A Semi-Automatic Method on a Small Italian Sample for Estimating Sex Based on the Shape of the Crown of the Maxillary Posterior Teeth

Teeth are known to be reliable substrates for human identification and are endowed with significant sexual dimorphism not only in the size but also in the shape of the crowns. In the preliminary phase of our study (already published in 2021), a novel sex estimation method based on dental morphometric geometric (GMA) analysis combined with the artificial neural network (ANN) was developed and validated on a single dental element (first upper premolar) with an accuracy rate of 80%. This study aims to experiment and validate the combination of GMA-ANN on the upper first and second left premolars and the upper left first molar to obtain a reliable classification model based on the sexual dimorphic traits of multiple maxillary teeth of Caucasian Italian adults (115 males and 115 females). A general procrustes superimposition (GPS) and principal component analysis (PCA) were performed to study the shape variance between the sexes and to reduce the data variations. The set-aside approach was used to validate the accuracy of the proposed ANN. As the main findings, the proposed method correctly classified 94% of females and 68% of males from the test sample and the overall accuracy gained was 82%, higher than the odontometric methods that similarly consider multiple teeth. The shape variation between male and female premolars represents the best dimorphic feature compared with the first upper molar. Future research could overcome some limitations by considering a larger sample of subjects and experimenting with the use of computer vision for automatic landmark positioning and should verify the present evidence in samples with different ancestry.

Automated summary

This study developed and validated a new method for sex estimation using dental morphometric geometric analysis and artificial neural network. By training and validating a classification model based on the sexual dimorphic traits of multiple maxillary teeth, the study achieved a high accuracy rate of 82%.