Original Computational Framework for Forensic Age Estimation from Dental Radiographs in Nigerian Paediatric Populations

Forensic dental age estimation from radiographic imaging is critical for determining the legal age of individuals without documentation, a common challenge in juvenile justice, asylum adjudication, and unidentified remains cases in Nigeria. This doctoral study develops an original computational framework for forensic age estimation from dental radiographs specifically calibrated for Nigerian paediatric populations aged four to eighteen years. A multi-centre experimental design was employed, collecting 840 dental panoramic radiographs from children and adolescents of known age and ancestry from twelve dental centres across all six geopolitical zones, extracting Demirjian stage classifications and quantitative morphometric measurements from each radiograph, and developing a convolutional neural network model trained on seventy percent of the sample and validated on the remaining thirty percent. The model's accuracy was benchmarked against the standard Demirjian and Willems methods applied to the same sample. The study demonstrates that the computational model achieves a mean absolute error of 0.68 years for age estimation in the four-to-eighteen age range, compared to 1.24 and 1.09 years for the Demirjian and Willems methods respectively on the Nigerian sample. Population-specific performance differentials were identified between North and South geopolitical zones, attributed to nutritional and developmental factors. The study makes original contributions to forensic dentistry, paediatric biometric science, and computational forensic methodology. An open-access model deployment is proposed for use by Nigerian forensic odontologists. Keywords: dental age estimation, computational forensics, paediatric populations, Nigeria, convolutional neural network.