Application of Cone-Beam Computed Tomography-Generated Cephalograms in Children and Adolescents
Article information
Trans Abstract
This study investigates the potential of cone-beam computed tomography (CBCT)-generated cephalograms as a replacement for conventional lateral cephalograms (LCs) in children and adolescents. This retrospective study included 60 individuals, equally divided into permanent and mixed dentition groups. Both groups underwent conventional LCs and CBCT scans on the same day. LCs were then derived from CBCT scans. The same examiner performed digital measurements twice, with a week’s interval, identifying landmarks and obtaining 7 angular and 5 linear measurements. In the permanent dentition group, significant differences were observed between the two imaging modalities for 6 angular and 2 linear measurements. In the mixed dentition group, significant differences were observed for 3 angular and 2 linear measurements. However, none of these differences exceeded the clinically acceptable limit of 2.0º or 2.0 mm. No significant differences in any measurement were found between the two groups (p < 0.05). CBCT-generated LCs demonstrated comparable results with good reliability in both dentition groups, suggesting their potential as suitable alternatives for children and adolescents who require CBCT for clinical purposes.
Introduction
Since Broadbent’s introduction of cephalometric radiography in 1931, it has served as the gold standard in orthodontics [1,2]. Lateral cephalograms (LCs) have become essential for the diagnosis, treatment planning, and monitoring craniofacial growth. This technique is critical for analyzing craniomaxillofacial development and assessing orthodontic treatment outcomes by examining the interactions among skeletal, dental, and soft tissue structures [3,4]. Despite its utility, the challenge of representing a three-dimensional (3D) structure in two dimensions leads to several issues in two-dimensional (2D) cephalometry, including distortion from structural displacement, differential magnification, patient positioning errors, and inaccuracies in identifying landmarks [5,6].
Introduced to dentistry in 1998, cone-beam computed tomography (CBCT) has been widely used in orthodontic diagnosis, offering significant benefits over traditional CT, such as reduced radiation dose, lower cost, and superior spatial resolution [7,8]. 3D-CBCT complements the limitations of 2D cephalograms, enabling accurate diagnosis of critical elements, such as impacted teeth, temporomandibular joint issues, and airway problems. Additionally, CBCT images can be reconstructed into 2D images, such as lateral, posteroanterior, and panoramic cephalometric radiographs [9].
In patients who require CBCT for clinical purposes, utilizing reconstructed CBCT images for orthodontic diagnosis can reduce additional radiation exposure and financial costs. Despite previous research on 3D diagnostics, normative data for 3D CBCT volumes and emerging 3D paradigms remain undefined. Therefore, during this transitional phase from 2D to 3D analysis, numerous studies have assessed the alignment between CBCT and conventional radiography [10,11]. Previous studies suggested that CBCT-generated LCs might serve as a substitute for conventional LCs [3,12-14].
However, previous in vivo studies have been conducted with a limited number of participants or without categorization according to patient age. Few studies have specifically targeted children and adolescents, and to the best of our knowledge, no study has divided children and adolescents into dental development stages for a comparative analysis.
In children, the recognition of anatomical features in cephalograms is more challenging than that in adults because of immature bones, low contrast with soft tissues, variations in growth and development, and the coexistence of primary and permanent tooth germs [15]. Therefore, it can be anticipated that comparative studies are needed to evaluate whether CBCT-generated LCs can replace conventional LCs in children and adolescents.
Furthermore, children and adolescents are more susceptible to ionizing radiation than adults. Proving that CBCT-generated cephalograms can replace conventional cephalograms, thereby avoiding extra radiation from conventional radiography, could be particularly beneficial for minimizing radiation exposure in younger populations [16].
Therefore, this study aimed to verify the applicability of CBCT-generated LCs, a potential alternative to conventional LCs, in children and adolescents. The objective was to assess the clinical utility of CBCT-generated LCs by comparing measurements between two imaging modalities in both the permanent dentition (PD) and mixed dentition (MD) groups.
