thesis on cephalometric analysis

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• My Account Login
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 29 November 2022

Effectiveness of cone-beam computed tomography-generated cephalograms using artificial intelligence cephalometric analysis

• Eun-Ji Chung 1 , 5   na1 ,
• Byoung-Eun Yang 2 , 5 , 6 , 7   na1 ,
• In-Young Park 3 , 5 , 6 ,
• Sangmin Yi 2 , 5 , 6 ,
• Sung-Woon On 5 , 6 , 8 ,
• Young-Hee Kim 4 , 5 , 6 ,
• Sam-Hee Kang 1 , 5 , 6 &
• Soo-Hwan Byun 2 , 5 , 6 , 7  

Scientific Reports volume  12 , Article number:  20585 ( 2022 ) Cite this article

2456 Accesses

10 Citations

2 Altmetric

Metrics details

• Medical research
• Outcomes research

Lateral cephalograms and related analysis constitute representative methods for orthodontic treatment. However, since conventional cephalometric radiographs display a three-dimensional structure on a two-dimensional plane, inaccuracies may be produced when quantitative evaluation is required. Cone-beam computed tomography (CBCT) has minimal image distortion, and important parts can be observed without overlapping. It provides a high-resolution three-dimensional image at a relatively low dose and cost, but still shows a higher dose than a lateral cephalogram. It is especially true for children who are more susceptible to radiation doses and often have difficult diagnoses. A conventional lateral cephalometric radiograph can be obtained by reconstructing the Digital Imaging and Communications in Medicine data obtained from CBCT. This study evaluated the applicability and consistency of lateral cephalograms generated by CBCT using an artificial intelligence analysis program. Group I comprised conventional lateral cephalometric radiographs, group II comprised lateral cephalometric radiographs generated from CBCT using OnDemand 3D, and group III comprised lateral cephalometric radiographs generated from CBCT using Invivo5. All measurements in the three groups showed non-significant results. Therefore, a CBCT scan and artificial intelligence programs are efficient means when performing orthodontic analysis on pediatric or orthodontic patients for orthodontic diagnosis and planning.

Similar content being viewed by others

Effectiveness of 2D radiographs in detecting CBCT-based incidental findings in orthodontic patients

Jin-Young Choi, Song Hee Oh, … Gerald Nelson

Evaluation of accuracy and reliability of OneCeph digital cephalometric analysis in comparison with manual cephalometric analysis—a cross-sectional study

Akshay Mohan, Arvind Sivakumar & Prasad Nalabothu

Assessing mandibular body changes in growing subjects: a comparison of CBCT and reconstructed lateral cephalogram measurements

Cinzia Maspero, Marco Farronato, … Andrea Abate

Introduction

Since the introduction of the “new X-ray technique” for cephalometric analysis by Broadbent in 1931, cephalograms have been widely used for measuring the size and shape of craniomaxillofacial structures and evaluating their growth and development 1 . Lateral cephalograms are the representative tool in the evaluation of craniofacial growth, orthodontic diagnosis, treatment planning, assessment of treatment results, and craniofacial growth prediction 2 , 3 . However, since the conventional cephalometric radiograph displays a three-dimensional (3D) structure on a two-dimensional (2D) plane, it may produce inaccurate results when quantitative evaluation is required. For example, when structures on both sides overlap and have distinct magnifications, it is difficult to distinguish between the left and right sides. This may result in inter-examiner discrepancy and differences according to time between the same inspectors. In addition, depending on the transmission of radiation, the structures in the midsagittal region may have an ambiguous shape, thereby lowering the measurement accuracy in the overlapping structures.

Recently, owing to the innovative development of 3D radiographic techniques, such as cone-beam computed tomography (CBCT), 3D images have been used for orthodontic diagnosis. The indications of CBCT, with some evidence on its clinical efficacy, include impacted teeth, severe craniofacial anomalies, planning and evaluation of combined orthodontic-surgical treatments and bone irregularities, and temporomandibular joint malformation with accompanying signs and symptoms. CBCT has minimal image distortion because there is no difference in magnification according to the region, and important parts can be observed in detail without overlapping images. Furthermore, compared to conventional radiographs, CBCT has a higher resolution and can distinguish between tissues when there is only a difference of 10% in tissue density. In addition, images can be refined using multiplanar reformatting, surface rendering, and volume rendering through computer reconstruction, and evaluation in various directions is possible through image rotation. Clearly, CBCT has advantages over plain lateral cephalometric radiographs, but conventional cephalograms are easier to access than CBCT in many ways. CBCT has been able to address some of the limitations of conventional CT and provides high-resolution images at low radiation dosage and cost, but still exposes patients to greater doses of radiation than conventional lateral cephalometric radiographs 4 . With the rapid improvement in CBCT technology, the gap between accessible scientific data and the lawful use of CBCT is narrowing. This holds especially true for children who are more sensitive to radiation and frequently present with difficult diagnoses 5 , 6 . The two basic principles of radiological protection of patients should always be followed when considering radiation exposure for diagnostic purposes: justification and optimization 5 , 7 .

A 2D image, such as the conventional lateral cephalometric radiograph, can be obtained by reconstructing the Digital Imaging and Communications in Medicine (DICOM) data obtained from CBCT. The advantage of this technique is that there is no additional need to record a lateral cephalometric radiograph, skull anteroposterior radiograph, or submentovertex radiograph. Furthermore, when recording a conventional radiograph, the position of the radiographic film is fixed, but in CBCT, the position of the image can be modified using software. In addition, it is possible to reduce the error caused by the magnification of the left and right sides of a conventional 2D image. A CBCT-generated 2D image can be obtained with minimal magnification by using a converting program in the CBCT system.

Lateral cephalogram measurements can be performed manually or with a computer. Manual measurement methods are time-consuming, have a large measurement error, and are greatly affected by the expertise of the operator. In addition, although cephalometric analysis is typically performed by orthodontists trained in clinical practice, there have been many reports of significant intra- and inter-observer variability 8 , 9 . In computer-assisted cephalometric analysis, computerized cephalometric tracing programs, such as V Ceph (CyberMed, Inc., Seoul, Korea), Rainbow Ceph (Dentium Co, Gyeonggi-do, Korea), and Dolphin Imaging Version 8.0 (Dolphin Imaging, Chatsworth, CA), automatically evaluate the selected landmarks and calculate the distance and angles, thereby reducing inaccuracies that can arise with manual measurement 10 . However, an error might still occur in identifying the landmarks according to the skill level of the examiner in using such software 11 .

Therefore, the need for a fully automated tracing software program to improve the accuracy and reliability of cephalometric measurements is continuously increasing. Artificial intelligence (AI) is widely used in everyday applications. AI-based algorithms are found in almost every technology and used in spam filtering or online voice assistants, internet search engines, and image recognition on social media platforms. Several AI-based programs for automatically identifying anatomical measurement points are being studied currently. These include AI-based orthodontic and orthognathic online platforms, such as WebCeph (Assemble Circle, Gyeonggi-do, Korea), WeDoCeph (Audax, Ljubljana, Slovenia), and Ceph X (ORCA Dental AI, Las Vegas, NV). These are gaining popularity because of their ability to plan orthodontic treatment and obtain patient information quickly. WebCeph includes automated cephalometric tracing, cephalometric analysis, automatic superimposition, visual treatment simulation, photo gallery, and image archiving. Additionally, it enables manual landmark modification and automatic measurement computation.

In this study, conventional lateral cephalograms and lateral cephalograms generated from CBCT data were analyzed using the AI-based landmark measurement program WebCeph. The purpose of this study was to evaluate the applicability of lateral cephalograms generated from CBCT images using an AI-based cephalometric analysis program.

