construction of 3d human distal femoral surface models using a 3d statistical deformable model

construction of 3d geometric surface models of human knee joint is always a challenge in biomedical engineering. this study introduced an improved statistical shape model (ssm) method that only uses 2d images of a joint to predict the 3d joint surface model. the ssm was constructed using 40 distal femur models of human knees. in this paper, a series validation and parametric analysis suggested that more than 25 distal femur models are needed to construct the ssm; each distal femur should be described using at least 3000 nodes in space; and two 2d fluoroscopic images taken in 45° directions should be used for the 3d surface shape prediction. using this ssm method, ten independent distal femurs from 10 independent living subjects were predicted using their 2d plane fluoroscopic images. the predicted models were compared to their native 3d distal femur models constructed using their 3d mr images. the results demonstrated that using two fluoroscopic images of the knee, the overall difference between the predicted distal femur surface and the mr image-based surface was 0.16±1.16 mm. these data indicated that the ssm method could be a powerful method for construction of 3d surface geometries of the distal femur.