Semi-supervised automatic dental age and sex estimation using a hybrid transformer model

Area of Science:

• Forensic dentistry
• Radiology
• Artificial Intelligence

Background:

• Dental age and sex estimation is crucial for forensic and archaeological investigations.
• Current manual methods are subjective and rely heavily on observer experience.
• Automated methods are needed to improve accuracy and efficiency.

Purpose of the Study:

• To develop and validate a deep learning model for automatic age and sex estimation from orthopantomograms (OPGs).
• To compare the performance of the deep learning model against manual estimation techniques.

Main Methods:

• A hybrid deep learning model combining convolutional neural networks and transformer models was trained on 15,195 OPGs.
• The model's performance was evaluated on an independent test set of 1,413 OPGs.
• External validation was performed on an additional 100 OPGs, comparing results to manual methods.

Main Results:

• The deep learning model achieved a Mean Absolute Error (MAE) of 2.61 years for age estimation on the test set.
• Sex estimation accuracy was 95.54% with an Area Under the Curve (AUC) of 0.984.
• Heatmaps identified premolar and molar crowns/pulp chambers as key age-related features. External validation showed MAEs of 3.28 years (males) and 3.79 years (females).

Conclusions:

• The developed deep learning model offers a highly accurate and automated solution for dental age and sex estimation.
• This AI-powered tool has the potential to significantly assist radiologists and forensic experts.
• The model demonstrates superior performance compared to traditional manual methods.