Cephalometric radiographs of 30 participants (15 male and 15 female) were evaluated. The distribution of skeletal malocclusion was as follows: 13 cases of class I, 14 of class II, and 3 of class III. Figure  1 shows the measurements used in this study. The results of the one-way analysis of variance (ANOVA) are shown in Table 1 . In Tukey’s post-hoc test, all measurements were distributed within a 95% confidence interval.

Measurements used in this study. ( A ) Conventional lateral cephalogram, ( B ) lateral cephalogram generated from OnDemand 3D, and ( C ) lateral cephalogram generated from Invivo5.

Table 2 shows comparisons of landmark detection between the three groups. When comparing the measured values in 2D cephalograms and regenerated 2D cephalograms from CBCT images, none of the measurements were statistically significant. The maximum differences for the angular measurements were in the L1-MP angle, whereas for the linear measurements, the maximum difference was in the Upper Lip–E line. Between groups I and II, the greatest difference was in the Upper Lip–E line, and the least difference was in the U1-NPog and L1-NPog lines. Between groups I and III, the greatest difference was in the Upper Lip–E line, and the least was in the SN-MP angle. Likewise, between groups II and III, the greatest difference was in the SNB angle, and the least was in L1-NPog.

Conventional CBCT should not be used in general orthodontic practice because of the higher radiation dose. Lateral cephalograms, anteroposterior cephalograms, panoramic radiographs, and temporomandibular joint radiographs, which are usually required to formulate the orthodontic treatment plan, can be generated by our system from a single CBCT session. Although the radiation dose associated with a single CBCT session is higher than that of all the four radiographs combined, as CBCT needs to be performed only once, it can decrease patient discomfort and provide accurate 3D images 12 . Before the widespread use of CBCT, clinicians often missed critical factors such as temporomandibular joint problems, severely impacted tooth, and airway problems 13 , 14 . As dental clinics now widely use CBCT, many orthodontists and maxillofacial surgeons order CBCT to assess the temporomandibular joint area, position of the third molar, and maxillary sinus for accurate diagnosis. Particularly, if the patient requires orthognathic surgery, CBCT is essential for identifying the position of the anatomic structure. In addition, this study included patients who required both a cephalometric radiograph and CBCT. Therefore, an additional cephalometric radiograph was required when orthodontic analysis was performed in patients who underwent CBCT. If adequate data are obtained with CBCT for orthodontic analysis, additional cephalometric radiographs would not be necessary.

Several studies have been conducted on CBCT use in cephalometry, but most of these studies were on the effectiveness of 3D cephalometric analysis. Three-dimensional cephalometric analysis can be performed using various 3D cephalometric analysis programs. However, most clinicians are familiar with 2D cephalometric analysis, which are mainly based on 2D data. In addition, 3D cephalometric analysis is more complex than 2D cephalometric analysis. Therefore, most clinicians prefer to use 2D cephalometric analysis.

A significant novelty of our study is that it focused on the effectiveness of a CBCT-generated 2D cephalogram, for which there are only a limited number of studies.

The null hypothesis was not rejected based on the statistical analysis results. The lack of significant differences among all evaluated measurements indicates that lateral cephalograms generated from CBCTs are similar to conventional lateral cephalograms. These findings add to the argument that CBCT alone can be used for diagnosis in orthodontics. If there were statistically significant differences, one conclusion would be that lateral cephalograms generated from CBCTs would still be insufficient for clinical use because in this study, the CBCT-generated cephalograms were evaluated on the premise that conventional cephalograms analyzed with AI were correct. For this premise, entire reference points were not compared, and only reliable points were selected and used.

Lateral cephalograms are indispensable for the examination of the relationship between soft tissues, dental tissues, and skeletal structures as well as the diagnosis of anteroposterior and vertical variation in these structures 15 . Therefore, the procedure for cephalometric analysis must be precise, safe, and repeatable. We considered manual cephalometry performed by a single examiner. However, as the 2D cephalogram and CBCT-generated cephalogram images were markedly different, it was impossible to perform a blind test. Thus, AI-based cephalometric analysis may be more objective than manual cephalometry.

Since the development of the first automatic measurement point identification method by Cohen and Linney et al. 16 in 1984, various studies to improve the automatic measurement point recognition accuracy have been reported, and most of the measurement points showed a high correlation with the measurement results of the examiner. The development of AI has significantly influenced image analysis, particularly medical image analysis 17 . Several algorithms have been developed to automatically recognize these anatomical indicators using various AI models, and dentistry is no exception. These algorithms enable inexperienced clinicians to consistently detect landmark points and analyze them. The AI of WebCeph uses a deep learning algorithm. The deep learning algorithm uses a convolutional filter and pooling layer to extract features from an image and analyze the patterns. Filter sizes, regions, categorization, combinations, and so on have been used to enhance and develop many deep learning models. Since they leverage spatially on local correlation by enforcing local connection patterns, convolution neural networks are particularly well-suited for image processing and recognition applications. Consequently, it is expected that when the diagnostic image data are evaluated using deep learning, the empirical knowledge gained from examining the image data would be better reflected. Clinicians appreciate time-saving and convenience of use as two of the many benefits of digital cephalometry. Measurement reproducibility is required to determine the accuracy of any method of analysis. The use of computers in treatment planning is predicted to eliminate the incidence of errors caused by fatigued operators and offer a uniform, quick, and effective evaluation with a high rate of repeatability 18 . According to recent studies, AI can identify landmarks as accurately as human examiners, and it might be a viable choice for repeated recognition of numerous cephalometric landmarks 19 . The successful detection rates of 19 skeletal landmarks with a 2-mm range 20 , 21 , 22 , which has usually been acknowledged as a clinical error range in AI performance 23 , have traditionally been used to compare the performance of an automated identification system.

In general, differences between the three groups were not statistically significant in any of the assessments. Differences in the linear measurements were larger than changes in the angular measurements, possibly because of image distortion or calibration. This is in line with the findings of a previous study 15 . According to the study by Chen et al. 18 , the menton, gonion, lower incisor apex, orbitale, and porion are the most questionable and unpredictable points irrespective of the method utilized for locating the landmarks. In addition, Lagravere et al. 24 reported that the menton, nasion, and posterior nasal spine also result in errors. Hwang et al. 25 described that the orbitale and PNS displayed higher standard deviation values when detected by AI because such landmarks are difficult to recognize owing to overlapping cranial base structures.

The accuracy depends on the size of the training dataset, which increases as the number of landmarks to be detected increases 26 . Most studies evaluating these software programs used lateral cephalometric radiographs obtained from a small number of cephalography equipment 27 . In the present study, WebCeph was used to evaluate 2D conventional lateral cephalograms obtained from specific devices, not lateral cephalograms generated from CBCT. WebCeph could not identify the landmark around the mandibular symphysis area, which could have affected the results of this study. This may be because the anterior region of the mandible of each patient was stabilized by a chin holder when CBCT was acquired. Apart from the inability to locate a specific point, landmarks may not be identified in more cases using WebCeph depending on what kind of radiographic device was used. In some cases, WebCeph deviates completely from a specific location, especially in the mandibular symphysis area. The widest possible failure category in training the software is the mandibular symphysis.

In summary, most inconsistencies were observed in linear measurements. The findings of the present study showed that tracing with the automatic WebCeph resulted in obvious inaccuracies, such as in landmark detection, where points were frequently identified outside the bone or at the wrong location; in soft tissue outline tracing, where the tracing line was clearly drawn away from the soft tissue outline; and in the detection of the average of bilateral points. These are all significant concerns that can have a direct influence on the analysis outcome and were identified in all the radiographs examined in this study.

