O.-T. DAVID*, OANA MUNTEANU*, R.-A. TUCE*, A. NEAGU*,**#, F. BÎRSĂŞTEANU***
*Center for Modeling Biological Systems and Data Analysis, “Victor Babeş” University of Medicine and Pharmacy, Timişoara, Romania
**Department of Physics and Astronomy, University of Missouri, Columbia, U.S.A.
***Discipline of Radiology and Medical Imaging, “Victor Babeş” University of Medicine and Pharmacy, Timişoara, Romania
Preoperative assessment of bone quality in the dental arches is important in implantology. Fan-beam computed tomography (CT) proved to be successful in this respect, although it involves high costs and excessive radiation exposure. Bone density estimated by CT correlates well with the mechanical bone density scale proposed by Misch [14]. Ranges of CT numbers, expressed in Hounsfield units (HU), were associated to various bone density classes. About an order of magnitude cheaper and less harmful than CT, cone-beam computed tomography (CBCT) has attracted much attention as an alternative technique for bone quality estimation. CBCT voxel values, however, are less informative than CT numbers because they depend on position, on instrument settings and type. This paper proposes a mathematical procedure for making corrections to CBCT voxel values in order to eliminate their position-dependence and to bring them closer to CT numbers. An anthropomorphic spine phantom, as well as granular CaCl2 and bone meal were scanned using both CT and CBCT. We found a mathematical relationship between CT numbers and CBCT voxel values, and used nonlinear least squares methods to find the model parameters that provided the best fit to experimental data. We validated our procedure for a few materials and one type of CBCT unit. Further research will be needed to test the feasibility of this procedure for other types of CBCT instruments. The approach proposed here might become clinically useful if model parameters will be identified for hard tissues from the oral environment.
Key words: CBCT voxel values, bone quality, implantology.
Corresponding author’s e-mail: neagu@umft.ro.
Full text: PDF