Open access
Jul 2026
Automated assessment of gingival biotype using deep learning on intraoral photographs
The proposed deep learning model, particularly the Vision Transformer, demonstrated high diagnostic accuracy and robust performance across gingival biotype classes, and showed improved recall and F1-scores, and better class separability in the confusion matrix, confirming its robustness and statistical reliability.
Lisa Liu, Erkang Tian, Shuqi Quan et al.
· PeerJ Computer Science · 0 citations