Periapical dental X-ray image classification using deep neural networks.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Dental Diagnostics

Background:

• Dental diseases like caries and bone loss are prevalent globally, impacting overall health.
• Delayed diagnosis of dental conditions can lead to mistreatment and adverse health outcomes.
• Accurate and timely detection of dental diseases is crucial for effective patient care.

Purpose of the Study:

• To develop and evaluate a Deep Neural Network (DNN) approach for detecting dental diseases from periapical X-ray images.
• To compare the performance of AlexNet against other DNN models (Res-Net-18, ResNet-34) for dental disease classification.
• To assess the efficacy of a pipelined DNN approach in identifying healthy versus non-healthy dental X-ray images.

Main Methods:

• A pipelined Deep Neural Network (DNN) approach was implemented using optimized Convolutional Neural Networks.
• Three state-of-the-art DNN models, Res-Net-18, ResNet-34, and AlexNet, were utilized for disease classification.
• The models were trained and evaluated on a large dataset comprising 16,000 dental X-ray images.

Main Results:

• AlexNet demonstrated superior performance, achieving an accuracy of 0.852 in classifying dental X-ray images.
• AlexNet outperformed Res-Net-18 and ResNet-34, with precision, recall, and F1 scores of 0.850.
• The Area Under the ROC Curve indicated low false-positive and false-negative rates for the AlexNet model.

Conclusions:

• The AlexNet model effectively generalizes to unseen data, proving its utility in diagnosing various dental diseases.
• The proposed pipelined DNN approach, particularly using AlexNet, shows significant potential for improving dental disease detection.
• This AI-driven method can aid clinicians in the early and accurate diagnosis of dental conditions, benefiting both oral and overall health.