Adults' dental cone beam computed tomography images dataset for detecting and classifying missing teeth

Area of Science:

• Radiology
• Dental Imaging
• Artificial Intelligence in Medicine

Background:

• High-quality Cone Beam Computed Tomography (CBCT) datasets with Three-Dimensional (3D) annotations are scarce.
• Existing datasets lack precise 3D localization of tooth loss sites, especially with metal artifacts or implants.
• This limits the development of automated dental implant planning systems.

Purpose of the Study:

• To curate a comprehensive CBCT dataset with 3D annotations of missing teeth.
• To facilitate the training of deep learning models for automated dental implant planning.
• To establish baselines for AI-driven dental rehabilitation research.

Main Methods:

• A CBCT dataset was curated from 158 patients across three institutions.
• Cases were screened for image quality and annotated using 3D Slicer, with artifact characteristics delineated.
• 85 volumes with 114 missing tooth sites were selected for detailed 3D annotation (4,994 slices).

Main Results:

• A novel CBCT dataset with 3D annotations of 114 missing tooth sites was created.
• Comprehensive baselines were established using clinical evaluations and deep learning models.
• The dataset includes detailed information on artifact characteristics.

Conclusions:

• The curated dataset provides a valuable resource for training AI models in automated dental implant planning.
• Publicly available data will foster research in AI-driven dental rehabilitation.
• This work addresses a critical need for high-quality annotated data in dental AI research.