In this dissertation, we present novel computational modeling techniques that enable us to (1) create 3D models without medical imaging data in order to study muscle fascicle behavior during contraction, and (2) directly compare 2D ultrasound muscle architecture measurements to 3D model architecture, allowing us to validate model-predicted changes in architecture during contraction as well as study and improve our understanding of these commonly used ultrasound measurements.

Muscle architecture – the arrangement of fascicles (fiber bundles) within a muscle – determines a muscle’s ability to contract to produce force and enable movement. Cadaver dissections can determine these properties ex vivo, and magnetic resonance diffusion tensor imaging (MR-DTI) is used to study in vivo three-dimensional (3D) muscle architecture. However, neither of these methods can inherently measure changes in fascicle behavior during contraction. Brightness-mode ultrasound (US) imaging is commonly used to measure changes in architecture in contracting muscle. However, these US measurements are two-dimensional (2D) while muscles’ architecture is 3D. Physics-based computational methods enable us to model 3D muscle form and architecture in order to study fascicle behavior contraction.

In project 1, we developed a method to create simple 3D CAD muscle models using cadaver architecture data. With a simple model of the medial gastrocnemius (MG), we demonstrated that we can capture the varied fascicle lengths and angles throughout a muscle and a apply a realistic material model to perform simulations of contraction in order to study muscle mechanics. In project 2, we created a method to simulate 2D ultrasound measurements using an MRI-based 3D MG model we developed. Our model successfully captured fascicle behavior that agreed with in vivo ultrasound data, enabling us to use our model for virtual experiments. In project 3, we applied our US simulation method to examine how the simplifications made during typical 2D ultrasound architecture measurements impact the accuracy of the 2D measurements relative to 3D architecture. We found that the difference between 2D/linearized and 3D fascicle length decreases as the percentage of the fascicle in the US field of view increases. This suggests that linearized fascicle lengths (as in US architecture measurements) more accurately represent 3D fascicle lengths when majority of the fascicle is captured in the FOV.

My dissertation research advances our ability to 1) create 3D muscle models without in vivo data, and 2) explore the impact of current limitations of ultrasound on the interpretation of 2D architecture measurements of 3D muscle architecture.