A neural network has been developed in the US to detect cavities in dental X-rays

American mathematicians and doctors have developed a neural network that detects almost invisible traces of tooth decay in dental X-rays. This breakthrough simplifies the work of dentists and forms the foundation for mobile apps aimed at facilitating initial self-diagnosis.

This advanced neural network is an adapted version of the popular YOLOv7 machine vision open system, trained on photos of hundreds of patients’ upper and lower jaws.

With an impressive accuracy of 79.5%, this algorithm identified the destructive process, finding more than 83% of affected teeth, and accurately determined the boundaries of caries on almost 80% of enamel surfaces.