Evaluation of root position during orthodontic treatment via multiple intraoral scans with automated registration technology

This study aimed to introduce a method for dynamically monitoring root position with intraoral scans using automated crown registration and root segmentation with artificial intelligence technology and to evaluate its accuracy using a novel semiautomatic root apical distance measurement procedure.The sample consisted of 412 teeth from 16 patients whose intraoral scans and cone-beam computed tomography (CBCT) were obtained before and after treatment. Crowns from intraoral scans and roots segmented from CBCT with artificial intelligence technology before treatment were registered, integrated, and divided into individual teeth. With an automated registration program, the virtual root was constructed by crown registration before and after treatment. The distance deviation of the root position at the apex between the virtual root and the actual root, which served as a control, was measured and decomposed into the distance deviation in the mesiodistal and buccolingual directions.The shell deviation of crown registration between CBCT and oral scan before treatment was 0.19 ± 0.04 mm and 0.22 ± 0.04 mm in the maxilla and mandible, respectively. The apical root position distance deviations were 0.27 ± 0.12 mm in the maxilla and 0.31 ± 0.11 mm in the mandible. There was no significant difference between root position in mesiodistal and buccolingual directions.Applying automated crown registration and root segmentation with artificial intelligence technology in this study improved the accuracy and efficiency of monitoring root position. In addition, the innovative semiautomatic distance measurement procedure can more precisely distinguish the root position discrepancy.