The naso-orbito-ethmoid (NOE) region comprises complex anatomy, and as such, NOE fractures present with a challenge during reconstruction. Restoring the pretraumatic facial form is often challenging, and patients may experience re-deformities. We evaluated 2 computer-aided methods to estimate pretraumatic facial form in silico . A sample of 3D facial images of adult Caucasian males and females was utilized to evaluate 2 geometric morphometric-based facial landmark estimation methods in silico . These include thin-plate spline (TPS) and multivariate regression. Six anthropometric facial landmarks ( nasion , pronasale , bilateral endocanthion , and bilateral alare ) representing the regional anatomic projections were artificially removed and subsequently estimated based on the location of 18 other landmarks. Between-group principal component analysis was run with permutation testing to uncover impact on facial form between original and estimated landmark configurations. Thin-plate spline-based estimation presented with superior accuracy. Conversely, the regression-based method introduced a statistically significant artificial deformity and was thus excluded from further consideration. Estimation error in the TPS-based method was evaluated by measuring Euclidean distances between the estimated and original landmarks. Mean error was highest at nasion (up to 3.34±1.7 mm). Thin-plate spline-based estimation showed promising accuracy, and the computer-assisted pipeline is time-feasible and yields a 3D-printable aid to be used intraoperatively during reconstructive surgery. Given that the current in silico study that has not been clinically translated yet, our findings, therefore, await replication and future clinical translation.
Sergani et al. (Wed,) studied this question.