Deep Learning on Dental RGB and OPG Images for Public Health and Clinical Decision Support

Abstract

The rapidly increasing rates of unrecognized dental issues, along with the ongoing barriers to timely dental health monitoring, have raised the need for the creation of Machine Learning (ML) based dental systems that are accurate and can be used by both patients and dental professionals. In an effort to bridge the circuit, a combined Deep Learning (DL) architecture analysis is implemented, which takes advantage of two imaging sources that are very different yet complementary: Intraoral Red-Green-Blue (RGB) images for the monitoring of dental health in general purposes, and panoramic Orthopantomogram (OPG) radiographs for the support of clinical decisions. The approach is composed of two specifically designed frameworks that include an RGB-based classification system that also includes the detection of early cases of dental caries, which is for teeth health monitoring for public use, and an OPG-based system for conceiving and supporting dentists in radiograph interpreting as a supporting tool in case of complex patterns in OPG radiographs.