Evaluation and Initial Validation Studies of Anatomical Structure Morphing

Anatomical structure morphing is the process of estimating the patient-specific 3D shape of a given anatomy from a few digitized surface points. This provides an appropriate intra-operative 3D visualization without pre or intra-operative imaging. Our method fits a statistical deformable model to the digitized landmarks and bone surface points which are usually sparse. The statistical deformable model is constructed using principal component analysis (PCA) from an appropriate training set of objects. Our proposed technique extrapolates the 3D shape by computing a Mahalanobis distance weighted least-squares fit of this model to the minimal sparse 3D data. In this paper we present evaluation and initial validation studies of our morphing technique on 9 dry cadaver femur bones. The influence of size of the initial training set on the morphing performance is also evaluated by repeating our experiments on two different training sets of varying sizes