On cone-beam computed tomography artifacts induced by titanium implants

To briefly review the mathematical background of beam-hardening artifacts in cone-beam computed tomography (CBCT)-reconstruction and to investigate geometrical properties relevant for these reconstruction errors. By means of simulated and experimental results, beam-hardening effects caused by titanium implants are evaluated.

The geometrical and physical properties of the acquisition process of the projections used for 3D reconstruction are investigated and their effects on the CBCT images in the presence of titanium implants are derived. Beam-hardening effects are computed for a simplified polychromatic situation (three energy subsets of 80 and 110 kV) and compared with experimental results from a hard-plaster phantom containing two 'implants' (pure titanium rods; 4 mm diameter) exposed in two CBCT machines.

Massive absorption within a typical implant body (diameter: 4 mm) was computed for the low-energy subset of both energies (80 kV: 99.7%; 110 kV: 90.9%), whereas the high-energy subsets are only marginally absorbed (80 kV: 14.8%; 110 kV: 11.3%). Accordingly, phantom data revealed drastically reduced gray values in artifact-affected regions (3DAccuitomo: -46% to -51%) or (3DExam: -55%) plus increased noise (+67% vs. +73%), when compared with unaffected regions.

Our theoretical and experimental results prove massive beam-hardening artifacts for a typical implant diameter and typical energies of up-to-date CBCT machines. Meaningful artifact reduction has to be based on more sophisticated mathematical modeling of the actual physical image acquisition process rather than on postprocessing of the erroneous results obtained from the rather crude reconstruction algorithms used presently.