Bone Morphing: 3D Reconstruction without Pre- or Intra-Operative Imaging

Abstract

first group, the model can be obtain pre-operatively from radiographs or CT scans [2–4], or intra-operatively from fluoroscopic images [5]. In the second group, we will describe two subgroups: in the first one, called a geometric model, the data consist of several landmarks digitized on the patient [6, 7]. In the second group, the 3D shapes of the bones are built from data collected with a 3D optical localizer in relative coordinate systems attached to the bones, using clouds of points and deformable statistical models. This is the bone morphing technology, invented by PRAXIM (patents pending), after several years of research in deformable and statistical modeling performed at Grenoble University (TIMC Laboratory). Orthopaedic surgery can be defined as a 3D action (the operation) in a 3D space (the operating room) on a 3D object (the patient). Therefore, there is an absolute need for 3D models and we believe that models based on 2D images are of little present-day interest and without promise for the future. While geometric models, based on landmarks only, may be useful in some cases [7], they have several limitations and may not meet the entire surgeon’s requirement. Therefore, several applications based on 3D morphologic models have been developed and marketed in the late nineties; however, all these models used CT to provide the data. Building a 3D model from CT has some drawbacks: (1) The high volume of orthopaedic procedures performed each day makes CT scans as routine preoperative procedures an unacceptably expensive exercise. Also, the radiation burden to the patients would be high. (2) Performing a CT scan takes 20 min per patient. To this has to be added the time it takes to transfer the data to the  