AI-based Volumetric Measurements in Subjects with Unilateral Palatal Canine Impactions: A Proof of Concept

Introduction: To establish a novel artificial intelligence (AI)-assisted cone-beam computed tomography (CBCT) segmentation method to investigate asymmetry in the maxilla in unilateral palatal canine impactions. Methods: Craniofacial CBCT datasets of subjects (n = 11) with unilateral palatal canine impactions were processed with a combination of AI-assisted automatic and user-guided segmentation techniques. Post-segmentation, three investigators independently measured the voxel-based volumes of specific maxillary structures, including the impacted and non-impacted maxillary sides and the maxillary canines. Results: High inter- and intra-investigator reliability in the volumetric measurements was seen. No significant right-left differences in the volumetric measurements of the skeletal maxillary halves (P = 0.3) or maxillary canines (P = 0.8) was observed in subjects with unilateral palatal canine impactions. Conclusions: This pilot feasibility study establishes a reliable method for future AI-assisted investigations of canine impactions using CBCT datasets. Within study limitations, right-left maxillary volumetric symmetry is observed in subjects with unilateral palatal canine impactions.