Another factor that influenced the results of this study was the use of 2D radiographs generated from CBCT for analysis. Compared with conventional cephalograms, during image acquisition, the errors due to faulty positioning of the patients could be modified in CBCT datasets by repeated correction and reassessment. The inherent 3D properties of the CBCT dataset enable the generation of an endless number of reformatted images and orthogonal cephalograms 28 , 29 . Moreover, it is possible to represent both sides of the skull, preventing the superimposition of the left and right structures. However, the development of 2D skull landmarks and virtual 3D CBCT models remains an obstacle 30 . Owing to the characteristics of the 3D anatomical structures, landmarks are often missing in 2D. In 3D images, the acute edges observed in conventional lateral cephalograms are replaced by surfaces and curves. For example, the ear rods used in conventional cephalograms point to the location of the external auditory meatus; however, the anatomic porion differs from the external opening. According to van Vlijmen et al. 31 , the sella, upper incisor apex, incision inferius, and lower incisal apex are difficult to recognize using the 3D model. Since all these points are contained within the 3D model, CBCT slices should be selected to be able to designate their position 32 . However, several in-vitro and in-vivo investigations found no difference 29 , 30 , 31 .

Since the landmark measurements obtained from the CBCT-generated images are comparable with the virtual distances and angles between skull locations and the measurements made on conventional lateral cephalograms, the need for further conventional lateral cephalograms can be reduced, thereby avoiding additional radiation exposure to the patient. In addition, AI in CBCT analysis will be a beneficial addition and should be studied further in future research 33 . We are in the process of developing low-dose CBCT using various algorithms to reduce exposure rate. We plan to evaluate the significance of AI-based low-dose CBCT acquisition in orthodontic diagnosis. The present study could act as a starting point for a CBCT-based orthodontic diagnosis system.

Cephalograms generated from CBCT should be used by recognizing their limitations and considering the advantages in terms of radiation dose, convenience, and cost. In addition, through the development of AI and integration of CBCT, it can be expected that orthodontic diagnosis and treatment planning will be easier in the future.

Study participants

CBCT and lateral cephalograms were recorded for patients who visited the Department of Dentistry, Hallym University Sacred Heart Hospital. The study participants comprised 15 male and 15 female patients, with a mean age of 16.57 years and an age range of 7–41 years. Informed consent was obtained from all subjects involved in the study. Specific consent was obtained to publish the images of participants in an online open-access publication. Written informed consent has been obtained from a patient and/or legal guardian for minors to publish this paper. The inclusion criteria were as follows:

A.Patients with systemic diseases that were medically well-controlled

B.Patients who required cephalometric radiograph and CBCT

C.Patients without any maxillofacial deformity

D.Patients with erupted incisors and first molars

Patients for whom radiographs could not be recorded were excluded from the study. The radiographs were categorized into three groups. Group I included conventional lateral cephalograms, group II included cephalometric radiographs generated from CBCTs using OnDemand 3D (Cybermed Co., Seoul, Korea), and group III included cephalometric radiographs generated from CBCTs using Invivo5 (KaVo Co., Biberach, Germany). The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of Hallym University Sacred Heart Hospital (IRB Approval No. 2021-07-016-005). In the entire research, the personal information of the patients was not disclosed.

CBCT Protocol

CBCT scans were recorded using Alphard Vega (Asahi Roentgen Inc., Kyoto, Japan), with a slice thickness of 0.39 mm, 3 voxel size level, 20.0 × 17.9 cm exposure area, 4 mA, 80 kV, and 17 s exposure time. The collected data were imported into OnDemand 3D and Invivo5 as DICOM files.

Lateral cephalometric radiograph protocol

Lateral cephalograms were obtained using Rayscan Alpha (Ray Co., Gyeonggi-do, Korea). During imaging, both ear rods of the head restraint were inserted into the participant’s ear hole, and the head was fixed. During imaging, the tube current was 4–17 mA, tube voltage was 60–90 kV, and exposure time was 3.8–9.9 s (group I). The radiographs were saved as JPEG files for easy comparison with groups II and III.

Generation of lateral cephalometric radiographs from CBCT data

After recording the CBCT, the stored DICOM file was reconstructed into 2D lateral cephalograms using the X-ray generation module in the OnDemand 3D and Invivo5 programs. The midsagittal plane of the patients was aligned vertically using the axial view, the transporionic line was positioned horizontally using the coronal view, and the Frankfort plane was oriented horizontally using the sagittal perspective. The reconstructed images were saved as JPEG files .

Landmark identification

Conventional lateral cephalometric radiographs and lateral cephalometric radiographs generated from the CBCT data were automatically measured using Webceph. Figure  2 shows the automatic tracing of the measurement points in WebCeph using AI. Each analyzed image was saved individually. On each lateral cephalometric radiograph, 17 measurement points were indicated, and 11 measurements representing the skeletal, dental, and soft-tissue characteristics were evaluated, including six angular and five linear measurements. The bilateral structures were averaged to create a single measurement point. The measurements used in the study were as follows: SNA (°), SNB (°), ANB (°), SN-MP (°), U1-MaxP (°), L1-MP (°), N-Me (mm), U1-NPog (mm), L1-NPog (mm), upper lip-E line (mm), and lower lip-E line (mm).

Major landmarks used in this study with AI-based tracing program.

Statistical analysis

Statistical analysis was performed using Statistical Package for Social Sciences for Windows (version 25.0; SPSS Inc., Chicago, Illinois, USA). The data are presented as the mean, standard error, standard deviation, and significance values. To compare the differences in measured values between the three groups, a one-way ANOVA and Tukey’s post hoc test were used. The significance level was set at p  < 0.05, and the results of the study group were estimated with a 95% confidence interval.

Data availability

The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request.

Broadbent, B. H. A new X-ray technique and its application to orthodontia. Angle Orthod. 1 , 45–66. https://doi.org/10.1043/0003-3219(1931)001%3c0045:ANXTAI%3e2.0.CO;2 (1931).

Article   Google Scholar  

Albarakati, S. F., Kula, K. S. & Ghoneima, A. A. The reliability and reproducibility of cephalometric measurements: A comparison of conventional and digital methods. Dentomaxillofac. Radiol. 41 , 11–17. https://doi.org/10.1259/dmfr/37010910 (2012).

Article   CAS   PubMed   PubMed Central   Google Scholar  

Nijkamp, P. G., Habets, L. L., Aartman, I. H. & Zentner, A. The influence of cephalometrics on orthodontic treatment planning. Eur. J. Orthod. 30 , 630–635. https://doi.org/10.1093/ejo/cjn059 (2008).

Article   PubMed   Google Scholar  

Pauwels, R. et al. Effective dose range for dental cone beam computed tomography scanners. Eur. J. Radiol. 81 , 267–271. https://doi.org/10.1016/j.ejrad.2010.11.028 (2012).

De Grauwe, A. et al. CBCT in orthodontics: A systematic review on justification of CBCT in a paediatric population prior to orthodontic treatment. Eur. J. Orthod. 41 , 381–389. https://doi.org/10.1093/ejo/cjy066 (2019).

Berkhout, W. E. The ALARA-principle. Backgrounds and enforcement in dental practices. Ned. Tijdschr. Tandheelkd. 122 , 263–270. https://doi.org/10.5177/ntvt.2015.5.14227 (2015).

