CBCT has been proven to be a valid 3D representation of the skull that is suitable for clinical and laboratorial use. To evaluate mandibular change properly, CBCT superimposition is supposed to be the best method. The voxel-based method was first described in dentistry by Cevidanes et al.  and became the most popular. Voxel-based superimposition matches the grayscale values of the voxels to superimpose the CBCT images. Ruellas et al.  suggested superimposing the mandible in growing patients on the mandibular body (mandible without teeth, alveolar bone, rami and condyles) using 3D regional registration. Koerich et al.  showed that this was a precise method for 3D mandibular superimposition in growing subjects.
However, there may be concerns about using the entire mandibular body to superimpose growing patients because certain areas of the mandibular body are remodeled during growth, such as the chin and the lingual tuberosity, which may not qualify as reference structures . Erroneous information regarding patterns of bone growth and remodeling would be obtained if biologically incorrect superimposition protocols are used . Ideally, superimposition should be based on the most stable, verified structure with good reliability.
In past studies using 2D cephalograms, stable structures in the mandible were identified with the aid of metallic implants [1-3], which can unlikely be repeated nowadays using CBCT. Fortunately, the inner cortical structure of the inferior border of the mandibular symphysis, located on the facial midline and having no transversal change during growth, was verified by implants to be stable sagittally and vertically. In this study, we used this structure as an indirect reference to identify other stable landmarks or structures in the mandible.
After several trials and errors, we found eight relatively stable landmarks. Among them, points C, D, E and Pog were located on the stable regions, which was in agreement with Nguyen’s study. With the aid of bone plates and screws, Nguyen et al. found that the chin and symphysis regions were stable areas for 3-dimensional mandibular regional superimpositions . In addition, two new landmarks (MF and MFA) on the mental foramina on each side, which cannot be viewed in lateral cephalograms, were identified. In the present study, the positional stability of C, D, and E through time was statistically tested. All three points showed high stability over a 4.6-year time interval. In addition, C turned out to be the most stable point and was used as the reference point to measure the stability of other landmarks in the mandible. The average changes in distance from C to all of the reference landmarks were less than 1.0 mm, and the change in width between the right and left mental foramina (MF- MF) was 0.44 mm, which is a relatively small change during 4.6 years of growth.
In this study, a three-way ANOVA, which gave more detailed insight into how this difference may be occurring (Table 5, Fig 6), was used to analyze the distance change between T1 and T2. The statistical analysis of the change in C-D, C-E, C-Pog, C-MF, C-MFA, and MF-MF showed that time was not an important factor in contributing to the distance change, indicating the stability of these landmarks during growth.
Reliability is another important characteristic of the reference landmark. The Cronbach α values for the 8 relatively stable landmarks were all above 0.94 in all 3 dimensions and did not exhibit scale distortion or discrepancy bias.
In this study, we estimated the dimensional change of the mandibular canal by quantifying the positional stability of the mandibular foramen (starting point of the canal) and mental foramen (terminal point of the canal). The position of the mental foramina appears stable in all 3 dimensions, with less than 1.0 mm of change. However, the position of the mandibular foramina shows some changes, such as an increase of 2.5±2.1 mm transversally and an increase of about 3.0 mm linearly from C to the right and left mandibular foramina. This finding is in agreement with the result of Krarup’s study, which analyzed normal mandibular growth using medical CTs in 10 children with Apert syndrome from 1 week to 14.5 years . The mandibular canals were relocated laterally; therefore, we should be careful in including the posterior part of the mandibular canal as a reference structure for mandibular regional superimposition. This information can provide an important foundation for mandibular regional superimposition using both methods based on landmarks or voxels.
No landmark on the molar germs was mentioned in this study, since in most cases the roots of the third molars started developing at T2. Furthermore, it is difficult to place a reliable landmark on the lower contour of the molar germ, which is a smooth and curved surface in the early stages.