Superimposing orthodontic records at different time points has been used widely to determine the craniofacial changes. The cornerstone of superimposition is stable structure. Identification of stable structures in jaws without having external references in growing patients is extremely challenging. In history, metal implants have been used as reference in 2D cephalograms to explore natural stable structure [2,3]. In 3D era, implants should continually play a crucial role in CBCT [8,9,10] and digital dental model superimposition [14,15]. However, implants produce artifact both in CBCT and IOS image, which will degrade the image quality and bring errors. The experimental animal skulls are the substitution of human skulls, which are more feasible and less expense. This study, we used goat heads to evaluate the reliability and validity of linear and angle measurements of 3D miniscrews on CBCT and IOS with actual values. We think the results of our study apply to humans as well, because the goat heads are merely used to provide a platform for miniscrew implantation, rather than simulation of human skulls. Moreover, the study is ethically impossible to be conducted on patients because of the amount of radiation exposure using CBCT at different resolutions.
As our study revealed, statistically significant overestimations of linear measurements were obtained on CBCT both at 0.12 (0.27±0.24mm) and 0.3 (0.14±0.22mm) voxels compared with actual measurements. Our results are, to some extent, consistent with several studies. Moshfeghi et al.  using gutta-percha, reported an enlargement by 0.10±0.99 mm in axial section and 0.27±1.07 mm coronal section at 0.3 voxels. However, the values for standard deviation were greater than our data. Tolentino et al.  used silica markers, but they didn’t observe statistical difference among voxels at 0.25, 0.3 or 0.4 mm. These contradictory results may be owing to the different materials as references used in studies. Schulze et al.  pointed that an extreme artifact could be produced by titanium implants. Instead of upgrading resolution, they suggested a more sophisticated reconstruction algorithm for meaningful reduction of artifacts. Moreover, when using linear measurement to evaluate the stability of miniscrews, the systematic error should be taken into consideration.
Secondly, miniscrews at two voxels presented reliable and accurate results on angle measurements with actual values. Our result supported it in clinical applications, which is important on measuring the angle stability of miniscrews after orthodontic loading. 
On account of radiation exposure, CBCT is limited for evaluation of short-term treatment effects. Instead, chairside IOS is promising on this purpose. DeLong et al.  analyzed factors influencing IOS performance, and suggested that a smooth textured surface (such as the titanium miniscrews used in our study) could worsen the digitizing performance due to spectral reflection. However, our study confirmed the clinical reliability and validity of IOS for linear and angular measurements of miniscrews, which were consistent with other studies, but was different with respect to systematic errors and their tendencies [22,23,24,25,26,27]. Our result supported that the evaluation of tooth movement of serial digital dental models from IOSs during growth or after orthodontic intervention is operable. Apart from that, we also found it quite interesting that the mean bias on the homolateral side was significantly larger than that on the opposite, implying unequal magnification in sagittal and transverse directions. Anh et al.  claimed that regions imaged later would generate more errors during configuration than regions imaged earlier. Thus, the scanning sequence could be one of the reasons for the unequal amplification effect observed in our study, and a modification is required when miniscrews are involved.
Above all, in accordance with results of literatures and this study, several suggestions are proposed when miniscrews are used to superimpose 3D image: 1.The positional stability of miniscrews should be evaluated firstly, and the linear and angular measurements on 3D models are both reliable and clinically valid. 2. Same CBCT machine with same scanning setting are required. 3. Systematic errors of miniscrew measurements on CBCT image and digital dental model from IOS should be consider when the stable structures are explored.
However, our study had limitations. First, this is an experiment on goat jaw bone and we did not include the motion artifacts during imaging. Second, the study was conducted for a single experimental condition by testing systematic errors on a single type of miniscrew, a single CBCT machine and one IOS. Whether the results of this study are suitable for other miniscrews, other CBCT machines at different voxel sizes and other IOSs is not known.