Article   CAS   PubMed   Google Scholar  

Frush, D. P. Justification and optimization of CT in children: How are we performing?. Pediatr. Radiol. 41 (Suppl 2), 467–471. https://doi.org/10.1007/s00247-011-2097-0 (2011).

Kanno, H. & Fujii, H. Clinical significance and interpretation of red cell enzyme analyses. Rinsho Byori 41 (116), 139–147 (2001).

Google Scholar  

da Silveira, H. L. & Silveira, H. E. Reproducibility of cephalometric measurements made by three radiology clinics. Angle Orthod. 76 , 394–399. https://doi.org/10.1043/0003-3219(2006)076[0394:ROCMMB]2.0.CO;2 (2006).

Tsorovas, G. & Karsten, A. L. A comparison of hand-tracing and cephalometric analysis computer programs with and without advanced features–accuracy and time demands. Eur. J. Orthod. 32 , 721–728. https://doi.org/10.1093/ejo/cjq009 (2010).

Baumrind, S. & Miller, D. M. Computer-aided head film analysis: The University of California San Francisco method. Am. J. Orthod. 78 , 41–65. https://doi.org/10.1016/0002-9416(80)90039-1 (1980).

Choi, J. Y. et al. Effectiveness of 2D radiographs in detecting CBCT-based incidental findings in orthodontic patients. Sci. Rep. 11 , 9280. https://doi.org/10.1038/s41598-021-88795-3 (2021).

Article   ADS   CAS   PubMed   PubMed Central   Google Scholar  

Kuhlberg, A. J. & Norton, L. A. Pathologic findings in orthodontic radiographic images. Am. J. Orthod. Dentofacial Orthop. 123 , 182–184. https://doi.org/10.1067/mod.2003.4 (2003).

Bondemark, L., Jeppsson, M., Lindh-Ingildsen, L. & Rangne, K. Incidental findings of pathology and abnormality in pretreatment orthodontic panoramic radiographs. Angle Orthod. 76 , 98–102. https://doi.org/10.1043/0003-3219(2006)076[0098:IFOPAA]2.0.CO;2 (2006).

Aksakalli, S., Yilanci, H., Gorukmez, E. & Ramoglu, S. I. Reliability assessment of orthodontic apps for cephalometrics. Turk. J. Orthod. 29 , 98–102. https://doi.org/10.5152/TurkJOrthod.2016.1618 (2016).

Article   PubMed   PubMed Central   Google Scholar  

Cohen, A. M., Ip, H. H. & Linney, A. D. A preliminary study of computer recognition and identification of skeletal landmarks as a new method of cephalometric analysis. Br. J. Orthod. 11 , 143–154. https://doi.org/10.1179/bjo.11.3.143 (1984).

Kim, Y. H. et al. Influence of the depth of the convolutional neural networks on an artificial intelligence model for diagnosis of orthognathic surgery. J. Pers. Med. https://doi.org/10.3390/jpm11050356 (2021).

Celik, E., Polat-Ozsoy, O. & Toygar Memikoglu, T. U. Comparison of cephalometric measurements with digital versus conventional cephalometric analysis. Eur. J. Orthod. 31 , 241–246. https://doi.org/10.1093/ejo/cjn105 (2009).

Kim, H. et al. Web-based fully automated cephalometric analysis by deep learning. Comput. Methods Programs Biomed. 194 , 105513. https://doi.org/10.1016/j.cmpb.2020.105513 (2020).

Arik, S. O., Ibragimov, B. & Xing, L. Fully automated quantitative cephalometry using convolutional neural networks. J. Med. Imaging (Bellingham) 4 , 014501. https://doi.org/10.1117/1.JMI.4.1.014501 (2017).

Nishimoto, S., Sotsuka, Y., Kawai, K., Ishise, H. & Kakibuchi, M. Personal computer-based cephalometric landmark detection with deep learning, using cephalograms on the internet. J. Craniofac. Surg. 30 , 91–95. https://doi.org/10.1097/SCS.0000000000004901 (2019).

Park, J. H. et al. Automated identification of cephalometric landmarks: Part 1-comparisons between the latest deep-learning methods YOLOV3 and SSD. Angle Orthod. 89 , 903–909. https://doi.org/10.2319/022019-127.1 (2019).

Hwang, H. W. et al. Automated identification of cephalometric landmarks: Part 2- might it be better than human?. Angle Orthod. 90 , 69–76. https://doi.org/10.2319/022019-129.1 (2020).

Lagravere, M. O. et al. Intraexaminer and interexaminer reliabilities of landmark identification on digitized lateral cephalograms and formatted 3-dimensional cone-beam computerized tomography images. Am. J. Orthod. Dentofacial Orthop. 137 , 598–604. https://doi.org/10.1016/j.ajodo.2008.07.018 (2010).

Hwang, H. W., Moon, J. H., Kim, M. G., Donatelli, R. E. & Lee, S. J. Evaluation of automated cephalometric analysis based on the latest deep learning method. Angle Orthod. 91 , 329–335. https://doi.org/10.2319/021220-100.1 (2021).

Moon, J. H. et al. How much deep learning is enough for automatic identification to be reliable?. Angle Orthod. 90 , 823–830. https://doi.org/10.2319/021920-116.1 (2020).

Kim, J. et al. Accuracy of automated identification of lateral cephalometric landmarks using cascade convolutional neural networks on lateral cephalograms from nationwide multi-centres. Orthod. Craniofac. Res. https://doi.org/10.1111/ocr.12493 (2021).

Cevidanes, L. H., Styner, M. A. & Proffit, W. R. Image analysis and superimposition of 3-dimensional cone-beam computed tomography models. Am. J. Orthod. Dentofacial Orthop. 129 , 611–618. https://doi.org/10.1016/j.ajodo.2005.12.008 (2006).

Cattaneo, P. M., Bloch, C. B., Calmar, D., Hjortshoj, M. & Melsen, B. Comparison between conventional and cone-beam computed tomography-generated cephalograms. Am. J. Orthod. Dentofacial Orthop. 134 , 798–802. https://doi.org/10.1016/j.ajodo.2008.07.008 (2008).

Kumar, V., Ludlow, J. B., Mol, A. & Cevidanes, L. Comparison of conventional and cone beam CT synthesized cephalograms. Dentomaxillofac. Radiol. 36 , 263–269. https://doi.org/10.1259/dmfr/98032356 (2007).

Kumar, V., Ludlow, J., Soares Cevidanes, L. H. & Mol, A. In vivo comparison of conventional and cone beam CT synthesized cephalograms. Angle Orthod. 78 , 873–879. https://doi.org/10.2319/082907-399.1 (2008).

van Vlijmen, O. J. et al. A comparison between 2D and 3D cephalometry on CBCT scans of human skulls. Int. J. Oral Maxillofac. Surg. 39 , 156–160. https://doi.org/10.1016/j.ijom.2009.11.017 (2010).

Yu, H. J. et al. Automated skeletal classification with lateral cephalometry based on artificial intelligence. J. Dent. Res. 99 , 249–256. https://doi.org/10.1177/0022034520901715 (2020).

Download references

Acknowledgements

E.-J.C. and B.-E.Y. have contributed equally to this work.

