Related Experiment Video
Updated: Jul 5, 2025
09:10
Digital Hybrid Model Preparation for Virtual Planning of Reconstructive Dentoalveolar Surgical Procedures
Published on: August 5, 2021
1.8K
Artificial Intelligence for 3D Reconstruction from 2D Panoramic X-rays to Assess Maxillary Impacted Canines
Sumeet Minhas1, Tai-Hsien Wu1, Do-Gyoon Kim1
1Division of Orthodontics, The Ohio State University College of Dentistry, Columbus, OH 43210, USA.
Diagnostics (Basel, Switzerland)
|January 22, 2024
Summary
This study explored using AI to reconstruct 3D canine positions from 2D X-rays. While AI shows promise, further development is needed for accurate dental imaging in impacted canine treatment.
Area of Science:
- Dental Imaging
- Artificial Intelligence
- 3D Reconstruction
Background:
- Maxillary impacted canines pose diagnostic challenges.
- Accurate 3D localization is crucial for treatment planning.
- Current 2D imaging limitations necessitate advanced assessment methods.
Purpose of the Study:
- To assess the feasibility of 3D reconstruction for impacted canine position using 2D panoramic X-rays.
- To evaluate a generative AI algorithm for pseudo-3D image creation.
- To determine the accuracy of AI-predicted canine locations.
Main Methods:
- Utilized Cone-Beam Computed Tomography (CBCT) data from 123 patients (74 impacted, 49 controls).
- Generated paired 2D panoramic X-rays and pseudo-3D images.
- Trained a deep-learning generative AI algorithm to predict canine position (buccal/lingual, mesial/distal, apical/coronal).
Main Results:
- AI predicted buccal/lingual position with 41% accuracy and mesial/distal with 55% accuracy.
- Mean Structure Similarity Index Measure (SSIM) for reconstruction quality was 0.71 (range 0.63-0.84).
- This is the first AI application for multidisciplinary diagnosis of impacted canines.
Conclusions:
- AI-driven 3D reconstruction from 2D X-rays is feasible but requires further refinement.
- The developed algorithm shows potential for aiding orthodontists, periodontists, and maxillofacial surgeons.
- Enhancing deep-learning algorithms is essential for robust dental reconstruction applications.
