Accurate gingival segmentation from 3D images with artificial intelligence: an animal pilot study

Area of Science:

• Dental Diagnostics
• Biotechnology
• Medical Imaging

Background:

• Gingival phenotype is crucial for dental diagnosis and treatment planning.
• Traditional methods for determining gingival phenotype are invasive and time-consuming.
• A novel, non-invasive technology using 3D soft tissue reconstruction is proposed.

Purpose of the Study:

• To evaluate the feasibility and accuracy of a novel, non-invasive technology for predicting gingival biotype.
• To assess the use of 3D soft tissue reconstruction from intraoral scanning and cone beam computed tomography (CBCT) for virtual gingival thickness measurements.

Main Methods:

• Yorkshire pig mandibles were scanned using CBCT and intraoral scanners.
• A deep-learning model reconstructed teeth and bone structures in 3D.
• CBCT and intraoral scans were overlaid for virtual soft tissue thickness measurements.
• Virtual measurements were compared with clinical measurements using periodontal probes and calipers.

Main Results:

• Clinical and virtual measurements showed a strong positive correlation (r=0.9656, P<0.0001).
• Clinically insignificant differences (0.066±0.223 mm) were observed between virtual and clinical assessments.
• Greater agreement was found at buccal sites compared to lingual sites.

Conclusions:

• AI-based virtual measurement offers an innovative technique for accurate soft tissue thickness assessment.
• This non-invasive method aids in comprehensive dental diagnosis and optimizes treatment planning.
• The technology utilizes routine 3D imaging systems for predictable clinical outcomes.