Deep learning-based detection of dental prostheses and restorations

Area of Science:

• Dentistry
• Artificial Intelligence
• Computer Vision

Background:

• Dental restorations and prostheses are crucial for oral health.
• Accurate identification of these dental elements is important for diagnosis and treatment planning.
• Automated recognition systems can aid dental professionals.

Purpose of the Study:

• To develop and evaluate a deep learning (DL) model for recognizing 11 types of dental prostheses and restorations.
• To assess the performance of the DL model using metrics like mean average precision (mAP) and mean intersection over union (mIoU).

Main Methods:

• A dataset of 1904 oral photographic images was utilized.
• A DL model was developed using TensorFlow and Keras.
• The model was trained to recognize 11 distinct types of dental restorations.

Main Results:

• The DL system achieved a mean average precision (mAP) of 0.80 and a mean intersection over union (mIoU) of 0.76.
• High accuracy (over 80%) was observed for detecting metallic dental prostheses.
• Moderate accuracy (around 60%) was achieved for tooth-colored prostheses.

Conclusions:

• Deep learning models can effectively recognize metallic dental prostheses with high accuracy.
• Current DL models demonstrate moderate accuracy in identifying tooth-colored dental restorations.
• Further research may be needed to improve the recognition of tooth-colored dental restorations.