External validation of cloud-based AI for anterior tooth segmentation on multicenter CBCT

Cloud-based AI segmented anterior teeth on CBCT from five scanners with 90.1% adequacy, Dice coefficients 0.96-0.97, and <0.07mm median distance. Endodontic treatment, brackets, and high-density artifacts increased refinement odds.

• Automated segmentation was adequate in 90.1% of cases across five CBCT systems, with Dice similarity coefficients of 0.96–0.97 and median absolute distance <0.07 mm.
• Endodontic treatment (OR=4.43), orthodontic brackets (OR=3.74), and adjacent high-density artifacts (OR=7.88) significantly increased the need for manual refinement.
• Automated segmentation was substantially faster than manual refinement.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.