Date of Award
Kevin B. Shelburne, Ph.D.
Finite element, Muscle, Musculoskeletal, Simulation
Computational modeling is a powerful tool which has been used to inform decisions made by engineers, scientists, and clinicians for decades. Musculoskeletal modeling has emerged as a computational modeling technique used to understand the interaction between the body and its surroundings. There are several common approaches used for musculoskeletal modeling which take advantage of different model formulations to obtain information of interest. Unfortunately, models with different joint formulations inherit disparities in representations of ligament, muscle, and cartilage at joints of interest. These differences affect the way the joint functions and limit the insight it provides through computational analysis. Musculoskeletal models with high fidelity joint representations in a finite element framework have become increasingly viable in recent years, but three challenges limit progression: model personalization, modeling infrastructure, and computational efficiency. The goal of musculoskeletal modeling is almost entirely to understand the motion of the body, the mechanics of the joints, and the strain on the tissues in subjects performing various activities. These interests require models that act as the subject’s body would – a very complex task. Improving on methods in model personalization for calibrating joint strength, soft tissue response, and modeling geometry will continue to drive this work toward true subject specificity. Previously, software has been released which provides a modeling infrastructure for musculoskeletal modeling using rigid body dynamics. No such framework exists to build and perform musculoskeletal modeling with high fidelity joint representations in a finite element environment. A computational framework which provides methods to scale models and estimate joint kinematics and muscle forces directly from laboratory data would improve the accessibility and usability of these complex techniques. Developing tools which promote computational efficiency and manage effective parallelization of simulation and optimization will help improve the usability of musculoskeletal finite element modeling. The purpose of this work was to improve upon methods in musculoskeletal finite element modeling by developing novel techniques to evolve the current state-of-the-art in this area of research. Specifically, the first study calibrated the knee strength response of a musculoskeletal model of the lower limb to healthy data collected from subjects. The model was then used in the second study to perform concurrent estimation of muscle forces and tissue strain in subjects performing two activities. The third study considered markerbased motion and compared it to kinematics obtained from stereo radiography-based bone tracking. As part of this study a new set of polynomial splines describing the motion in 5 degrees of freedom at the knee were provided. Lastly, a computational framework was developed which served to scale a generic musculoskeletal finite element model and perform estimations of joint kinematics and muscle forces directly from laboratory data. The goal of this dissertation was to increase the accessibility of a powerful modeling approach to researchers around the globe by developing and advancing techniques which improve the usability of these methods.
Copyright is held by the author. User is responsible for all copyright compliance.
Hume, Donald R., "Translating Data from the Laboratory into Simulation: A Computational Framework for Subject-Specific Finite Element Musculoskeletal Simulation" (2018). Electronic Theses and Dissertations. 1499.
Received from ProQuest
Donald R. Hume