Mandible morphological estimation: A reliability and population-specific study on a White American virtual sample

The mandible plays a critical role in dictating facial appearance. Therefore, estimation of the mandible for craniofacial approximation is crucial to ensure positive identification, as the element is commonly missing due to disarticulation. This study assessed the reliability and accuracy of existing mandibular estimation methods on a White American (WA) population and the development of population-specific standards. The sample consists of 150 WAs (75 female, 75 male) from the New Mexico Decedent Image Database. 11 mandibular and 26 cranial measurements were collected in 3D Slicer, and intra- and inter-observer errors were evaluated. A validation study was conducted, testing Ide and Rynn (2020) and Namiki et al. (2023) methods on the target sample. Sexual dimorphism was tested, and regression analysis was conducted for creating prediction formulae, with cross-validation performed on an internal hold-out validation set. Repeatability and reproducibility were achieved. Sexual dimorphism was found in most of the mandibular parameters under investigation. The comparison between errors obtained through the validation of existing methods and the newly generated population-specific formulae suggested that the latter reached higher accuracy, ranging from 2.90 to 6.15 mm for the total sample. The sex-specific model did not provide a lower error rate. This is the first study using a large WA population sample using virtual data, proving that mandibular and craniometrics can be used reliably, with some parameters demonstrating similar patterns, suggesting underlying biological and physiological causes. Population-specific regression models for WA individuals enhanced mandibular estimation accuracy, providing a foundation for future research using larger, more diverse population datasets.