Musculoskeletal disorder | EPFL Graph Search

Musculoskeletal disorders (MSDs) are injuries or pain in the human musculoskeletal system, including the joints, ligaments, muscles, nerves, tendons, and structures that support limbs, neck and back. MSDs can arise from a sudden exertion (e.g., lifting a heavy object), or they can arise from making the same motions repeatedly repetitive strain, or from repeated exposure to force, vibration, or awkward posture. Injuries and pain in the musculoskeletal system caused by acute traumatic events like a car accident or fall are not considered musculoskeletal disorders. MSDs can affect many different parts of the body including upper and lower back, neck, shoulders and extremities (arms, legs, feet, and hands). Examples of MSDs include carpal tunnel syndrome, epicondylitis, tendinitis, back pain, tension neck syndrome, and hand-arm vibration syndrome. Causes MSDs can arise from the interaction of physical factors with ergonomic, psychological, social, and occupationa

Task-driven hierarchical deep neural network models of the proprioceptive pathway

Pranav Mamidanna, Mackenzie Mathis, Alexander Mathis, Kai Jappe Sandbrink

Biological motor control is versatile and efficient. Muscles are flexible and undergo continuous changes requiring distributed adaptive control mechanisms. How proprioception solves this problem in the brain is unknown. Here we pursue a task-driven modeling approach that has provided important insights into other sensory systems. However, unlike for vision and audition where large annotated datasets of raw images or sound are readily available, data of relevant proprioceptive stimuli are not. We generated a large-scale dataset of human arm trajectories as the hand is tracing the alphabet in 3D space, then using a musculoskeletal model derived the spindle firing rates during these movements. We propose an action recognition task that allows training of hierarchical models to classify the character identity from the spindle firing patterns. Artificial neural networks could robustly solve this task, and the networks’ units show directional movement tuning akin to neurons in the primate somatosensory cortex. The same architectures with random weights also show similar kinematic feature tuning but do not reproduce the diversity of preferred directional tuning nor do they have invariant tuning across 3D space. Taken together our model is the first to link tuning properties in the proprioceptive system to the behavioral level.

Neuroscience2020

Musculoskeletal Model of the Human Shoulder for Joint Force Estimation

David Ingram

Human beings like all organisms, are subject to a variety of diseases. Musculoskeletal diseases such as arthritis, that affect our muscles and bones, are particularly debilitating in that they considerably limit our ability to interact with our environment. The symptoms of arthritis are joint pain and loss of movement. There is deterioration of the cartilage in our articulations. The precise determination of the underlying cause of the deterioration is a challenging task. It is believed that it is caused by excessive force in the joints due to inappropriate muscle forces. Since only forces in muscles just beneath the skin can be measured, the force hypothesis remains unproven. Musculoskeletal models are essential in analysing musculoskeletal diseases because they address the lack of information on the forces involved. Such models are used to estimate muscle and joint reaction forces. Determining the key elements in a musculoskeletal model to assess its quality raises several challenges. In this thesis, a musculoskeletal model of the shoulder is presented. The model is governed by the laws of rigid-body mechanics and is similar to a model of a cable-driven mechanism. The model contains both the kinematic and dynamic aspects of the shoulder. Applying the theory of rigid body mechanics requires a certain level of rigour to ensure compatibility between the kinematic and dynamic parts of the model. Therefore, a considerable part of the thesis is devoted to presenting the details of the model's construction. The model is designed specifically for estimating muscle and joint-reaction forces in quasi-static and dynamic situations. The muscle-force estimation problem is defined as a nonlinear program and solved in this thesis using a two-step approach. In a first step, the kinematics is constructed and inverse dynamics is used to estimate the associated joint torques. In a second step, the nonlinear program is solved using null-space optimisation. An initial solution to the estimation problem is obtained by taking a pseudo-inverse of the moment-arms matrix. The solution is then corrected using the matrices null-space to satisfy the constraints. This approach redefines the estimation problem as a quadratic program and considerably reduces the time required to find a solution. Once the muscle-forces are estimated, the joint reaction forces are deduced from the dynamic model. Muscle and joint-reaction forces are compared to other results from the literature.

EPFL2015

A Closed-Loop EMG-Assisted Shoulder Model

Ehsan Sarshari

The human shoulder is a complex musculoskeletal system. Knowledge about its kinematics and dynamics can help improve associated treatments. However, to date direct measurements of these quantities can be only granted through invasive investigations or expensive imaging techniques. Musculoskeletal shoulder models provide useful predictions of shoulder kinematics and dynamics. Nevertheless, there remain significant gaps between the model predictions and behaviors of the real system. This thesis aims at extending an existing shoulder musculoskeletal model for patient-specific clinical applications. To this end, number of improvements are considered. The initial model only considered an outstretch arm. Therefore, the elbow and the muscles spanning it are added in the extended model. To this end, the bone morphologies of the ulna and the radius and muscles architectures are obtained from MRI scans. The elbow is modeled using two hinge joints replicating its flexion/extension and pronation/supination motions. The model is developed based on anthropometric data of a single subject. Given anthropometric variabilities among subjects, it cannot predict inter-individual differences. Therefore, scaling routines are developed to scale the model to a specific subject. The model's bone segment inertial properties, skeletal morphologies, and muscles architectures are scaled according to any specific subject. The effects of anthropometric parameters on glenohumeral (GH) joint reaction force predictions are evaluated. Humeral head translations (HHT) play a crucial role in the GH joint functions. Given that the model is developed based on inverse dynamics, it falls short of predicting the HHT. Therefore, a framework is developed allowing forward-dynamics simulation of the model with a six DOF GH joint. A deformable articular contact is included in the framework defining the GH joint contact force in terms of the joint rotations and translations. A videogrammetry systems is used for recording upper extremity motions. It measures trajectories of skin-fixed markers. However, it cannot practically track scapula motions and the GH joint center. Therefore, a method is developed estimating the GH joint center and consequently scapula motions. Multi-segment optimization is used to reconstruct the measured motions in terms of joints angles. A musculotendon model is a key component for muscle-driven applications of the model. A Hill-type musculotendon model is developed. However, the initial state of the Hill-type model is not provided. Therefore, singular perturbation analysis is used to propose a method providing an initial state for the developed Hill-type model. Given that the model is over-actuated, an optimal load-sharing is used to predict muscle forces. It overlooks antagonistic muscle co-contractions. However, muscle co-contractions play crucial roles in the GH joint stability. Therefore, the load-sharing is modified such that measured electromyography (EMG) data can be incorporated. It is hypothesized that inclusions of the measured EMG can improve model predictions of muscle co-contractions. The developed model provides predictions of joints angles, muscles forces, and GH joint force and translations that are in good agreements with in vivo studies. It could be populated with pre/post operative patients of total shoulder arthroplasty to answer clinical questions regarding treatments of GH joint osteoarthritis.

EPFL2018