| Muscle architecture is a primary determinant of the muscle function associated with body movement. An assessment of muscle architecture is therefore of great importance, not only for investigating anatomical aspects of muscle but also for predicting its functional capacity. Most muscles have a variable complexity in their architectures, making it challenging to accurately assess them. Previous cadaveric approaches only take into account limited portions of architecture. On the other hand, conventional radiological approaches, such as ultrasonography and MRI, examine two-dimensional projected images. Neither of these approaches provides a thorough understanding of the entire muscle architecture. This may lead to under- or overestimation of architectural parameters that are significant for both clinical and computational studies. Therefore, the purpose of this thesis is to develop a computational modeling approach to facilitate quantification and reconstruction of complex muscle architecture. Cadaveric specimen data are used to investigate muscle architecture and to reconstruct accurate models. Associated geometric complexity and variation are carefully examined to yield consistent estimation of architectural parameters. This method demonstrates robustness against non-uniformity in the data and consistency over various types of muscle architecture; less than 10% error in PCSA estimation. By incorporating ultrasonographic assessment, this method is extended to approximate muscle architecture in living subjects, which enables estimation of PCSA for in vivo muscle in a more consistent manner. Validation experiments demonstrate 0:4 -- 8:4 % estimation errors between the original architecture and its approximation, depending on the anatomical complexity, which provides a practical insight into the quantification of PCSA for in vivo muscle. |
