The Accuracy of Tooth Image Reconstruction from Spiral CT, Micro-CT, and Cone-Beam CT Scans

Objectives The aim of this study was to compare 3D accuracy of tooth image reconstruction from three kinds of CT scans using 3D superimpositional method. Methods 18 sound extracted human teeth were scanned by 3D optical system, spiral CT, micro-CT, and cone-beam CT scanner. The digital teeth images reconstructed from three kinds of CT scans were superimposed onto the standard image from optical scans respectively. Distribution patterns of shape discrepancy were presented using histogram, as well as showed in different colors on the superimposed imagines. The ratio of voluminal discrepancy versus the volume of the standard image (RVD/VS) was calculated and analyzed, using the matched-pair t-test and rank sum test. Results Compared with the standard tooth image, the average RVD/VS of digital teeth images by the micro-CT, cone-beam CT, spiral CT scans were 5.11%, 20.73%, 24.60% respectively, and there were statistically significant difference among the three kinds of CT scans (P<0.01). Significant difference were also found among the anterior teeth, bicuspids, and molars (P<0.01). Histogram gave the description about the counts and magnitude of the discrepancies. Marked by difference colors, the superimposed images could give visualized information about the magnitude and distribution patterns of discrepancies. Conclusions The digital teeth models reconstructed from the spiral CT, micro-CT, cone-beam CT images are inhomogeneous enlarged, compared with the original models. As the only realizable way to individualized FEM analysis, tooth modeling by CT scans needs more efforts and refinements to improve its accuracy.