Abstract:
Statistical shape modeling has been extensively used to model the 3-D shape variations of many biological structures. Using statistical methods such as principal component analysis to reduce the dimensionality of 3-D data, complex geometric variations can be modeled using just a few parameters. These models have been used to greatly improve the robustness and efficiency of patient-specific model generation from images and sparse or incomplete data. This article will outline the basic theory of statistical shape modeling, and describe how it has been applied to the human femur as an example of its application. Further examples highlighting the lower lumbar spine, foot, and carpometacarpal joints are also presented.