Teeth Lesion Detection Using Deep Learning and the Internet of Things Post-COVID-19

Area of Science:

• Artificial Intelligence in Healthcare
• Medical Informatics
• Internet of Things (IoT)

Background:

• The post-COVID-19 era necessitates remote healthcare solutions.
• Pandemic preparedness requires innovative, non-contact diagnostic tools.
• Current dental diagnostics may require physical interaction, posing risks.

Purpose of the Study:

• To develop an automated, non-contact healthcare diagnosis model using IoT.
• To enable layperson use of AI for dental issue identification and treatment options.
• To leverage deep learning for accurate dental lesion classification.

Main Methods:

• Utilized Internet of Things (IoT) devices for data acquisition (periapical X-rays).
• Employed data augmentation, transfer learning, and deep learning (AlexNet CNN).
• Applied unsupervised neural networks, Support Vector Machine (SVM), and K-nearest neighbors (KNN) for classification.

Main Results:

• The model achieved 98% accuracy in classifying five different types of dental lesions.
• AlexNet feature extraction followed by SVM classification proved highly effective.
• Data augmentation and transfer learning improved model accuracy and prevented overtraining.

Conclusions:

• The proposed IoT-based automated model offers an effective, non-contact solution for dental diagnosis.
• The system's user-friendly interface allows layperson operation without expert assistance.
• This AI-driven approach demonstrates significant potential for remote healthcare and pandemic preparedness.