UBC Theses and Dissertations
Towards dynamic, patient-specific musculoskeletal models Levin, David Isaac William
This thesis focuses on the development of tools to aid in producing dynamic simulations from patient specific volumetric data. Specifically, two new computational methods have been developed, one for image acquisition and one for simulation. Acquiring patient-specific musculoskeletal architectures is a difficult task. Our image acquisition relies on Diffusion Tensor Imaging since it allows the non-invasive study of muscle fibre architecture. However, musculoskeletal Diffusion Tensor Imaging suffers from low signal-to-noise ratio. Noise in the computed tensor fields can lead to poorly reconstructed muscle fibre fields. In this thesis we detail how leveraging a priori knowledge of the structure of skeletal muscle can drastically increase the quality of fibre architecture data extracted from Diffusion Tensor Images. The second section of this thesis describes a simulation technique that allows the direct simulation of volumetric data, such as that produced by the denoising algorithm. The method was developed in response to two key motivations: first, that the medical imaging data we acquire is volumetric and can be difficult to discretize in a Lagrangian fashion, and second that many biological structures (such as muscle) are highly deformable and come into close contact with each other as well as the environment. In response to these observations we have produced an Eulerian simulator that can simulate volumetric objects in close contact. The algorithm intrinsically handles large deformations and potential degeneracies that can result in contacting scenarios. Extending the simulator to produce complex musculoskeletal simulations is also discussed. These two algorithms address concerns in two stages of a proposed pipeline for generating dynamic, patient specific musculoskeletal simulations.
Item Citations and Data
Attribution-NonCommercial-ShareAlike 3.0 Unported