Construction of 3D Human Knee Joint Models Using a 3D Statistical Deformable Model

Statistical shape model (SSM) has been established as a useful method for reconstruction of patient-specific 3D surface models, such as the hip or proximal femur using a single radiographic image of the joint [1, 2]. However, there are few studies that have reconstructed patient-specific 3D models of the entire knee joint. We propose to utilize the strong embedded spatial information in a 2D knee joint radiographic image to predict the 3D human knee joint shape model using the SSM method. We also present a preliminary study to test the accuracy of this method when applied to predict human knee joint shapes.