The Validity and Reliability of Automatic Tooth Segmentation Generated Using Artificial Intelligence

Area of Science:

• Orthodontics
• Dental Imaging
• Artificial Intelligence in Dentistry

Background:

• Accurate dental models are crucial for effective orthodontic treatment planning.
• Traditional methods for creating segmented tooth models can be time-consuming.
• Evaluating novel AI-driven approaches for dental model generation is essential.

Purpose of the Study:

• To assess the precision of artificial intelligence (AI)-generated segmented tooth models (STMs) using the CephX® software.
• To compare the accuracy of STMs produced by CephX® with those derived from intraoral scans (IOS) and Insignia™ outcomes.
• To evaluate root volume measurements between CephX® and Insignia™ systems.

Main Methods:

• 10 patients with Class I malocclusion and mild-to-moderate crowding underwent nonextraction orthodontic therapy.
• Intraoral scans (IOS) and CBCT scans were acquired before treatment.
• AI (CephX®) generated 280 STMs in DICOM and STL formats; root volumes were measured and compared using OnDemand3D and Geomagic® Control X™ software.
• Intraclass correlation (ICC), paired t-tests, and ANOVA were used for statistical analysis.

Main Results:

• The study demonstrated good agreement between measurements using ICC analysis.
• No significant differences were found in measurements between tooth types on the right and left sides.
• There were no significant differences in accuracy between the DICOM, CephX® (STL), and IOS (STL) tooth models.

Conclusions:

• The automatic AI approach using CephX® shows high precision for segmented tooth models.
• CephX® is a fast and effective tool suitable for clinical practice in patients with mild crowding and no restorations.
• AI-driven dental modeling offers a promising advancement in orthodontic workflow efficiency.