Materials and Methods
This retrospective study was approved by the Institutional Review Board (IRB) of Ajou University Hospital (IRB File No.: AJOUIRB-DB-2023-538).
1. Study population
This study included 60 children and adolescents who visited Ajou University Dental Hospital for orthodontic diagnoses, randomly allocated to the PD and MD groups. PD and MD groups each comprised 13 male and 17 female participants, with mean ages of 13.4 years (range, 10 - 19 years) and 9.2 years (range, 7 - 11 years), respectively.
The inclusion criteria were as follows: (1) patients who underwent both conventional LC and CBCT on the same day by the same operator; (2) patients with erupted central incisors and first molars; (3) patients whose conventional LC and CBCT images were of sufficient quality for analysis; and (4) patients without maxillofacial deformities. Patients with orthodontic appliances, large metal restorations, or blurred radiographic images were excluded.
This study included patients who required both conventional LC and CBCT. CBCT data was necessary for their diagnosis or treatment planning for the following reasons: (1) anomalies in dental position (impacted teeth, ectopic eruptions, and/or impacted supernumeraries); (2) asymmetry (chin deviation, dental midline deviation, and/or occlusal cant discrepancies); (3) anteroposterior discrepancies (skeletal class II or III malocclusion); (4) vertical discrepancies (anterior open bite, deep bite, and/or vertical maxillary deficiency or excess); (5) transverse discrepancies (lingual or buccal crossbites or excessive dental compensation of posterior teeth); (6) temporomandibular joint signs and/or symptoms; (7) tooth resorption (root resorption, external or internal resorption); (8) cysts of the jaws. Out of the total 60 patients included in the study, 2 patients had one factor, 25 patients had two factors, and 33 patients had three or more factors.
2. Methods
1) Radiography
LCs were obtained using a CS 8100SC (Carestream Dental Limited Company, Atlanta, GA, USA) with the following specifications: tube current, 2 - 15 mA; tube voltage, 60 - 90 kV; and exposure time, 2 - 12.5 s. Images were downloaded from INFINITT Dental PACS (Infinitt Co., Seoul, Korea) and saved in the JPEG format. The patients’ head position was stabilized using ear rods and a head holder, and images were captured while maintaining the Frankfort Horizontal (FH) plane parallel to the floor. CBCT scans were recorded using Dinnova3 (HDX, Seoul, Korea) under the following conditions: slice thickness, 1.3 mm; voxel size level, 0.3 mm × 0.3 mm × 0.3 mm; field of view, 20 × 19 cm; 7 mA; 120 kV; scan time, 20 s. The patients’ heads were stabilized using a chin and forehead rest. All radiographs were acquired by the same radiologic technologist.
2) Reconstruction of lateral cephalograms from CBCT scans
The CBCT scans were imported into the OnDemand3D software (Cybermed, Daejeon, Korea) as DICOM files. The “X-ray generator” menu of the software was used to reconstruct LCs, and the FH plane-based method was selected as the reorientation method. When setting up the X-ray generation method, a perspective reconstruction method matching the magnification and distortion of conventional LCs was chosen [17]. The orientation of the CBCT images was achieved by aligning the FH plane horizontally in the sagittal view, adjusting the midsagittal plane vertically in the coronal view, and orienting the transporionic line horizontally in the axial view (Fig. 1). The generated LCs were stored as JPEG files for comparison with the conventional LCs (Fig. 2, 3).
Three-plane orientation for generating the lateral cephalogram from CBCT scans. (A) Sagittal view, (B) Coronal view, (C) Axial view.
CBCT: cone-beam computed tomography.
Illustrative images obtained from a patient with permanent dentition. (A) Conventional lateral cephalogram, (B) CBCT-generated cephalogram.
CBCT: cone-beam computed tomography.
Illustrative images obtained from a patient with mixed dentition. (A) Conventional lateral cephalogram, (B) CBCT-generated cephalogram.
CBCT: cone-beam computed tomography.
