Tooth-Related Disease Detection System Based on Panoramic Images and Optimization Through Automation: Development...

Area of Science:

• Artificial intelligence in dentistry
• Medical image analysis
• Dental diagnostics

Background:

• Early detection of tooth diseases is crucial for patient dental health and preventing complications.
• Subtle visual cues in panoramic images often lead to missed diagnoses by dentists.
• Five key diseases (coronal caries/defect, proximal caries, cervical caries/abrasion, periapical radiolucency, residual root) can be identified.

Purpose of the Study:

• To design an AI-powered real-time model for assessing five specific tooth-related diseases in panoramic images.
• To provide an auxiliary diagnostic tool for dentists, enhancing telemedicine capabilities.
• To reduce the time required for treatment planning.

Main Methods:

• Trained AI models on 10,000 panoramic images covering five distinct dental conditions.
• Utilized Fast Region-based Convolutional Network (Fast R-CNN), ResNet, and Inception models.
• Developed individual detection models for each disease and integrated them to improve overall accuracy due to indistinct features in images.

Main Results:

• The Fast R-CNN model demonstrated the highest diagnostic accuracy, exceeding 90% for the five targeted tooth diseases.
• This model facilitates real-time diagnosis of visually challenging dental conditions from radiographs.
• The system assists dentists by providing rapid insights for treatment planning.

Conclusions:

• The Fast R-CNN model offers high accuracy for real-time dental disease diagnosis, supporting dentists and shortening treatment planning.
• Continuous updates to the web service's panoramic image database are expected to further enhance diagnostic accuracy.
• The developed system diagnoses five diseases from a single panoramic image in approximately 2 minutes, proving effective for dental treatment scheduling.