Do image modality and registration method influence the accuracy of craniofacial navigation?

PURPOSE

Cone-beam computed tomography (CBCT) is increasingly used in craniofacial imaging and may be an interesting option for navigated surgery. The purpose of this study was to evaluate the accuracy of various registration techniques in CBCT compared with intraoperative and diagnostic multislice spiral computed tomography (MSCT).

MATERIALS AND METHODS

High-resolution images of an anthropomorphic skull phantom with target markers were obtained using 2 CBCT machines (KaVo 3D eXam, ILUMA), an intraoperative MSCT (Sensation Open), and the standard diagnostic MSCT (LightSpeed VCT). Bone markers, a registration template, and an external registration frame were used for registration with an optical-based navigation system. Target registration errors (TREs) were evaluated and statistically analyzed in SPSS (P < .05).

RESULTS

The mean ± standard deviation for overall TREs of the KaVo 3D eXam, ILUMA, Sensation Open, and LightSpeed VCT devices were 1.37 ± 0.54, 1.67 ± 0.65, 1.27 ± 0.52, and 1.31 ± 0.30 mm, respectively. The KaVo 3D eXam showed no significant differences compared with the MSCTs. The ILUMA imaged the external registration frame only marginally and showed significant higher TREs compared with the other registration methods (P < .001). In the 2 MSCTs, no significant differences between the registration methods were found.

CONCLUSIONS

CBCT and intraoperative MSCT may show comparable TREs as standard diagnostic MSCT. Bone markers are the gold standard. Registration templates and external registration frames are valuable alternatives. When using only external registration frames, CBCTs with a large scan field are recommended.