Evaluation of Dental Prosthesis Colors Using a Neural Network

One of the most important tasks in the work of dental technicians is to reproduce the color of dental prostheses. Experienced dental technicians can correctly estimate the true color from a photograph of the patient's teeth taken by the dentist, but this is a very difficult process for inexperienced dental technicians. In this study, using machine learning techniques, we aimed to develop an automatic color evaluation system that possesses the same skill as experienced dental technicians. We prepared 62 images of the patient's teeth that were annotated by experienced dental technicians, and we attempted to estimate the true color by using a neural network. We confirmed that it is difficult to completely evaluate the true tooth color, but some color candidates could be obtained correctly.