Research in Medicine

Tooth decay is one of the most common chronic diseases in Iran and the world. Accurate and early diagnosis of caries, especially on proximal (interdental) surfaces, is not possible through routine clinical examination and requires radiography. However, the interpretation of radiographs by dentists can be subject to diagnostic error. The use of artificial intelligence (AI) software based on deep learning can serve as an auxiliary tool alongside the dentist to increase the accuracy of caries diagnosis. The present study shows that the EfficientNet model has excellent performance with an accuracy of 94.7%, sensitivity of 96.7%, and specificity of 96.2%. To use this technology in Iran, the following are needed:

• Provision of software and hardware infrastructure (servers, internet, cloud storage)

• Training dentists to work with the software

• Integration with the Picture Archiving and Communication System (PACS) of dental schools and healthcare centers

• Clinical validation and obtaining approval from regulatory organizations (such as the Food and Drug Organization)