Estimating dental age based on panoramic radiographs is a key aspect of age estimation for forensic and civil purposes. For children and young adults, this is typically assessed based on the development of several permanent teeth of lower jaw, providing a maturity scoring according to Demirijan’s method. Although its wide usage, this assessment procedure is time-consuming, subjective and requires a trained expert.
Together with the Department of Prosthetic Dentistry in Dresden, we attempt to improve the speed, accuracy and reliability of age estimation based on panoramic radiographs (orthopantomograms, OPG) using convolutional neural networks. We are developing methods for image preprocessing (segmentation, registration) and age estimation.
We combine convolutional neural networks and methods from Bayesian statistics and variational inference to estimate both the age and its aleatoric and epistemic uncertainties.
collaborators
- Sebastian Seurig, MD student, TU Dresden, Carl Gustav Carus Faculty of Medicine, TU Dresden.
- Dr. med. Birgit Marré, Department of Prosthetic Dentistry, TU Dresden, University Hospital Carl Gustav Carus.