3) Measurements
A single examiner measured a total of 120 cephalograms, saved in JPEG format, using V-Ceph version 8.5 (Osstem, Seoul, Korea). To ensure intra-examiner reliability, the same examiner reassessed these measurements after a week. The analysis involved seven angular and five linear measurements, widely used for skeletal sagittal, skeletal vertical, dentoalveolar, and soft tissue analyses (Fig. 4, Table 1). The measurements were based on landmarks that encompassed both the midsagittal and bilateral anatomical structures.
A depiction of landmarks and planes used in the lateral cephalometric analysis in this study. (A) Landmarks: N, Nasion; S, Sella; Po, Porion; Or, Orbitale; ANS, Anterior nasal spine; PNS, Posterior nasal spine; A, A point; B, B point; U1, Upper central incisor tip; U1R, Upper central incisor root apex; L1, Lower central incisor tip; L1R, Lower central incisor root apex; Pog, Pogonion; Me, Menton; Go, Gonion; P, Pronasale; Ls, Labialis superior; Li, Labialis inferior; Pog’, Soft tissue pogonion. (B) Planes: S-N, Sella-Nasion; Po-Or, Frankfort horizontal (FH) plane; ANS-PNS, Maxillary plane (MxP); Go-Me, Mandibular plane (MP); N-Pog, Facial plane; P-Pog’, Esthetic plane of Ricketts (E-line).
Definitions of the cephalometric measurements used in this study
3. Statistical analysis
All statistical analyses were conducted using R software (version 4.2.3; R Foundation for Statistical Computing, Vienna, Austria), with a threshold for statistical significance set at p < 0.05. The normality of the measurements in each group was verified using the Shapiro-Wilk test. Because the measurements in each group did not present a normal distribution, nonparametric statistics were conducted. The Wilcoxon signed-rank test was used to compare measurements between imaging modalities, while the Wilcoxon rank-sum test was used to compare differences in measurements between the imaging modalities across the PD and MD groups. To determine the intra-examiner reliability, the concordance correlation coefficient (CCC) was calculated.
Results
1. Reproducibility of repeated measurements in the permanent and mixed dentition groups
The CCC values, indicating the intra-examiner reliability, are listed in Table 2. Although the intra-examiner reliability was relatively higher in the PD group, both the PD and MD groups demonstrated high overall reliability for all measurements (CCC > 0.8).
Intra-examiner reliability of repeated measurements in the permanent and mixed dentition groups
2. Discrepancies in measurements between conventional and CBCT-generated lateral cephalograms in the permanent and mixed dentition groups
Statistical comparisons of the 12 measurements between the conventional and CBCT-generated LCs in the PD group are shown in Table 3. Statistically significant differences were observed in six angular measurements (SNB, ANB, FMA, SN-MP, U1-MxP and L1-MP) and two linear measurements (Upper lip-E line and Lower lip-E line). Comparisons of the measurements between the two imaging modalities in the MD group are presented in Table 4. Fewer measurements showed statistically significant differences compared to the PD group, with significance demonstrated in 3 angular measurements (ANB, FMA, and L1-MP) and 2 linear measurements (U1-NPog and Lower lip-E line). Although some measurements revealed significant differences, the average differences between the two imaging modalities were less than 2.0° or 2.0 mm, which is below the standard error for repeated measurements.
Discrepancies in measurements between conventional and CBCT-generated lateral cephalograms in the permanent dentition group
Discrepancies in measurements between conventional and CBCT-generated lateral cephalograms in the mixed dentition group
3. Comparison of differences between two imaging modalities across the permanent and mixed dentition groups
A comparison of the differences between the two groups revealed no statistically significant differences in any of the measurements (Table 5). In the PD group, the mean differences between the two imaging modalities ranged from -0.8º to 0.8º and -1.6 mm to 0.0 mm. In the MD group, the ranges were -1.4º to 0.7º and -1.0 mm to 0.1 mm. While the N-Me measurement, which is relatively longer, showed differences of -1.6 mm and -1.0 mm, discrepancies in linear measurements for relatively shorter items were all within ±1.0 mm.
