Thank you for your interest in the Forensic dentistry Hub.
As the Collection has now closed, this Hub is also closing. You can continue to explore the latest research and related content through the Collection page, where you’ll find published articles and more information connected to this topic.
Want to keep discovering relevant discussions and updates? Explore the Research Communities and follow related topics to find more content on Forensic medicine and Dentistry, as well as the journal page.
Thank you for joining the conversation and helping build connections around this important area of research!
Contribution highlights
Assessment of evidence from teeth and the alveolar bone of skulls
The study analysed tooth loss, dental trauma, and alveolar bone changes in 28 historical skulls to evaluate how reliably these features can inform forensic assessments. It found that most teeth were lost antemortem, dehiscence and fenestration were the most common trabecular bone pathologies, and although fracture‑related tissue patterns offer some clues, determining perimortem events remains difficult.
Developmental asymmetry of the clavicles and third molars and its implications for forensic age estimation
The study examined whether developmental asymmetry in clavicles and third molars affects the accuracy of forensic age estimation. It found strong bilateral correlations for both structures but notable asymmetry in a minority of cases, leading the authors to recommend bilateral clavicle imaging and side‑specific third‑molar selection to improve precision around legal age thresholds.
An autoencoder and vision transformer based interpretability analysis on the performance differences in automated staging of second and third molars
The paper presents a hybrid autoencoder–Vision Transformer framework that both improves automated staging accuracy for second and third molars and explains why the two teeth differ in model performance. It shows that remaining errors stem largely from high morphological variability in third molars, and demonstrates that relying on a single interpretability method can mask data‑driven uncertainty, making the proposed multi‑layered approach more reliable for forensic age estimation.
Sex estimation from lateral cephalograms via a hybrid multimodel convolutional neural network
The study developed a hybrid multimodel CNN that integrates landmark‑based measurements from DenseNet169 with image‑based classification from EfficientNetB3 to automate sex estimation from lateral cephalograms. It achieved very high accuracy and strong external performance, showing that combining supervised and unsupervised models through majority voting produces a more robust and reliable sex‑estimation framework for forensic applications.