This research was supported by a grant from the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (Grant Number: HI20C2114). This work was supported by the Nano-Convergence Foundation ( www.nanotech2020.org ) funded by the Ministry of Science and ICT (MSIT, Korea) & the Ministry of Trade, Industry and Energy (MOTIE, Korea). [Project Name: Dental implant placement guide robot system based on permanent magnet positioning device/Project Number: 20014921]. This work was supported by the Korea Medical Device Development Fund grant funded by the Korea government (the Ministry of Science and ICT, the Ministry of Trade, Industry and Energy, the Ministry of Health & Welfare, Republic of Korea, and the Ministry of Food and Drug Safety) (Project Number: KMDF_PR_20200901-0237, 1711138501). This work was supported by the Korea Medical Device Development Fund grant funded by the Korea government (the Ministry of Science and ICT, the Ministry of Trade, Industry and Energy, the Ministry of Health & Welfare, the Ministry of Food and Drug Safety) (Project Number: KMDF001409351G0003232).

Author information

These authors contributed equally: Eun-Ji Chung and Byoung-Eun Yang.

Authors and Affiliations

Department of Conservative Dentistry, Hallym University Sacred Heart Hospital, Anyang, 14068, Korea

Eun-Ji Chung & Sam-Hee Kang

Department of Oral and Maxillofacial Surgery, Hallym University Sacred Heart Hospital, Anyang, 14068, Korea

Byoung-Eun Yang, Sangmin Yi & Soo-Hwan Byun

Department of Orthodontics, Hallym University Sacred Heart Hospital, Anyang, 14068, Korea

In-Young Park

Department of Oral and Maxillofacial Radiology, Hallym University Sacred Heart Hospital, Anyang, 14068, Korea

Young-Hee Kim

Graduate School of Clinical Dentistry, Hallym University, Chuncheon, 24252, Republic of Korea

Eun-Ji Chung, Byoung-Eun Yang, In-Young Park, Sangmin Yi, Sung-Woon On, Young-Hee Kim, Sam-Hee Kang & Soo-Hwan Byun

Institute of Clinical Dentistry, Hallym University, Chuncheon, 24252, Republic of Korea

Byoung-Eun Yang, In-Young Park, Sangmin Yi, Sung-Woon On, Young-Hee Kim, Sam-Hee Kang & Soo-Hwan Byun

Dental Implant Robotic Center, Hallym University Sacred Heart Hospital, Anyang, 14068, Korea

Byoung-Eun Yang & Soo-Hwan Byun

Department of Oral and Maxillofacial Surgery, Hallym University Dongtan Sacred Heart Hospital, Hwaseong, 18450, Korea

Sung-Woon On

You can also search for this author in PubMed   Google Scholar

Contributions

S.-H.B. and B.-E.Y. conceptualized the study. E.-J.C. performed data curation. S.-H.B. and B.-E.Y. funding acquisition. E.-J.C. were responsible for investigation. S.Y. and Y.-H.K. provided resources for the study. I.-Y.P., S.Y., Y.-H.K. and S.-H.K. supervised this study. B.-E.Y., S.Y. and S.-H.B. performed data visualization. E.-J.C. and S.-H.B. wrote the original draft. I.-Y.P., S.-H.K. and S.-W.O. contributed to the review and editing of the manuscript. All authors read and approved the final version of the manuscript.

Corresponding author

Correspondence to Soo-Hwan Byun .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Chung, EJ., Yang, BE., Park, IY. et al. Effectiveness of cone-beam computed tomography-generated cephalograms using artificial intelligence cephalometric analysis. Sci Rep 12 , 20585 (2022). https://doi.org/10.1038/s41598-022-25215-0

Download citation

Received : 07 March 2022

Accepted : 28 November 2022

Published : 29 November 2022

DOI : https://doi.org/10.1038/s41598-022-25215-0

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

This article is cited by

An artificial intelligence study: automatic description of anatomic landmarks on panoramic radiographs in the pediatric population.

• Elif Bilgir

BMC Oral Health (2023)

By submitting a comment you agree to abide by our Terms and Community Guidelines . If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Advertisement

On imaging modalities for cephalometric analysis: a review

• Published: 21 March 2023
• Volume 82 , pages 36837–36858, ( 2023 )

Cite this article

• Abhishek Gupta   ORCID: orcid.org/0000-0002-8592-9964 1  

276 Accesses

3 Citations

1 Altmetric

Explore all metrics

Cephalometrics is an integral part of orthodontic diagnosis and treatment planning. It has been extensively used to study variation in human face and craniofacial growth. Cephalometrics is an established and valuable tool to assess outcome of orthodontic and orthognathic surgical procedures, follow up and relapse. Cephalometric has also been used a research instrument for huge number of investigations. Cephalometric measurement techniques has progressed over the years from a manual tracing of analog X-Ray film over acetate tracing sheets to the modern practice of on-screen computerized cephalometric analysis on a digital two-dimensional (2-D) image. Cephalometric analysis can also be performed on-screen on image derived from CT scans or CBCT. Each imaging modality is associated with its own quality features of the X-Ray image, and radiation protocol. The objective of this review was to critically analyze diagnostic limitations associated with three types of imaging modality being used for cephalometric analyses. These limitations can vary in terms of accuracy, repeatability, reproducibility, reliability, feasibility of craniofacial landmark localization and radiation exposure to patient.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price includes VAT (Russian Federation)

Instant access to the full article PDF.

Rent this article via DeepDyve

Institutional subscriptions

Similar content being viewed by others

Cephalometric Radiography

In vivo comparison of MRI- and CBCT-based 3D cephalometric analysis: beginning of a non-ionizing diagnostic era in craniomaxillofacial imaging?

Alexander Juerchott, Christian Freudlsperger, … Tim Hilgenfeld

Comparison of a tridimensional cephalometric analysis performed on 3T-MRI compared with CBCT: a pilot study in adults

Cinzia Maspero, Andrea Abate, … Marco Farronato

Adams GL, Gansky SA, Miller AJ, Harrell WE Jr, Hatcher DC (2004) Comparison between traditional 2-dimensional cephalometry and a 3-dimensional approach on human dry skulls. Am J Orthod Dentofac Orthop 126:397–409

Google Scholar  

Ahlqvist J, Eliasson S, Welander U (1986) The effect of projection errors on cephalometric length measurements. Eur J Orthod 8:141–148

Available (n.d.) http://www.instrumentariumdental.com

Baumrind S, Frantz RC (1971) The reliability of head film measurements. 1. Landmark identification. Am J Orthod 60:111–127

Baumrind S, Miller DM (1980) Computer-aided head film analysis: the University of California san Francisco method. Am J Orthod 78:41–65

Bayome M, Park JH, Kook YA (2013) New three-dimensional cephalometric analyses among adults with a skeletal class I pattern and normal occlusion. Korean J Orthod 43:62–73

Berco M, Rigali PH Jr, Miner RM, DeLuca S, Anderson NK, Will LA (2009) Accuracy and reliability of linear cephalometric measurements from cone-beam computed tomography scans of a dry human skull. Am J Orthod Dentofac Orthop 136:17–18

Bholsithi W, Tharanon W, Chintakanon K, Komolpis R, Sinthanayothin C (2009) 3D vs. 2D cephalometric analysis comparisons with repeated measurements from 20 Thai males and 20 Thai females. Biomed Imaging Interv J 5:1

Brown AA, Scarfe WC, Scheetz JP, Silveira AM, Farman AG (2009) Linear accuracy of cone beam CT derived 3D images. Angle Orthod 79:150–157

Cattaneo PM, Bloch CB, Calmar D, Hjortshoj M, Melsen B (2008) Comparison between conventional and cone-beam computed tomography-generated cephalograms. Am J Orthod Dentofac Orthop 134:798–802

Cevidanes L, Oliveira AE, Motta A, Phillips C, Burke B, Tyndall D (2009) Head orientation in CBCT-generated cephalograms. Angle Orthod 79:971–977