Discrepancies in measurements between conventional and CBCT-generated lateral cephalograms across the permanent and mixed dentition groups
Discussion
Herein, we evaluated the applicability of CBCT-generated LCs in substituting conventional LCs in children and adolescents. Several studies have explored the use of CBCT in cephalometry, proposing a shift towards direct 3D cephalometric analysis. However, its application in orthodontic practice remains limited owing to the lack of established 3D normative data [18,19]. Despite the high accuracy and reliability of 3D CBCT cephalometry, most clinicians often favor 2D cephalometric analysis, finding it more familiar and comfortable [20]. Therefore, many researchers have sought to determine whether CBCT-generated cephalograms can replace conventional cephalograms during the transition from 2D to 3D analyses in orthodontics [21].
In vitro studies using dry skulls have found that the consistency and accuracy of CBCT-generated cephalograms are consistent with those of conventional cephalograms [5,17,18,22]. Similar results have been observed in in vivo studies [3,12,23,24]. Several studies have shown that measurements on CBCT-generated cephalograms do not significantly differ from those on conventional cephalograms, suggesting that CBCT-generated cephalograms could successfully replace conventional ones in clinical practice [25-28]. Some researchers have reported that CBCT-generated cephalograms demonstrate superior reproducibility and accuracy compared to conventional cephalograms [10,29-31].
Previous studies have indicated that for patients requiring a CBCT scan for diagnostic purposes, additional 2D radiographs are not required. However, it has not yet been proven whether CBCT-generated cephalograms are also applicable to children in the mixed dentition stage or adolescents in the permanent dentition stage. Recognizing anatomical features can be more challenging in pediatric cephalometry than in adults owing to developmental differences [32]. This complexity may limit the ability of CBCT-generated cephalograms to replace conventional ones in these populations. Since standard norms for 3D CBCT volumes have not yet been established in the pediatric and adolescent population, patients requiring CBCT data had to obtain conventional LCs despite the additional radiation exposure. If CBCT data alone were sufficient for orthodontic analysis in children and adolescents, obtaining additional conventional LCs would be unnecessary [4,12,20]. Therefore, this study evaluated the potential of CBCT-generated LCs to substitute conventional LCs in children and adolescents. Additionally, because dental development is more easily assessed than skeletal development on cephalograms, the samples were divided into two groups based on dental development stages for a more lucid analysis [15].
Although the PD group showed significant differences in more categories when comparing the two imaging modalities, the mean differences did not exceed 2.0º or 2.0 mm, which are unlikely to hold clinical significance. Previous studies have also stipulated that even if statistically significant differences are observed, differences within 2.0º or 2.0 mm are considered to be within the standard error of repeated measurements and are clinically acceptable [12,19]. However, for shorter distances, a 2.0 mm difference can be more significant for clinicians. Generally, a 2.0 mm difference is more impactful for shorter distances and less so for longer distances [12]. In this study, only the N-Me value, which is relatively longer, showed differences of -1.6 mm and -1.0 mm, while the other shorter measurements showed discrepancies within ±1.0 mm.
In cephalometric analysis, the use of ratios between linear measurements is more relevant for orthodontic diagnosis and treatment planning than absolute lengths, as they provide a standardized assessment of craniofacial proportions. Ratios help to normalize individual variations in growth and development, offering more consistent diagnostic insights. Therefore, length differences might be less critical compared to the diagnostic value of proportional relationships.
The diagnostic value of cephalometric analysis relies heavily on the accuracy and precision of anatomical landmark identification [33]. “Accuracy” ensures the absence of measurement errors, while “precision” indicates minimal deviation across repeated measurements of the same subject [34]. In this study, intra-examiner reliability or all values was high, consistently exceeding 0.8, demonstrating the precision of repeated measurements.
