Related Experiment Video
Updated: Jan 10, 2026
Author Spotlight: 3D Movement Assessment of Maxillary Posterior Teeth in Clear Aligner Treatment
Published on: February 23, 2024
Accuracy of Automatic Tooth Segmentation via Three Different CAD/CAM Orthodontic Software Packages: A Comparative
Objective:
To investigate and compare the accuracy of the automatic orthodontic tooth segmentation process performed by three commercial CAD software packages.
Materials And Methods:
Three typodonts, representing different degrees of crowding in both arches (mild, moderate and severe), were scanned and the files imported into three professional CAD software packages (ArchForm, OrthoSystem and VISION). In addition, 28 removable teeth, used for construction of each typodont, were scanned and their STL. files used for reference purposes. After tooth segmentation, the STL. file generated for each tooth by each software was 3D-superimposed onto its reference, and measurements were analyzed for each tooth subgroup in both arches. Both 2D discrepancies at 13 different points (mm) and central mesiodistal width (mm), and 3D discrepancies in area and volume (mm3) were investigated. Tooth segmentation accuracy was compared with respect to the reference STL. files and among software packages. The amount of crowding was investigated for its influence on the tooth segmentation process.
Results:
Most linear and 3D measurements displayed some degree of inaccuracy, with a general tendency to oversize segmented teeth. The degree of crowding did not appear to affect the accuracy of tooth segmentation, considering both arches and each individual tooth group (p > 0.05). Different software packages performed differently, though likely not to a clinically significant degree for most of the discrepancies found. In particular, OrthoSystem and VISION demonstrated better accuracy compared to ArchForm.
Conclusions:
Automatic tooth segmentation software presents variable degrees of inaccuracy, irrespective of crowding severity. The different software packages evaluated demonstrated varying levels of accuracy, with no clinically significant differences among them, except in a few cases. While most single-point discrepancies can be considered clinically irrelevant (< 0.1 mm), exceptions include the mesial, mesiogingival, and distogingival points. In contrast, discrepancies in central mesiodistal width often exceed 0.1 mm and tend to result in overestimated measurements. This overestimation can compromise the accuracy of interproximal reduction (IPR) during clear aligner therapy, potentially leading to excessive enamel removal. Regarding volume and area discrepancies, the automatic segmented teeth are generally oversized, without a significant clinical impact.
