Abstract
Corrective orthopedic surgeries results are difficult to be predicted and, unfortunately, sometimes unsuccessful. Other diseases resulting from a motor disability as bedsores are still poorly understood, despite a significant prevalence in the population. However, studies on these topics still insufficient especially for the analysis considering the muscle as a soft tissue volumetric organ. Muscle fascicule architectures and their correlation with movement efficiency is poorly documented, it lack of the detailed information regarding its volumetric deformations and stiffness changes along with muscle contractions.Muscle volumetric modeling, would provide a powerful tool for the personalized accurate simulation of body stresses of disabled or SCI patients during prolonged or friction contacts with standard medical devices non-adapted to particular morphologies, but also the planning of surgeries or functional electrical stimulation sequences.There is currently no software for automatic reconstruction of the architecture of fascicles, aponeurosis and tendons from MRI acquisitions of a specific subject. Actual volumetric muscle modeling is expensive in computational time, and not effective for real-time simulations of musculoskeletal system behavior with representation of physiological functions. The objective of this thesis is directed by the many contributions that have yet to make in the area. The current modeling methods based on the conventional finite element method are complex, inflexible or inaccurate in real-time. We propose a multi-model based on barycentric mapping approach that decouples the muscle strain density energy function into a set of independent less complex models, with the following objectives:- Improve complex muscle architecture reconstruction from the MRI acquisitions in term of complexity and flexibility.- Split muscle modeling into simple independent models, to offer more flexibility and reducing complexity of modeling which allows to have independent resolutions between deformable elements and muscle fiber elements..- By reducing the number of finite elements ensuring consistency of results of force and deformations, we reduce the computation time required for each simulation.Our methods are inspired by the previous work on the three-dimensional representation of the geometry and the complex architecture of muscles [Blemker and Delp, 2005]. In addition, the mathematical definition is studied [Chi et al., 2010] to characterize the energy density of deformations of skeletal muscle.Related with the above methods, we demand the following advances:- Improved three-dimensional representation of specific patients with muscle architecture and complex geometry from MRI measurement for personalized modeling. The method is more flexible and faster than previous.- A novel modeling method for muscle deformation via decoupled modeling of solid and muscle fiber mechanics is established. This new modeling allowed independent definitions between deformable elements and fiber force generation elements while keeping its muscle deformation accuracy. The performance is compared to conventional FEM method. - We reach high computational speed on standard machines for muscle complex simulations compared to FEM. Real-time simulation of specific person’s muscle strain and force is performed with an activation input updated in real-time from surface EMG measures.- Muscle modeling requires interdisciplinary knowledge from different research team members. The multi-model approach allows collaborative work, where each specialist focuses only on its area of expertise thanks to the modular designed modeling.