Accuracy and Validity of Stitching Sectional Cone Beam Computed Tomographic Images

To determine whether three-dimensionally reconstructed images of skulls created by stitching multiple cone beam computed tomographic (CBCT) images are as accurate as single images obtained using multidetector computed tomography (MDCT), 10 skull models were scanned using an optical three-dimensional scanner, MDCT, and CBCT. Cone beam CT images at 3 different levels of the skull were manually superimposed and stitched. The reconstructed CBCT images at each level were aligned and fused using computer software and then compared to the nominal reference image obtained from the optical three-dimensional scanner by determining positional errors. The reconstructed MDCT images were also compared, and the differences in the mean errors for the 3 image types compared with the nominal reference image data were evaluated. There were no significant differences between the MDCT images and the manually merged CBCT images (Wilcoxon signed rank test, P = 0.017). In contrast, there were significant differences between the MDCT images and the software-aligned CBCT images (P = 0.005). Manual stitching of CBCT sectional images at different levels can provide accurate anatomic details of the oral and maxillofacial regions.