• Samuel Kadoury, Farida Cheriet, Jean Dansereau, and Hubert Labelle.
• Ecole Polytechnique de Montréal, PO Box 6079, Succursale Centre-ville, Montreal, Quebec, Canada. samuel.kadoury@polymtl.ca
• J Spinal Disord Tech. 2007 Apr 1; 20 (2): 160-7.

AbstractCurrent three-dimensional (3D) reconstruction methods based on explicit or implicit calibration procedure require a calibration object to generate calibrated x-rays for the 3D reconstruction of the human spine and the pelvis. However, to conduct retrospective studies where no 3D technology is available, 3D reconstruction must be performed from x-ray images where no calibration object was used. The current state of the art offers a variety of methods to obtain a personalized 3D model of a patient's spine, however, none have presented a clinically proven method which allows a 3D reconstruction using uncalibrated x-rays. The main objective of this study was to propose a self-calibration method using only the anatomic content of the x-ray images and evaluate its clinical feasibility on uncalibrated x-ray images for the 3D reconstruction of the scoliotic spine and pelvis. The rationale for proposing a 3D reconstruction method from uncalibrated x-rays is to allow access to 3D evaluation of spinal deformities in any standard clinical setup and to enable the conduct of retrospective studies of any kind. To assess the validity of the 3D reconstructions yielded by the proposed algorithm, a clinical study using 60 pairs of digitized x-rays of children was conducted. The mean age for this group of 60 patients was of 14+/-3 (range 8 to 18) years old. All the children in the study group had scoliosis, with an average Cobb angle on the frontal plane of 25 degrees (range 3 to 70 degrees). For each case, a 3D reconstruction of the spine and pelvis was obtained using both explicit and self-calibration methods, from calibrated and uncalibrated x-rays, respectively. Results show that 3D reconstructions obtained with the proposed method from uncalibrated x-ray images yield- geometrical models that exhibit insignificant differences for 2D and 3D clinical indices commonly used in the evaluation of spinal deformities. This allows a 3D clinical assessment of scoliotic deformities from standard x-rays without the need for calibration, and providing access to this technology in any clinical setup and allowing to perform retrospective studies, which were previously impossible.

21 keywords...

hide…