Tooth aging monitoring system based on secondary dentin analysis using ultrasound and artificial intelligence

Teeth serve as valuable indicators for assessing biological age or determining the age of unidentified individuals due to the distinct age-related changes they undergo. However, previous studies have highlighted limitations in using dental radiographs for age estimation, such as measurement errors due to evaluator expertise, radiation exposure, and the requirement for bulky equipment. In this study, we propose a novel method for dental age estimation that is highly accurate, non-invasive, and utilizes compact equipment. We acquired ultrasound signals from teeth and analyzed the thickness of secondary dentin using machine learning. We achieved classification accuracies of 99% and 92% in grouping tooth samples by pulp-to-dentin ratio using MobileNetV2 on spectrogram images and a one-dimensional convolutional neural network on time-series data, respectively. The natural characteristics of ultrasound allow for the non-invasive acquisition of dental signals. The compact, single-element ultrasound transducers enable the acquisition of signals for machine learning without generating ultrasound images. Thus, this new method, leveraging machine learning analysis of ultrasound signals, offers a simpler and safer age estimation system that provides objective, reliable results without the risk of radiation exposure.