Comparison of in vivo 3D cone-beam computed tomography tooth volume measurement protocols

Background: The objective of this study is to analyze a set of previously developed and proposed image segmentation protocols for precision in both intra-and inter-rater reliability for in vivo tooth volume measurements using cone-beam computed tomography (CBCT) images. Methods: Six 3D volume segmentation procedures were proposed and tested for intra-and inter-rater reliability to quantify maxillary first molar volumes. Ten randomly selected maxillary first molars were measured in vivo in random order three times with 10 days separation between measurements. Intra-and inter-rater agreement for all segmentation procedures was attained using intra-class correlation coefficient (ICC). Results: The highest precision was for automated thresholding with manual refinements. Conclusions: A tooth volume measurement protocol for CBCT images employing automated segmentation with manual human refinement on a 2D slice-by-slice basis in all three planes of space possessed excellent intra-and inter-rater reliability. Three-dimensional volume measurements of the entire tooth structure are more precise than 3D volume measurements of only the dental roots apical to the cemento-enamel junction (CEJ).