Comparative Evaluation of Deep Learning Models for the Classification of Impacted Maxillary Canines on Panoramic...

Area of Science:

• Dentistry
• Artificial Intelligence
• Medical Imaging

Background:

• Accurate identification of impacted maxillary teeth is crucial for dental treatment planning.
• Manual interpretation of radiographs is time-consuming and variable.
• Deep learning offers a potential solution for automated analysis.

Purpose of the Study:

• To develop and evaluate a deep learning approach for automated classification of impacted maxillary canines.
• To compare the performance of different convolutional neural network (CNN) architectures.

Main Methods:

• A retrospective study using 694 annotated panoramic radiographs.
• Transfer learning was applied to four pre-trained CNNs: ResNet50, Xception, InceptionV3, and VGG16.
• Models were evaluated using accuracy, precision, recall, specificity, and F1-score.

Main Results:

• VGG16 achieved the highest performance with 99.28% accuracy and 99.43% F1-score.
• A prototype diagnostic interface was created for potential clinical use.
• The models showed effectiveness in classifying impacted maxillary canines.

Conclusions:

• Deep learning models, especially VGG16, show promise for enhancing dental diagnostic workflows.
• Further multi-center validation is needed to confirm generalizability in clinical settings.
• Automated classification can improve efficiency and accuracy in identifying impacted teeth.