Generation of a statistical model of the anatomy of human pelvises

Abstract: Osteoarthritis and osteoporosis are two medical conditions involving the hip which affect the life quality of many people worldwide. These two diseases are diagnosed with 2D imaging by analysis of radiological measures, bone mineral density and joint space. Computed Tomography (CT) can provide 3D images of the hip, but has higher cost and imposes a higher radiation dose to the patient. Another option (which the Biomechanics group in Lund is working on) is to utilize statistical models to construct a 3D model from a 2D image. The Biomechanics group has developed a statistical model of the anatomical variability of the human femur. Adding an equivalent model for the pelvis would then allow to fully represent the hip joint. In this study, CT scans from 26 male and 21 female patients scheduled for hip replacement surgery were used to create a Statistical Shape Model (SSM) to describe the shape of pelvis. To be able to generate the SSM, the shapes of all bones were defined by identical meshes. A template mesh was created based on one of the available anatomies and it was then registered to each hip bone. The registered bones were then used to create the SSM. The registration method was evaluated by a point-to-surface distance difference. For the SSM, the shape variation and the reconstruction of the hip bones were evaluated for the whole group and for the male and female patient cohorts within the group. The SSM created during the study was able to represent the shape variation of both male and female bones. Visually, the gender variance was associated to the width and thickness of the bone, corresponding with the known differences of the pelvic bone between the genders. The results indicate that the model can represent the shape of the bone accurately, independent of gender. Combined with a statistical model for the femur, the SSM created in this study can be used to provide a 2D to 3D reconstruction of the hip from clinical diagnostic images.