Implementation of artificial intelligence in chronological age estimation from orthopantomographic X-ray images of archaeological skull remains

Abstract

One of the primary steps in forensic dental analysis is age estimation. Alongside sex estimation, this is offers basic categorization of subjects. Whether it is used in person-identification or archaeological analysis and research, a forensic dentist will observe these parameters when starting his work. Orthopantomographic x-ray images offer a lot of data and basically represent the golden standard for identification in forensic stomatology. Deep convolutional neural networks are establishing their presence in numerous fields of medicine and therefore we have explored the possibility of their implementation in age estimation in forensic dentistry. We developed a deep convolutional neural network, based on a dataset of 4035 orthopantomographic images, captured by and kindly provided by University of Zagreb’s, School of Dental medicine. A quick, automated and accurate model was formed that opens a new door in the field of forensic dentistry. The developed convolutional neural network was used to estimate the age of 89 archaeological skull remains. The skulls were scanned with an orthopantomography x-ray machine and the received images were used as a testing dataset.  The results offered a noteworthy 73% accuracy of placing the images in correct age groups.