Development of a novel method for estimating the missing mandible using three-dimensional homologous models.

BACKGROUND AND AIM: Skeletal remains are sometimes discovered with the mandible missing from the skull due to external factors such as environmental conditions or scavenger activity. The absence of the mandible reduces the accuracy of forensic identification methods, including facial reconstruction. Although we previously established a two-dimensional method for estimating mandibular measurements (e.g., bicondylar breadth) commonly used in forensic practice, no method has yet been developed for estimating the entire three-dimensional mandibular shape. Therefore, the aim of this study was to establish a method for three-dimensional estimation of a missing mandible from the remaining cranium using an image analysis technique based on homologous modeling theory.

MATERIALS AND

METHODS

Postmortem CT images of 200 Japanese subjects were used to create homologous models of the cranium (8,739 vertices) and mandible (4,255 vertices), ensuring that all samples contained the same number of vertices. Principal component regression analysis was then performed using the three-dimensional point cloud data obtained from each homologous model to construct a three-dimensional estimation model. Estimation accuracy was subsequently validated using separate samples.

RESULTS

The mean error between all corresponding vertices of the estimated mandible and the actual mandible (ground truth) was 2.324 ± 0.659 mm. In addition, visualization analysis using multidimensional scaling confirmed that most of the estimated shapes were distributed close to the ground truth.

CONCLUSION

The estimation method developed in this study demonstrated a satisfactory degree of accuracy, suggesting its potential usefulness for three-dimensional estimation of missing mandibular morphology.