FDTooth and CBCT Data for AI Orthodontic Risk Assessment

Keyuan Liu's PhD thesis deposit supports AI tools for fenestration and dehiscence screening in orthodontics. It includes code and data from three chapters, referencing the FDTooth dataset of 241 patients and containing CBCT sections from 97 anonymized cases and 250 internal HKU cases. The data was deposited on figshare in May 2026.

Use Cases

• Fine-tuning vision foundation models for dental imaging based on 983 midsagittal CBCT sections.
• Training AI models for landmark detection based on manual annotations in CBCT volumes.
• Developing risk assessment algorithms for alveolar fenestration and dehiscence based on annotated intraoral photographs and CBCT data.

Strengths

• References the FDTooth dataset with 241 patient cases and annotations.
• Includes 97 anonymized cases with 983 CBCT sections for model fine-tuning.
• Data has been fully de-identified at multiple layers (filenames, headers, metadata).
• Provides a comprehensive README.txt detailing structure, ethics, and licensing.

Limitations

• The full FDTooth dataset is hosted externally on PhysioNet and not duplicated here.
• Row count and specific file formats are unknown.
• Column-level documentation is absent; field semantics must be inferred after download.

Provenance

Source

Faculty of Dentistry, The University of Hong Kong.

Collection Method

Data collected from patient CBCT scans and intraoral photographs, manually annotated.

Freshness

Last updated 2026-05-26 03:20:06; freshness should be verified.