Using Deep Learning to Detect Early Enamel Demineralization
Author Information
Author(s): He Ketai, Zhang Rongxiu, Liang Muchun, Tian Keyue, Luo Kaihui, Chen Ruoshi, Ren Jianpeng, Wang Jiajun, Li Juan
Primary Institution: West China Hospital of Stomatology, Sichuan University
Hypothesis
Can deep learning models accurately detect and classify early enamel demineralization using intraoral photographs?
Conclusion
Deep learning can accurately segment tooth surfaces and lesion contours, enhancing the precision, accuracy, and efficiency of enamel demineralization diagnosis.
Supporting Evidence
- The model achieved an F1-score of 0.856 for detecting demineralized teeth.
- Junior dentists' F1-scores improved significantly with the model's assistance.
- The model accurately segmented tooth surfaces and detected demineralized areas.
Takeaway
This study shows that a computer program can help dentists find early signs of tooth decay more easily, making it better for patients.
Methodology
A retrospective analysis using 624 digital images from 208 patients to train a deep learning model based on Mask R-CNN.
Potential Biases
Potential bias due to the subjective nature of image assessments by dentists.
Limitations
The study was based on a single institution and involved a small sample size, which may limit the generalizability of the findings.
Participant Demographics
Patients aged 14 to 44, with a majority being female (approximately 70%).
Statistical Information
P-Value
p<0.05
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website