In both the PD and MD groups, the differences between the two imaging modalities were below the clinical threshold in this study; however, some measurements showed significant differences between the two imaging modalities in both groups. Systematic and random errors in cephalometric analyses may have influenced these findings [35]. Systematic errors are primarily due to projection inaccuracies and can arise when different examiners or analysis periods are involved [21,34]. Random errors involve landmark identification, tracing, and measurements, with landmark identification discrepancies being a major contributor [35]. In cephalometric studies, landmark identification errors are identified as the largest source of error, significantly exceeding projection errors and estimated to be five times greater than measurement diversity [36,37].
Landmark identification variability exhibited distinct patterns [38]. In both groups, significant differences were observed in ANB, FMA, L1-MP, and Lower lip-E line measurements. Although these measurements are defined by several landmarks, it appears that the identification of landmarks, such as B, Go, Po, Or, L1R, and Li, contributed to the significant differences between the images observed in this study. This was because other measurements such as SNA, N-ME, and L1-NPog were not significantly different between the two imaging modalities.
Previous studies have demonstrated that landmarks, such as B and Go, which are located along a gradual curve, are difficult to identify [38,39]. The landmarks Po and Or, which define the FH plane, are challenging to identify accurately due to the overlap of various anatomical structures [12,40]. Generally, L1R and Go are among the least reliably identified cephalometric landmarks on conventional LCs [19]. The inaccuracy in identifying the Lower lip-E line can be attributed to the reliance on the reproducibility of a neutral position of the movable parts of the profile [39]. Additionally, it is not possible to ensure that the position of the lips is the same across different imaging modalities [41].
Although not as impactful as errors in landmark identification, projection errors due to patient head positioning may have influenced the outcomes. CBCT scans allow for the correction of head-positioning errors when reconstructing LCs, a feature not available with conventional LCs [42]. This difference may contribute to discrepancies between CBCT-generated LCs and conventional LCs. To minimize such errors, a skilled radiologic technologist consistently used standardized criteria for all imaging procedures. Despite these efforts, children and adolescents may still struggle to maintain proper head positioning during imaging, which is a limitation. Nonetheless, all the observed differences were within clinically acceptable limits.
An additional factor contributing to projection errors is the misalignment of the X-ray source. Given that the patients were imaged at different times, the alignment of the X-ray source may have varied over time, despite regular checks of the X-ray unit. Such misalignments could introduce systematic errors into the cephalometric analysis [42].
There are various options for reconstructing CBCT images. In this study, based on the previous literature, we selected conditions that were most similar to those of conventional LCs. The RayCast technique, which is more reproducible than maximum intensity projection (MIP), was primarily used [3]. Additionally, to achieve magnification and distortion similar to those of conventional LCs, perspective projection was used with the image creation algorithm of the OnDemand 3D program. However, the actual magnification depends on the reconstruction algorithm of the OnDemand3D program, which may have caused discrepancies between the calculated and actual magnifications [17,43]. Since V-Ceph software does not support Dicom files, all images were converted to JPEG format. A previous study has indicated that converting DICOM images to JPEG files does not alter the reproducibility of landmark identification [44].
In this study, a single examiner performed all the measurements. As intra-examiner errors are generally smaller than inter-examiner errors in cephalometric analysis, it is justifiable for one examiner to conduct the analysis in such studies [26].
As previously mentioned, the differences between pediatric and adult patients must be considered in pediatric cephalometric analysis. Anatomical feature recognition around landmarks is more challenging in children and adolescents, who exhibit greater biological variability compared to adults. Consequently, an observer’s training level and experience can significantly affect the analysis [45]. Additionally, the cumulative effect of ionizing radiation and the greater vulnerability of children to radiation-induced cancers should also be considered. Therefore, clinicians should adhere to the “As Low As Reasonably Achievable” (ALARA) principle when selecting patients for CBCT imaging [46].
