Articular human joint modelling
Robotica
| Hi-index | 0.00 |
This dissertation presents a computational framework for integrating measured data—such as medical images, tracked motion, and anatomy-book knowledge—into the predictive modeling of anatomical joints. The framework is data-driven in the sense that it uses sampled motion data to infer soft-tissue geometry and behavior. The framework allows the generation of adaptable, quantifiable, predictive models and simulations of complex joints, surpassing current measuring limitations.I instantiate the framework in a collection of tools: (1) a sub-voxel accurate method for tracking bone-motion from sequences of medical images; (2) computational tools for estimating soft-tissue geometry and contact; and (3) a tool for the visual and quantitative exploration of joint biomechanics. The first tool attains accuracy improvements of more than 74% over current tracking methods, when compared to the ground truth computed from marked data; the accuracy improvement enables the analysis of soft-tissue deformation with motion in live individuals. The second tool enables us to overcome current soft-tissue in vivo imaging limitations. The third tool facilitates the quantitative and visual analysis of joint models and simulations.The resulting computational models are somewhat unusual in their hybridization of data representations. Each representation has strengths for various aspects of the modeling and I combine them in unique ways to achieve simple, elegant and accurate estimations of biologically relevant measurements. I demonstrate the application of this framework to the human wrist and forearm. The results generated through this framework have already impacted orthopedists' understanding of the many diseases afflicting human joints. With such a better understanding, improvements in treatment for injuries are possible as well as reductions in injuries.