Dental Caries diagnosis from bitewing images using convolutional neural networks

Area of Science:

• Artificial Intelligence in Medical Imaging
• Dental Radiology
• Machine Learning for Healthcare

Background:

• Dental caries (tooth decay) is a prevalent condition affecting all age groups, particularly children.
• Bacterial acid production erodes tooth structure, causing discoloration, pain, and sensitivity.
• Current manual diagnosis methods are time-consuming; AI offers improved accuracy and efficiency.

Purpose of the Study:

• To develop and evaluate a data-driven model for accurate dental decay diagnosis using Bitewing radiology images.
• To leverage convolutional neural networks (CNNs) for automated detection of tooth decay.
• To explore the potential of AI in enhancing dental diagnostic processes.

Main Methods:

• Utilized a dataset of 713 patient Bitewing X-ray images from June 2020 to January 2022.
• Processed images using four distinct CNN architectures: AlexNet, ResNet50, VGG16, and VGG19.
• Resized images to 100x100 and split into 70% training (4219 images) and 30% testing (1813 images) sets.

Main Results:

• The VGG19 CNN model demonstrated the highest diagnostic accuracy among the evaluated architectures.
• Achieved an impressive accuracy rate of 93.93% in identifying dental caries.
• This highlights the effectiveness of specific CNN models in dental image analysis.

Conclusions:

• The study indicates strong potential for an automated AI-based dental caries diagnostic system.
• Such a system could function as a mobile app or cloud-based service, aiding dentists.
• This AI tool could significantly improve the early detection and management of tooth decay.