A population-based study to assess two convolutional neural networks for dental age estimation

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Forensic Science

Background:

• Dental age estimation is crucial for clinical and forensic applications.
• The use of convolutional neural networks (CNNs) for dental age estimation is an emerging field.
• Previous research has not explored VGG16 and ResNet101 for this purpose.

Purpose of the Study:

• To investigate the efficacy of VGG16 and ResNet101 CNN models for dental age estimation.
• To evaluate AI-based methods for age estimation in an Eastern Chinese population.

Main Methods:

• Utilized 9586 orthopantomograms (OPGs) from individuals aged 6-20 years.
• Applied VGG16 and ResNet101 CNN models for automated dental age calculation.
• Evaluated model performance using accuracy, recall, precision, F1 score, and an age threshold.

Main Results:

• VGG16 outperformed ResNet101 in overall prediction performance for dental age estimation.
• VGG16 achieved higher accuracy (93.63%) than ResNet101 (88.73%) in the 6-8 year age group.
• VGG16 showed a smaller age-difference error, particularly in younger age groups, though performance dipped in the 15-17 group.

Conclusions:

• VGG16 is a more effective CNN model than ResNet101 for dental age estimation using OPGs in the studied population.
• CNNs, specifically VGG16, show significant potential for future applications in clinical dentistry and forensic science.