Применение нейронных сетей нового уровня для автоматического выявления кариозных поражений твёрдых тканей зубов

Abstract

This study is devoted to the development of an innovative system based on new-level neural networks for the automatic detection of carious lesions of hard dental tissues. Timely detection and treatment of lesions of hard dental tissues is crucial for maintaining oral health, but traditional diagnostic methods are often time-consuming and subject to subjective errors by the doctor.
To solve this problem, a neural network was created, trained on a set of panoramic dental X-rays. The model demonstrated high accuracy in detecting caries - 93.2% with a sensitivity of 85.7%. The results of testing on an independent data set showed that the proposed system is capable of identifying carious defects in real time with an accuracy comparable to or superior to the expert assessment of dentists.
The implementation of this system can significantly improve the efficiency of caries diagnostics, reduce the time and subjectivity of the process, and ensure timely detection and adequate treatment of the disease. This, in turn, will help improve the dental health of patients and reduce the burden of caries in the population. The application of the developed solution can significantly increase the efficiency of diagnostics of diseases of hard dental tissues, reduce the workload of specialists and increase the timeliness of dental care. Further testing of the system in clinical practice and its integration into dental information systems will be the next steps of this study.