Cevidanes LH, Oliveira AE, Grauer D, Styner M, Proffit WR (2011) Clinical application of 3D imaging for assessment of treatment outcomes. Semin Orthod 17:72–80

Chau ACM, Wong RWK, Hagg U (2010) 3-dimensional cone-beam computerized tomographic cephalometric database on jaw dimensions in Chinese. Open Anthropol 3:2–7

Cheung LK, Chan YM, Jayaratne YS, Lo J (2011) Three-dimensional cephalometric norms of Chinese adults in Hong Kong with balanced facial profile. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 112:12

Chidiac JJ, Shofer FS, Al-Kutoub A, Laster LL, Ghafari J (2002) Comparison of CT scanograms and cephalometric radiographs in craniofacial imaging. Orthod Craniofacial Res 5:104–113

Chien P, Parks E, Eraso F, Hartsfield J, Roberts W, Ofner S (2009) Comparison of reliability in anatomical landmark identification using two-dimensional digital cephalometrics and three-dimensional cone beam computed tomography in vivo. Dentomaxillofacial Radiol 38:262–273

Couceiro CP, Vilella ODV (2010) Imagens em 2D e 3D geradas pela TC Cone-Beam e radiografias convencionais: qual a mais confiável? Dental Press J Orthod 15:40–41

Damstra J, Fourie Z, Huddleston Slater JJ, Ren Y (2011) Reliability and the smallest detectable difference of measurements on 3-dimensional cone-beam computed tomography images. Am J Orthod Dentofac Orthop 140:e107–e114

de Oliveira AE, Cevidanes LH, Phillips C, Motta A, Burke B, Tyndall D (2009) Observer reliability of three-dimensional cephalometric landmark identification on cone-beam computerized tomography. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 107:256–265

Dibbets JM, Nolte K (2002) Effect of magnification on lateral cephalometric studies. Am J Orthod Dentofac Orthop 122:196–201

Dot G, Schouman T, Chang S, Rafflenbeul F, Kerbrat A, Rouch P, et al. (2022) "Automatic Three-Dimensional Cephalometric Landmarking via Deep Learning," medRxiv, p. 2022.01.28.22269989

Downs WB (1948) Variations in facial relationships; their significance in treatment and prognosis. Am J Orthod 34(10):812–840. https://doi.org/10.1016/0002-9416(48)90015-3

Farman AG, Scarfe WC (2006) Development of imaging selection criteria and procedures should precede cephalometric assessment with cone-beam computed tomography. Am J Orthod Dentofac Orthop 130:257–265

Farronato G, Perillo L, Bellincioni F, Briguglio F, Farronato D, Dominici AD (2012) Direct 3D cephalometric analysis performed on CBCT. J Inf Technol Software Eng 2:107–1–107-3

Fuyamada M, Nawa H, Shibata M, Yoshida K, Kise Y, Katsumata A et al (2011) Reproducibility of landmark identification in the jaw and teeth on 3-dimensional cone-beam computed tomography images. Angle Orthod 81:843–849

Grauer D, Cevidanes LS, Styner MA, Heulfe I, Harmon ET, Zhu H et al (2010) Accuracy and landmark error calculation using cone-beam computed tomography-generated cephalograms. Angle Orthod 80:286–294

Gray CF (2010) Practice-based cone-beam computed tomography: a review. Prim Dent Care 17:161–167

Greiner M, Greiner A, Hirschfelder U (2007) Variance of landmarks in digital evaluations: comparison between CT-based and conventional digital lateral cephalometric radiographs. J Orofac Orthop 68:290–298

Gribel BF, Gribel MN, Manzi FR, Brooks SL, McNamara JA Jr (2011) From 2D to 3D: an algorithm to derive normal values for 3-dimensional computerized assessment. Angle Orthod 81(1):3–10. https://doi.org/10.2319/032910-173.1

Gupta A (2019) "Current research opportunities of image processing and computer vision," Comput Sci; Vol 20, No 4, https://doi.org/10.7494/csci.2019.20.4.3163

Gupta A (2020) Challenges for computer aided diagnostics using X-ray and Tomographic reconstruction images in craniofacial applications. Int J Comput Vis Robot 10:360–371

Gupta A (2022) "RegCal: registration-based calibration method to perform linear measurements on PA (posteroanterior) cephalogram- a pilot study," Multimed Tools Appl, 2022/07/12

Gupta A (2022) "A lightweight deep neural network implemented on MATLAB without using GPU for the automatic monitoring of the plants," Multimed Tools Appl

Gupta A, Kharbanda O, Sardana V, Balachandran R, Sardana H (2015) A knowledge-based algorithm for automatic detection of cephalometric landmarks on CBCT images. Int J Comput Assist Radiol Surg 10(11):1737–1752

Gupta A, Kharbanda OP, Sardana V, Balachandran R, Sardana HK (2016) Accuracy of 3D cephalometric measurements based on an automatic knowledge-based landmark detection algorithm. Int J Comput Assist Radiol Surg 11:1297–1309

Gupta A, Kharbanda OP, Balachandran R, Sardana V, Kalra S, Chaurasia S et al (2017) Precision of manual landmark identification between as-received and oriented volume-rendered cone-beam computed tomography images. Am J Orthod Dentofac Orthoped 151:118–131

Gupta A, Sardana HK, Kharbanda OP, Sardana V (2019) "Method for automatic detection of anatomical landmarks in volumetric data," US Patent US10318839B2, 11–06-2019

Halazonetis DJ (2005) From 2-dimensional cephalograms to 3-dimensional computed tomography scans. Am J Orthod Dentofac Orthop 127:627–637

Hanzelka T, Foltan R, Horka E, Sedy J (2010) Reduction of the negative influence of patient motion on quality of CBCT scan. Med Hypotheses 75:610–612

Hassan B, van der Stelt P, Sanderink G (2009) Accuracy of three-dimensional measurements obtained from cone beam computed tomography surface-rendered images for cephalometric analysis: influence of patient scanning position. Eur J Orthod 31:129–134

Hassan B, Nijkamp P, Verheij H, Tairie J, Vink C, van der Stelt P et al (2013) Precision of identifying cephalometric landmarks with cone beam computed tomography in vivo. Eur J Orthod 35:38–44

Holberg C, Steinhauser S, Geis P, Rudzki-Janson I (2005) Cone-beam computed tomography in orthodontics: benefits and limitations. J Orofac Orthop 66:434–444

Horner K, Islam M, Flygare L, Tsiklakis K, Whaites E (2009) Basic principles for use of dental cone beam computed tomography: consensus guidelines of the European academy of dental and maxillofacial radiology. Dentomaxillofac Radiol 38:187–195

Ibragimov B, Likar B, Pernus F, Vrtovec T (2016) "Computerized Cephalometry by game theory with shape-and appearance-based landmark refinement,"

Kang SH, Jeon K, Kim H-J, Seo JK, Lee S-H (2020) Automatic three-dimensional cephalometric annotation system using three-dimensional convolutional neural networks: a developmental trial. Comput Methods Biomech Biomed Eng Imaging Visual 8:210–218

Kang SH, Jeon K, Kang S-H, Lee S-H (2021) 3D cephalometric landmark detection by multiple stage deep reinforcement learning. Sci Rep 11:17509

Kapila S, Conley RS, Harrell WE Jr (2011) The current status of cone beam computed tomography imaging in orthodontics. Dentomaxillofac Radiol 40:24–34

Katkar RA, Kummet C, Dawson D, Moreno Uribe L, Allareddy V, Finkelstein M, Ruprecht A (2013) Comparison of observer reliability of three-dimensional cephalometric landmark identification on subject images from Galileos and i-CAT cone beam CT. Dentomaxillofac Radiol 42:5

