Related Experiment Video
Updated: Jan 12, 2026
Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization
Published on: February 23, 2024
Influence of artificial intelligence-based registration and occlusal collision correction on the accuracy of occlusal
Panagiotis Ntovas1, Konstantinos Vazouras2, Abdul B Barmak3
1Adjunct Faculty, Department of Prosthodontics, School of Dental Medicine, Tufts University, Boston, Mass.; External coworker, Division of Fixed Prosthodontics and Biomaterials, University Clinics for Dental Medicine, University of Geneva, Geneva, Switzerland; Scientific affiliate, Department of Operative Dentistry, School of Dentistry, National and Kapodistrian University of Athens, Athens, Greece.
Statement Of Problem:
Artificial intelligence (AI)-driven programs and occlusal collision correction algorithms have been proposed to improve the accuracy of articulating intraoral scans in maximum intercuspal position (MIP). However, the effect of each post-processing method on the accuracy of the MIP relationship remains unclear.
Purpose:
The purpose of this clinical comparative study was to assess the influence of occlusal collision correction method and AI-driven registration on the accuracy of occlusal contacts recorded in MIP using an intraoral scanner (IOS).
Material And Methods:
Occlusal contacts in MIP were recorded using articulating film in 46 participants and digitized using maxillary and mandibular scans captured with an IOS (TRIOS 4; 3Shape A/S). For each participant, a bilateral virtual occlusal record was generated in MIP. The scans were duplicated and assigned to 3 groups: non-articulated (NA), IOS-non-corrected (INC), and IOS-corrected (IC). Using a computer-aided design (CAD) software program (Elefsina 3.2; exocad GmbH) the INC and the IC scans were post-processed implementing 3 different methods: Cast trimming (CT) and increasing vertical dimension either by moving the jaws apart (IVDM) or by opening articulator's incisal pin (IVDA). An AI-driven software program (Bite-Finder) was also used to post-process both the INC and IC scans. The same AI program was also used to articulate the NA scans in MIP, generating the NA-AI group. Occlusal contact accuracy was assessed by 2 calibrated examiners, using the digitized articulating film contacts as reference. One-way ANOVA was used to examine if average True positive contacts and false positive contacts were different across the post-processing methods (P<.05).
Results:
Significant differences were observed in both the proportion of true positive contacts and the mean number of false positive contacts among the evaluated groups (P<.05). The IC demonstrated a higher proportion of true positive contacts compared to the INC, but also exhibited a higher mean number of false positives (P<.05). The highest proportions of true positive contacts were found in the AI-registered groups (NA-AI, INC-AI, IC-AI) and the IOS-corrected groups (IC, IC-CT). AI-registered groups showed significantly fewer false positive contacts compared to their counterparts without CAD-based post-processing (IC, INC).
Conclusions:
Both occlusal collision correction via IOS or a CAD software program and AI-based registration significantly influenced the accuracy of occlusal contact representation in MIP. The choice of occlusal correction method affected the rate of true and false positive contacts.
