3-D reconstruction of a surface model of the proximal femur from digital biplanar radiographs

Twenty-two femurs (one plastic bone and twenty-one cadaveric bones) with both nonpathologic and pathologic cases were considered to validate a point distribution model based three-dimensional (3-D) reconstruction technique using digital biplanar radiographs. The 3-D reconstruction technique is based on an iterative non-rigid registration of the features extracted from a statistically instantiated 3-D surface model to those identified from the radiographs. The surface models reconstructed from the radiographs were compared to the associated ground truths derived either from a 3-D CT-scan reconstruction method or from a 3-D laser-scan reconstruction method. Our experimental results demonstrated that our biplanar reconstruction technique could accurately reconstruct the surface models of both nonpathologic and pathologic femurs (average error distance = 0.9 mm), even when the point distribution model that we used were constructed from surface models of nonpathologic femurs.