficient population-based approaches to derive skeletal models and muscle origins/insertions that are useful for population statistics and consistent creation of continuum models; (3) continuum muscle descriptions to account for complex muscle architecture including spatially varying material properties with muscle wrapping; (4) muscle and tendon properties specific to CP; and (5) neural-based electromyography-informed methods for muscle force prediction. This represents a novel modeling pipeline that couples for the first time electromyography extracted features of disrupted neuromuscular behavior with advanced numerical methods for modeling CP-specific musculoskeletal morphology and function. The translation of such pipeline to the clinical level will provide a new class of biomarkers that objectively describe the neuromusculoskeletal determinants of pathological locomotion and complement current clinical assessment techniques, which often rely on subjective judgment. WIREs Syst Biol Med 2017, 9:e1368. doi: 10.1002/wsbm.1368 For further resources related to this article, please visit the WIREs website.

Joint kinematic calculation based on clinical direct kinematic versus inverse kinematic gait models.

Most clinical gait laboratories use the conventional gait analysis model. This model uses a computational method called Direct Kinematics (DK) to calculate joint kinematics. In contrast, musculoskeletal modelling approaches use Inverse Kinematics (IK) to obtain joint angles. IK allows additional analysis (e.g. muscle-tendon length estimates), which may provide valuable information for clinical decision-making in people with movement disorders. The twofold aims of the current study were: (1) to compare joint kinematics obtained by a clinical DK model (Vicon Plug-in-Gait) with those produced by a widely used IK model (available with the OpenSim distribution), and (2) to evaluate the difference in joint kinematics that can be solely attributed to the different computational methods (DK versus IK), anatomical models and marker sets by using MRI based models. Eight children with cerebral palsy were recruited and presented for gait and MRI data collection sessions. Differences in joint kinematics up to 13° were found between the Plug-in-Gait and the gait 2392 OpenSim model. The majority of these differences (94.4%) were attributed to differences in the anatomical models, which included different anatomical segment frames and joint constraints. Different computational methods (DK versus IK) were responsible for only 2.7% of the differences. We recommend using the same anatomical model for kinematic and musculoskeletal analysis to ensure consistency between the obtained joint angles and musculoskeletal estimates.

Application of a falsification strategy to a musculoskeletal model of the lower limb and accuracy of the predicted hip contact force vector.

In the literature, lower limb musculoskeletal models validated against in vivo measured hip contact forces (HCFs) exhibit a tendency to overestimate the HCFs magnitude and predict inaccurate components of the HCF vector in the transverse plane. In order to investigate this issue, a musculoskeletal model was forced to produce HCFs identical to those measured and the resulting joint equilibrium equations were studied through both a general approach and a static optimization framework. In the former case, the existence of solutions to the equilibrium equations was investigated and the effect of varying the intersegmental moments and the muscle tetanic stress assessed: for a value of 100 N/cm(2) and moments calculated from an inverse dynamics analysis on average only 62% of analyzed frames were solvable for level walking and 70% for stair climbing. In the static optimization study, the model could reproduce the experimental HCFs but the recruited muscles were unable to simultaneously equilibrate the hip intersegmental moments without the contribution of reserve moment actuators. Without constraints imposed on the HCFs, the predicted HCF vectors presented maximum angle deviations up to 22° for level walking and 33° for stair climbing during the gait stance phase. The influence of the medio-lateral HCF component on the solvability of the equilibrium equations and the muscle recruitment alteration when the model was forced to produce the experimental HCFs suggest that a more accurate geometrical representation of the gluteal muscles is mandatory to improve predictions of the HCF vector yielded by the static optimization technique.

An open source lower limb model: Hip joint validation.

Musculoskeletal lower limb models have been shown to be able to predict hip contact forces (HCFs) that are comparable to in vivo measurements obtained from instrumented prostheses. However, the muscle recruitment predicted by these models does not necessarily compare well to measured electromyographic (EMG) signals. In order to verify if it is possible to accurately estimate HCFs from muscle force patterns consistent with EMG measurements, a lower limb model based on a published anatomical dataset (Klein Horsman et al., 2007. Clinical Biomechanics. 22, 239-247) has been implemented in the open source software OpenSim. A cycle-to-cycle hip joint validation was conducted against HCFs recorded during gait and stair climbing trials of four arthroplasty patients (Bergmann et al., 2001. Journal of Biomechanics. 34, 859-871). Hip joint muscle tensions were estimated by minimizing a polynomial function of the muscle forces. The resulting muscle activation patterns obtained by assessing multiple powers of the objective function were compared against EMG profiles from the literature. Calculated HCFs denoted a tendency to monotonically increase their magnitude when raising the power of the objective function; the best estimation obtained from muscle forces consistent with experimental EMG profiles was found when a quadratic objective function was minimized (average overestimation at experimental peak frame: 10.1% for walking, 7.8% for stair climbing). The lower limb model can produce appropriate balanced sets of muscle forces and joint contact forces that can be used in a range of applications requiring accurate quantification of both. The developed model is available at the website https://simtk.org/home/low_limb_london.