Kau CH, Richmond S, Palomo JM, Hans MG (2005) Current Products and Practice: Three-dimensional cone beam computerized tomography in orthodontics. J Orthod 32:282–293

Kim SG, Yi WJ, Hwang SJ, Choi SC, Lee SS, Heo MS, Huh KH, Kim TI, Hong H, Yoo JH (2012) Development of 3D statistical mandible models for cephalometric measurements. Imaging Sci Dent 42:175–182

Kim J-H, An S, Hwang D-M (2022) Reliability of cephalometric landmark identification on three-dimensional computed tomographic images. British J Oral Maxillofac Surg 60:320–321

Kragskov J, Bosch C, Gyldensted C, Sindet-Pedersen S (1997) Comparison of the reliability of craniofacial anatomic landmarks based on cephalometric radiographs and three-dimensional CT scans. Cleft Palate Craniofac J 34:111–116

Kumar V, Ludlow JB, Mol A, Cevidanes L (2007) Comparison of conventional and cone beam CT synthesized cephalograms. Dentomaxillofac Radiol 36:263–269

Kumar V, Ludlow J, Soares Cevidanes LH, Mol A (2008) In vivo comparison of conventional and cone beam CT synthesized cephalograms. Angle Orthod 78:873–879

Lagravere MO, Gordon JM, Guedes IH, Flores-Mir C, Carey JP, Heo G et al (2009) Reliability of traditional cephalometric landmarks as seen in three-dimensional analysis in maxillary expansion treatments. Angle Orthod 79:1047–1056

Lagravere MO, Low C, Flores-Mir C, Chung R, Carey JP, Heo G et al (2010) Intraexaminer and interexaminer reliabilities of landmark identification on digitized lateral cephalograms and formatted 3-dimensional cone-beam computerized tomography images. Am J Orthod Dentofac Orthop 137:598–604

Lamichane M, Anderson NK, Rigali PH, Seldin EB, Will LA (2009) Accuracy of reconstructed images from cone-beam computed tomography scans. Am J Orthod Dentofac Orthop 136:156–157

Lecomber AR, Yoneyama Y, Lovelock DJ, Hosoi T, Adams AM (2001) Comparison of patient dose from imaging protocols for dental implant planning using conventional radiography and computed tomography. Dentomaxillofac Radiol 30:255–259

Lee H, Park M, Kim J (2017) Cephalometric landmark detection in dental x-ray images using convolutional neural networks vol. 10134: SPIE

Lee SM, Kim HP, Jeon K, Lee S-H, Seo JK (2019) Automatic 3D cephalometric annotation system using shadowed 2D image-based machine learning. Phys Med Biol 64:055002

Lee J-H, Yu H-J, Kim M-j, Kim J-W, Choi J (2020) Automated cephalometric landmark detection with confidence regions using Bayesian convolutional neural networks. BMC Oral Health 20:270

Liedke GS, Delamare EL, Vizzotto MB, da Silveira HL, Prietsch JR, Dutra V et al (2012) Comparative study between conventional and cone beam CT-synthesized half and total skull cephalograms. Dentomaxillofac Radiol 41:136–142

Lindner C, Wang C-W, Huang C-T, Li C-H, Chang S-W, Cootes TF (2016) Fully Automatic System for Accurate Localisation and Analysis of Cephalometric Landmarks in Lateral Cephalograms. Sci Rep 6:33581

Lofthag-Hansen S (2009) Cone beam computed tomography radiation dose and image quality assessments. Swed Dent J Suppl 209:4–55

Lopes PM, Moreira CR, Perrella A, Antunes JL, Cavalcanti MG (2008) 3-D volume rendering maxillofacial analysis of angular measurements by multislice CT. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 105:224–230

Ludlow JB, Laster WS, See M, Bailey LJ, Hershey HG (2007) Accuracy of measurements of mandibular anatomy in cone beam computed tomography images. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 103:534–542

Ludlow JB, Gubler M, Cevidanes L, Mol A (2009) Precision of cephalometric landmark identification: cone-beam computed tomography vs conventional cephalometric views. Am J Orthod Dentofac Orthop 136:312–313

Ma Q, Kobayashi E, Fan B, Nakagawa K, Sakuma I, Masamune K et al (2020) Automatic 3D landmarking model using patch-based deep neural networks for CT image of oral and maxillofacial surgery. Int J Med Robot 16:e2093

Maken P, Gupta A (2022) "2D-to-3D: a review for computational 3D image reconstruction from X-ray images," Arch Comput Methods Eng, 2022/07/22

Martin CJ (2007) Effective dose: how should it be applied to medical exposures? Br J Radiol 80:639–647

Montúfar J, Romero M, Scougall-Vilchis RJ (2018) Automatic 3-dimensional cephalometric landmarking based on active shape models in related projections. Am J Orthod Dentofac Orthop 153:449–458

Montúfar J, Romero M, Scougall-Vilchis RJ (2018) Hybrid approach for automatic cephalometric landmark annotation on cone-beam computed tomography volumes. Am J Orthod Dentofac Orthoped 154:140–150

Moreira CR, Sales MA, Lopes PM, Cavalcanti MG (2009) Assessment of linear and angular measurements on three-dimensional cone-beam computed tomographic images. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 108:430–436

Moshiri M, Scarfe WC, Hilgers ML, Scheetz JP, Silveira AM, Farman AG (2007) Accuracy of linear measurements from imaging plate and lateral cephalometric images derived from cone-beam computed tomography. Am J Orthod Dentofac Orthop 132:550–560

Naji P, Alsufyani NA, Lagravère MO (2013) "Reliability of anatomic structures as landmarks in three-dimensional cephalometric analysis using CBCT," Angle Orthod

Nalcaci R, Ozturk F, Sokucu O (2010) A comparison of two-dimensional radiography and three-dimensional computed tomography in angular cephalometric measurements. Dentomaxillofac Radiol 39:100–106

Neelapu BC, Kharbanda OP, Sardana V, Gupta A, Vasamsetti S, Balachandran R, Rana SS, Sardana HK (2017) A pilot study for segmentation of pharyngeal and sino-nasal airway subregions by automatic contour initialization. Int J Comput Assist Radiol Surg 12:1877–1893

Neelapu BC, Sardana HK, Kharbanda OP, Sardana V, Gupta A, Vasamsetti S (2018) "Method And System For Automatic Volumetric-Segmentation Of Human Upper Respiratory Tract," US Patent US10699415B2

Neelapu BC, Kharbanda OP, Sardana V, Gupta A, Vasamsetti S, Balachandran R, Sardana HK (2018) Automatic localization of three-dimensional cephalometric landmarks on CBCT images by extracting symmetry features of the skull. Dentomaxillofacial Radiol 47:20170054

Neelapu BC, Kharbanda OP, Sardana HK, Gupta A, Vasamsetti S, Balachandran R et al The reliability of different methods of manual volumetric segmentation of pharyngeal and sinonasal subregions. Oral Surg Oral Med Oral Pathol Oral Radiol 124:577–587

Nervina JM (2012) Cone beam computed tomography use in orthodontics. Aust Dent J 1:95–102

Ngan DC, Kharbanda OP, Geenty JP, Darendeliler MA (2003) Comparison of radiation levels from computed tomography and conventional dental radiographs. Aust Orthod J 19:67–75

