2017 年 22 巻 2 号 p. 73-83
This paper primarily proposes a method of implementing statistical shape model (SSM) of distal femur by automatically determining subject-specific femoral coordinate system (FCS). These have a wide range of applications in both biomechanical, 3D image analysis, knee model design and surgical planning. The main challenge in implementing SSM of distal femur as being complex in shape and a part of the whole femur, is to find correspondence across subjects automatically for the purpose of aligning training subjects to a reference coordinate space. While conventional correspondence is observer-dependent and time consuming, this study is intended to tackle these drawbacks by proposing a fully user-independent method to accurately extracting the FCS using 3D magnetic resonance (MR) images of isolated knee because of its popularity in knee surgery due to high spatial resolution. The proposed methods are based on morphological analysis of distal femoral bone in 3D MR image to locate the anatomical features automatically. Afterward an SSM has been constructed by utilizing the implemented FCSs in a set of volumetric MR images. Images were reconstructed and synthesized in each dimension from the model, and finally were evaluated by benchmarks index-generalization ability. In addition, possible application of the implemented FCS could be volumetric image (CT/MR) matching of distal femur, etc.