Early detection of tooth decay using neural networks

Behrouz Alfi¹ Sarvin Shafadeh²

IOI Code : AAHD-AAHCEB
Publication : National Conference on Information Technology, Nanotechnology, Artificial Intelligence and Technological Futures Studies(itnaf.ir)
Abstract :
    Dental caries is one of the most common oral diseases in the world, which can lead to severe pain, infection, and destruction of tooth structure if not diagnosed and treated promptly. Traditional methods of caries diagnosis, including clinical examination and 2D radiography, have limitations such as dependence on the dentist s experience, low image quality, and difficulty in identifying early lesions. In recent years, significant advances have been made in the field of artificial intelligence and deep learning, which have enabled the development of intelligent systems for the automatic detection of dental lesions. In this study, an intelligent system based on convolutional neural networks1 and U-Net architecture2 is presented, which is capable of early detection of dental caries by simultaneously utilizing 2D bitewing radiography data and 3D CBCT data. In order to overcome the problem of data scarcity, adversarial generative networks3 have been used to generate artificial data and increase the volume of the dataset. Simulation results showed that the combined U-Net + GAN model achieved 91% accuracy, 92% sensitivity, and 90% specificity, which outperforms traditional methods and even the basic CNN and U-Net models. These findings indicate the high capacity of artificial intelligence in improving the accuracy of dental caries detection and assisting dentists in clinical decision-making. This system can also pave the way for the development of application software in digital dental clinics and early detection tools worldwide. Python software was used to simulate and output the results.

Keywords : tooth decay artificial intelligence convolutional neural network u net cbct