A Biomechanist’s Guide and Toolbox for Statistical Shape Modeling of Bones and Joints
Allison Clouthier, University of Ottawa; Anthony Gatti, Stanford University; Erin CS Lee, University of Waterloo; Michael J Rainbow, Queen’s University; Lauren Welte, University of Alberta.
Bone and joint geometry play a critical role in determining joint movement patterns and loading. Statistical shape modelling (SSM) serves as an umbrella term encompassing diverse approaches to quantifying shape and its variations. This workshop will introduce SSM, highlighting its many applications in biomechanics and adjacent fields. Attendees will gain insights through an overview of SSM methodologies and practical, hands-on demonstrations. Implemented via Google Colab, these demonstrations will allow participants to build their own models using supplied datasets or their own data. The workshop is designed to accommodate beginners and those with more experience, ensuring a valuable experience for all attendees.
Biophysical muscle models for musculoskeletal simulation
Lena Ting, Emory University and Georgia Tech; Surabhi Simha, Emory University and Georgia Tech; Hansol Ryu, Georgia Tech and Emory University; Tim van der Zee, KU Leuven; Friedl De Groote, KU Leuven.
To date, there are extremely few musculoskeletal movement simulations that use biophysics-based muscle models. Most simulations use phenomenological models of muscle force generation, i.e. “Hill-type models” that provide black-box input-output relationships between muscle activation, length, and force, offering no insight into biological processes of muscle. Because data underlying Hill-type models is collected in controlled isometric or isotonic experiments, Hill-type models have poor fidelity in behaviourally-relevant movements in which muscles’ operating conditions are very different from those in the experiments that underlie the model. For example, we previously needed to extend Hill-type models with a phenomenological description of short-range stiffness to capture the response of the muscle to stretch in clinical tests and during standing balance. Recently, we have developed biophysical muscle models based on muscle crossbridge dynamics from which both force-velocity and short-range stiffness properties emerge. In this workshop, we will present (1) these biophysical muscle models as well as (2) musculoskeletal simulations using these biophysical models using open-source software based on OpenSim. Hands-on tutorials will focus on simulating a clinical test of joint hyper-resistance and perturbed standing balance.
The workshop will consist of a seminar on biophysical muscle modeling followed by a hands-on session during which participants will perform musculoskeletal simulations using the biophysical muscle model.
“In the wild” movement analysis using dynamic simulations
Anne Koelewijn, Friedrich-Alexander-Universität Erlangen-Nürnberg; Ton van den Bogert, Cleveland State University.
Dynamic movement simulations have gained prominence in biomechanics in the last ten years, because improved computational techniques allow for them to be created within 10 minutes in two dimensions and within 45 minutes in three dimensions. These simulations describe a dynamically consistent movement of a musculoskeletal dynamics model entirely from its initial state and muscle stimulation inputs. They are created by solving an optimization problem that finds a simulation that minimizes an objective related to effort or energy expenditure minimization. This approach replicates the optimization used in the central nervous system to plan and execute movements, where energy/effort minimization is also the main goal during gait. Therefore, these simulations can be used to predict new movements without using any measurements or based on reference measurements, but they can also be used to reconstruct movements from incomplete or inaccurate gait measurements, which allows for movements to be measured accurately “in the wild”.
The goal of the tutorial is to teach participants how to use dynamics simulations to reconstruct movements from measured data and use this approach for “in the wild” movements. The tutorial’s focus is on inertial measurement units (IMUs), but the concept can be used for other sensors as well, such as optical motion capturing and force plates. The concept is to add virtual sensors to a musculoskeletal model, and then find a simulation that minimizes two objectives: (1) effort and (2) a tracking error between the measured data and the simulated data from the virtual sensors. This way, we can find the movement of the musculoskeletal model that best matches the experimental measurement, and thus the actual recorded movement. When experimental data is not available, the movement will be the most likely movement according to the first objective (effort minimization). Therefore, this approach can also be used to analyse incomplete data, e.g. when using a sparse sensor set with only a small number of IMUs. The tutorial will use our “BioSimToolbox”, which is mostly in MATLAB and some c-code.
ISEK Workshop: High Density Surface Electromyography – from recording to motor unit identification and analysis
Silvia Muceli, Chalmers University; Francesco Negro, University of Brescia; Francois Hug, Université Côte d’Azur; Simon Avrillon, Nantes University; Giacomo Valli, University of Brescia.
A robust and accurate estimation of muscle activation during voluntary contractions is crucial for understanding human movement. Traditionally, this has been achieved using bipolar surface electromyography (EMG), which provides only a limited perspective on overall muscle activation and precludes the possibility to identify individual motor unit activity. Recently, multichannel technology has been introduced to record EMG signals in humans during voluntary contractions. This involves high-density (HD) two-dimensional surface electrode arrays that can cover large areas of the investigated muscles. This non-invasive technology is easily adaptable to both laboratory and clinical settings and offers significant advantages over traditional bipolar recordings. It provides critical insights into both peripheral (physiological and geometrical muscle fiber properties) and central (neural drive) aspects of muscle function. Moreover, HD surface EMG recordings can be decomposed into individual motor unit activity, using advanced blind source separation techniques, offering a non-invasive, direct, and reliable measure of alpha motor neuron activity innervating various muscles. The objectives of this tutorial are: 1) To provide an overview of the characteristics of high-density surface EMG signals and best practices for its reliable detection during voluntary contractions in humans, 2) To explain the fundamentals of motor unit decomposition from high-density EMG signals, 3) To demonstrate dimensionality reduction techniques applied to motor unit spike trains and their relevance for understanding muscle control, 4) To offer hands-on experience on motor unit decomposition and processing motor unit spike trains using open-source tools.
