Detection of dental restorations using no-code artificial intelligence.

Area of Science:

• Artificial Intelligence in Dentistry
• Medical Image Analysis
• Computer Vision Applications

Background:

• Dental restoration segmentation on panoramic radiographs is crucial for diagnosis.
• Traditional methods require specialized expertise and significant time.
• No-code platforms offer a potential solution to streamline AI model development.

Purpose of the Study:

• To develop and evaluate a no-code computer vision model for segmenting dental restorations.
• To assess the diagnostic validity of the developed model on panoramic radiographs.

Main Methods:

• Utilized a no-code computer vision platform to train a segmentation model.
• 100 anonymized panoramic radiographs were labeled by dental experts.
• Employed data augmentation (horizontal/vertical flip) and trained for 40 epochs.
• Assessed model performance using sensitivity, specificity, accuracy, precision, F1-score, and ROC-AUC.

Main Results:

• The model achieved high pixel-level performance: 86.64% sensitivity, 99.78% specificity, 99.63% accuracy, and 0.844 F1-score.
• Tooth-level performance was even higher: 98.34% sensitivity, 98.13% specificity, 98.21% accuracy, and 0.98 F1-score.
• Achieved an ROC-AUC of 0.978, indicating excellent diagnostic capability.

Conclusions:

• No-code computer vision platforms can accurately segment dental restorations on panoramic radiographs.
• This technology has the potential to democratize AI in dental research and diagnostics.
• Further validation on larger, diverse datasets is recommended for broader applicability.