Digital Human: Three-dimentional face model reconstruction based on Hifi3dface model with clinical two-dimentional photos

Abstract

Background This study aims to explore the reconstruction of three-dimensional (3D) face model based on orthodontic front and lateral image, and compare with the face-scan 3D images (the gold standard) to assess the accuracy and clinical practicability of this method. Methods In this study, 23 patients (mean age 20.70 ± 5.36 years) were included to obtain the front and lateral facial images. Then we accomplished 3D face reconstruction with front and lateral photos based on the Hifi3dface which is an open-source application. The reconstruction model and corresponding face scan model were imported into Geomagic Control for preprocessing and repositioning. 3D deviation analysis was conducted to obtain the root mean square error (RMSE) of the facial and perioral areas, as well as the 3D error value of 8 landmarks. The clinical utility of this method for 3D face reconstruction was evaluated through measuring the 13 soft tissue measurements in two groups. Results According to the 3D deviation results: the RMSEs of the 3D face reconstruction model based on Hifi3dface was (2.00 ± 0.38mm, 95%CI:1.84-2.17mm); the RMSEs at the perioral region was (2.12 ± 0.51mm, 95%CI:1.90-2.34mm). Cheek, chin and nose exhibited the greatest deviation in the reconstruction model. The mean deviation of landmarks was less than 2mm, with the majority of the error occurring in the z-direction(Z > Y > X). As for soft tissue measurements, there were significant differences (P < 0.05) between the two groups in nasolabial angle, nasal angle and nasal frontal angle, while the remaining measurements had no statistical difference (P > 0.05). Conclusion 3D facial reconstruction based on Hifi3dface demonstrated an acceptable outcome, but there was still potential for improvement in terms of clinical utility. The reconstruction needs to improve, particularly in the z-axis(sagittal) direction.