Preliminary exploration of stable soft tissue areas in multi-timepoint 3D facial scans

To explore the errors of using different methods for multi-timepoint 3D facial scan registration and the stable soft tissue areas suitable for multi-timepoint 3D facial scan registration. Twenty samples with pre-treatment and post-treatment large-field-of-view CBCT and 3D surface scans were included to measure and compare the registration errors (RMSE) between the results of 3D facial scans direct registration using four registration methods (including global best-fit and local surface registration methods) and voxel-based anterior cranial base superimposition. Multivariate ANOVA showed that there were differences in the RMSE of extraction status ( P = 0.004) and registration methods ( P = 0.003). There was an interaction effect between extraction status and registration methods ( P = 0.001). One-way ANOVA showed that there was a significant statistical difference ( P < 0.001) in the RMSE using different registration methods in the extraction group. The pairwise comparison results indicated that the RMSE using the “Global Best Fit” method was larger than that using the “Stable Area 1–3” registration methods. For the extraction and non-extraction groups, the RMSE mean and standard deviation of the “Stable Area 3” group (extraction group: 1.98 ± 0.56 mm, non-extraction group: 1.76 ± 0.42 mm) were smaller than other methods. There are significant differences in registration errors when using different registration methods for multi-timepoint 3D facial scans. An interaction effect was found between extraction status and registration methods. The forehead, the upper nasal bridge, and the zygomatic area may serve as relatively stable areas for multi-timepoint 3D facial scan registration. The protocol of this study (PKUSSIRB-202392145) was registered at the Chinese Clinical Trial Registry (Chictr.org.cn) with the identifier ChiCTR2000034288 (01/07/2020).