The ALARA principle has evolved into “As Low As Diagnostically Acceptable” (ALADA), which underscores the importance of optimizing radiation doses to ensure adequate image quality. Additionally, the new “As Low As Diagnostically Achievable being Indication-oriented and Patient-specific” (ALADAIP) concept acknowledges the need for customized image quality based on individual patient scenarios, promoting tailored imaging assessments. Therefore, CBCT should be utilized selectively in cases where 2D radiographs fall short of the diagnostic requirements. According to the European Academy of Paediatric Dentistry (EAPD) policy and the American Academy of Oral and Maxillofacial Radiology (AAOMR) recommendations, CBCT is not the primary imaging modality of choice for children and adolescents. It should only be justified in specific clinical situations where cross-sectional imaging is essential for the diagnosis and treatment planning of permanent teeth. Indications for CBCT imaging may include clinical cases with impacted or ectopic teeth with eruption disturbance, asymmetry, anteroposterior, vertical, or transverse discrepancies, and TMJ signs and/or symptoms [16,46].
A single CBCT scan can generate essential orthodontic images, such as temporomandibular joint radiographs and lateral, posteroanterior, and panoramic cephalometric radiographs. Effective doses for panoramic radiographs and lateral cephalograms typically range from 26.9 to 30 μSv, whereas those for CBCT scans vary between 132 to 210 μSv. While CBCT scans initially present a higher radiation dose than individual radiographs, their use can substantially reduce the overall radiation exposure when justified. Additionally, employing ultra-low dose CBCT technology can further decrease the effective dose to as low as 18 μSv, optimizing radiation safety [19].
This study had certain limitations. First, it was conducted at a single institution, and a multicenter study would have yielded more generalizable conclusions [47]. Second, to minimize observer-induced errors in cephalometric analysis, this study used a method in which a single examiner performed the analysis twice at a set interval. However, the experience and proficiency of a single investigator could have influenced the results, and a learning curve is anticipated when transitioning from conventional 2D cephalometric radiography to CBCT-generated radiography[26]. Future studies could enhance reliability by involving multiple trained examiners and comparing their measurements. In addition, employing automated analysis systems to reduce human errors would be meaningful [48]. Third, as LCs are the gold standard in orthodontic diagnosis, this study initially focused solely on LCs. Further comparative studies with other 2D radiographs, such as frontal cephalograms, could expand the clinical applications of CBCT-reconstructed images and enhance the benefits of reduced radiation exposure and costs.
This study is significant because it is the first to assess whether CBCT-generated LCs can replace conventional LCs in children and adolescents. Our results demonstrated the clinical applicability of CBCT-generated LCs in children and adolescents. Therefore, unnecessary radiation exposure and costs can be minimized in patients requiring CBCT in clinical practice. Nevertheless, clinicians should interpret these findings with a comprehensive understanding of the limitations of this study to ensure the judicious application of these results in clinical practice.
As treatment progresses, serial lateral cephalograms can be taken and compared with the cephalogram generated from an initial CBCT scan, allowing for detailed analysis at much lower radiation doses than repeated CT scans [5]. However, caution must be exercised regarding magnification settings during CBCT reconstruction to ensure accurate comparisons between the different imaging modalities. Despite the additional effort required, this approach can significantly reduce unnecessary radiation exposure and costs for patients. Therefore, it is advisable for clinicians to consider using CBCT-generated cephalograms in orthodontic diagnosis for children and adolescents.
Conclusion
In our cephalometric analysis, the differences between the measurements derived from the landmarks identified on conventional LCs and those identified on CBCT-generated LCs were within clinically acceptable limits at both dental development stages. Statistical analysis comparing the differences between the two groups revealed no significant differences in any of the measurements, and the reliability of the repeated measurements was also high.
These findings suggest that CBCT-generated LCs could be used in place of conventional LCs in children and adolescents. Therefore, clinicians who understand the limitations of this study may selectively utilize CBCT-generated LCs for patients requiring CBCT scans, reducing unnecessary exposure to ionizing radiation and costs.
Notes
Conflicts of Interest
The authors have no potential conflicts of interest to disclose.