3D reconstruction of lumbar vertebrae from computed tomography images

INTRODUCTION: Accurate 3D reconstruction of bony structures is essential for many clinical applications, such as evaluation of bone deformation, surgical planning and clinical follow-up of patients. Computed tomography (CT) is a widely used medical imaging technique in clinical routine, especially for bones which provide high contrast. However accurate bone segmentation still remains challenging and often requires tedious manual processing, particularly for vertebrae. Some authors attempted to perform fully automatic multi-step labeling and segmentation of the whole spine (Klinder et al. 2009). However, such framework approach did not succeed in all cases, depending on the noise in images or presence of strong calcification. The aim of this study was to perform a semi-automatic and robust 3D segmentation of lumbar vertebral body and pedicles from axial CT images. A validation study was conducted both on non-pathological vertebrae and on vertebrae with a vertebral body compression fracture.