Basic treatment goal in patients with facial asymmetry is the correction of the deviated midline of the maxilla, mandible, and chin point. On the other hand, Yanez-Vico et al. reported that the angle of the mandibular ramus, on both frontal and lateral planes, determined apparent facial asymmetry [17]. Hwang et al. also commented that some patients complained of mandibular asymmetry even after successful correction of chin deviation, so the operators should pay attention to the improvement of the condylar axis, such as frontal and lateral ramal inclination [11]. Because there are a lot of limitations to evaluate 3D skull morphology by using frontal and lateral cephalometric X-rays, for successful correction of facial asymmetry, 3D evaluation of facial asymmetry by using 3D-CT is necessary. Various methods have been reported to evaluate facial asymmetry based on 3D-CT. Damstra et al. suggested a combined 3D and mirror-image analysis for the diagnosis of facial asymmetry [26]. The others reported the evaluation method using facial asymmetry index [13, 16, 17].
To use these methods, identifying reference landmarks and establishing appropriate reference planes are crucial steps for the evaluation of facial asymmetry. In 3D environments, when the reference planes are established, clinicians should consider not only the horizontal and vertical position of reference planes, but also the rotational position such as yaw, pitch, and roll [27].
Yanez-Vico et al. used mid-dorsal position of the foramen magnum, bilateral points of the external auditory meatus, and foramen spinosums which were located in the middle and posterior cranial base because they thought this area might be the most stable area during development [17].
Katsumata et al. used the plane which passed through sella, nasion, and dent as a midsagittal reference plane [13]. And two more planes perpendicular to this midsagittal plane were selected as horizontal and coronal reference planes.
In this study, nasion, sella, and MidZ were selected as the midsagittal reference points. MidZ was used as a reference point instead of basion or dent. If the landmarks like basion and dent which were located on the posterior part of the cranium were selected as the reference points, posterior cranial bone asymmetry could affect the evaluation of anterior mandibular asymmetry, such as chin point deviation. In PA cephalometric analysis, Trpkova et al. reported that the perpendicular line through midpoints between pairs of orbital landmarks showed excellent validity as the vertical reference line [2]. In CBCT analysis, Park et al. used bilateral Z points and orbitale in 3D reconstruction images and reported that the transverse reference line using these landmarks might be used even in patients with a severe asymmetry of the maxilla when this was used with reference to the clinical photos [28]. If the clinicians chose the posterior cranial landmarks as reference landmarks, it is difficult for clinicians to use them on clinical examination because they cannot see and measure the landmarks. If the clinicians used orbital landmarks as reference landmarks, the clinicians are able to compare the degree of asymmetry on CT images with that on clinical examination and photos. If the patient has obvious asymmetry in the orbital area, it is better to allow orbital asymmetry in setting the reference plane rather than using posterior cranial landmarks.
Therefore, the authors used MidZ as a midsagittal reference point instead of posterior cranial bone landmarks like basion or dent.
To define the inclusion criteria for the control group, previous researches about PA cephalometric analysis were used. Some researchers showed that the critical distance of menton that distinguished symmetry from asymmetry was approximately 4 mm [21, 23]. So, in this study, the patients who had the length of the perpendicular line from menton to the Na-S-MidZ plane under 4 mm were included as the control group.
Kwon et al. proposed the similarity index to evaluate three-dimensional asymmetry [29]. They evaluated the symmetry of the mandible using a mirror image. When overlapping the left and right of the mandible, the overlapping part is expressed by similarity index. The closer the similarity index is to zero, the more symmetrical it is. This is also a good way to evaluate facial symmetry. However, this method evaluates symmetry by dividing the mandible into two parts, ramus and body. Therefore, there is a limit in evaluating which anatomical landmarks have asymmetry.
To evaluate the preoperative facial asymmetry and postoperative improvement, reproducible identification of the landmarks is important [30]. In the studies of Katsumata and Yanez-Vico, there is no solution how to maintain the positional data of reference landmarks on the serial CT images. If the reference landmarks should be re-identified on the follow-up CT images, the index value of landmarks, such as orbitale, which were not changed by the treatment could be changed. Therefore, errors in identifying the landmarks and the change following the treatment could be mixed and represented as an index. Therefore, it may adversely affect the accuracy of the evaluation.
In this study, superimposition of serial CT images was done on the best fit of cranial base structures. In the VCeph 3D module of OnDemand 3D software, the positional data of selected landmarks on the preoperative CT images were saved and loaded on the postoperative CT images. Therefore, the position of reference landmarks was maintained on the postoperative 3D model. The measurement landmarks, which were changed following surgery, were moved to the new position on the postoperative 3D model, and the new positions were also checked on the multiplanar reconstruction (MPR) images. This method was able to minimize the error of identifying the landmarks in the follow-up CT images and improve the accuracy of postoperative evaluation.