Nur M, Kayipmaz S, Bayram M, Celikoglu M, Kilkis D, Sezgin OS (2012) Conventional frontal radiographs compared with frontal radiographs obtained from cone beam computed tomography. Angle Orthod 82:579–584

Olmez H, Gorgulu S, Akin E, Bengi AO, Tekdemir I, Ors F (2011) Measurement accuracy of a computer-assisted three-dimensional analysis and a conventional two-dimensional method. Angle Orthod 81:375–382

Olszewski R, Cosnard G, Macq B, Mahy P, Reychler H (2006) 3D CT-based cephalometric analysis: 3D cephalometric theoretical concept and software. Neuroradiology 48:853–862

Olszewski R, Reychler H, Cosnard G, Denis JM, Vynckier S, Zech F (2008) Accuracy of three-dimensional (3D) craniofacial cephalometric landmarks on a low-dose 3D computed tomograph. Dentomaxillofac Radiol 37:261–267

Olszewski R, Tanesy O, Cosnard G, Zech F, Reychler H (2010) Reproducibility of osseous landmarks used for computed tomography based three-dimensional cephalometric analyses. J Craniomaxillofac Surg 38(3):214–221. https://doi.org/10.1016/j.jcms.2009.05.005

Olszewski R, Frison L, Wisniewski M, Denis JM, Vynckier S, Cosnard G, Zech F, Reychler H (2013) Reproducibility of three-dimensional cephalometric landmarks in cone-beam and low-dose computed tomography. Clin Oral Investig 17:285–292

Oz U, Orhan K, Abe N (2011) Comparison of linear and angular measurements using two-dimensional conventional methods and three-dimensional cone beam CT images reconstructed from a volumetric rendering program in vivo. Dentomaxillofac Radiol 40(8):492–500. https://doi.org/10.1259/dmfr/15644321

Park SH, Yu HS, Kim KD, Lee KJ, Baik HS (2006) A proposal for a new analysis of craniofacial morphology by 3-dimensional computed tomography. Am J Orthod Dentofac Orthop 129:e23–e34

Park CS, Park JK, Kim H, Han SS, Jeong HG, Park H (2012) Comparison of conventional lateral cephalograms with corresponding CBCT radiographs. Imaging Sci Dent 42(4):201–205. https://doi.org/10.5624/isd.2012.42.4.201

Pauwels R, Beinsberger J, Collaert B, Theodorakou C, Rogers J, Walker A, Cockmartin L, Bosmans H, Jacobs R, Bogaerts R, Horner K, SEDENTEXCT Project Consortium (2012) Effective dose range for dental cone beam computed tomography scanners. Eur J Radiol 81(2):267–271. https://doi.org/10.1016/j.ejrad.2010.11.028

Periago DR, Scarfe WC, Moshiri M, Scheetz JP, Silveira AM, Farman AG (2008) Linear accuracy and reliability of cone beam CT derived 3-dimensional images constructed using an orthodontic volumetric rendering program. Angle Orthod 78:387–395

Qian J, Cheng M, Tao Y, Lin J, Lin H (2019) "CephaNet: An Improved Faster R-CNN for Cephalometric Landmark Detection," in 2019 IEEE 16th International Symposium on Biomedical Imaging (ISBI 2019), pp. 868–871

Ramirez-Sotelo LR, Almeida S, Ambrosano GM, Boscolo F (2012) Validity and reproducibility of cephalometric measurements performed in full and hemifacial reconstructions derived from cone beam computed tomography. Angle Orthod 82:827–832

Scarfe WC, Farman AG, Sukovic P (2006) Clinical applications of cone-beam computed tomography in dental practice. J Can Dent Assoc 72:75–80

Schendel SA, Lane C (2009) 3D orthognathic surgery simulation using image fusion. Semin Orthod 15:48–56

Schlicher W, Nielsen I, Huang JC, Maki K, Hatcher DC, Miller AJ (2012) Consistency and precision of landmark identification in three-dimensional cone beam computed tomography scans. Eur J Orthod 34:263–275

Shahidi S, Bahrampour E, Soltanimehr E, Zamani A, Oshagh M, Moattari M et al (2014) The accuracy of a designed software for automated localization of craniofacial landmarks on CBCT images. BMC Med Imaging 14:1471–2342

Song Y, Qiao X, Iwamoto Y, Chen Y-w (2020) Automatic cephalometric landmark detection on X-ray images using a deep-learning method. Appl Sci 10:2547

Spin-Neto R, Mudrak J, Matzen LH, Christensen J, Gotfredsen E, Wenzel A (2013) Cone beam CT image artefacts related to head motion simulated by a robot skull: visual characteristics and impact on image quality. Dentomaxillofac Radiol 42:27

Spolyar JL (1987) Head Positioning Error in Cephalometric Radiography. Angle Orthod 57:77–88

Swennen GR, Schutyser F (2006) Three-dimensional cephalometry: spiral multi-slice vs cone-beam computed tomography. Am J Orthod Dentofac Orthop 130:410–416

Swennen GR, Schutyser F, Barth EL, De Groeve P, De Mey A (2006) A new method of 3-D cephalometry part I: the anatomic Cartesian 3-D reference system. J Craniofac Surg 17:314–325

Titiz I, Laubinger M, Keller T, Hertrich K, Hirschfelder U (2012) Repeatability and reproducibility of landmarks--a three-dimensional computed tomography study. Eur J Orthod 34:276–286

Tomasi C, Bressan E, Corazza B, Mazzoleni S, Stellini E, Lith A (2011) Reliability and reproducibility of linear mandible measurements with the use of a cone-beam computed tomography and two object inclinations. Dentomaxillofac Radiol 40:244–250

Torosdagli N, Liberton DK, Verma P, Sincan M, Lee JS, Bagci U (2019) Deep geodesic learning for segmentation and anatomical Landmarking. IEEE Trans Med Imaging 38:919–931

Trivedi M, Gupta A (2022) A lightweight deep learning architecture for the automatic detection of pneumonia using chest X-ray images. Multimed Tools Appl 81:5515–5536

Tulunoglu O, Esenlik E, Gulsen A, Tulunoglu I (2011) A comparison of three-dimensional and two-dimensional cephalometric evaluations of children with cleft lip and palate. Eur J Dent 5(4):451–458.

Tweed CH (1953) Evolutionary trends in orthodontics, past, present, and future. Am J Orthod 39:81–108

van Vlijmen OJ, Berge SJ, Swennen GR, Bronkhorst EM, Katsaros C, Kuijpers-Jagtman AM (2009) Comparison of cephalometric radiographs obtained from cone-beam computed tomography scans and conventional radiographs. J Oral Maxillofac Surg 67(1):92–97. https://doi.org/10.1016/j.joms.2008.04.025

Vandaele R, Aceto J, Muller M, Péronnet F, Debat V, Wang C-W et al (2018) Landmark detection in 2D bioimages for geometric morphometrics: a multi-resolution tree-based approach. Sci Rep 8:538

Wang S, Li H, Li J, Zhang Y, Zou B (2018) Automatic Analysis of Lateral Cephalograms Based on Multiresolution Decision Tree Regression Voting. J Healthcare Eng 2018:1797502

Williams FL, Richtsmeier JT (2003) Comparison of mandibular landmarks from computed tomography and 3D digitizer data. Clin Anat 16:494–500

Wu MC, Cheng KS, Chen YT, Liu JK, Ting WH (2010) Three-dimensional analysis of biplanar cephalograms. Eur J Orthod 32:627–632

Yitschaky O, Redlich M, Abed Y, Faerman M, Casap N, Hiller N (2010) Comparison of common hard tissue cephalometric measurements between computed tomography 3D reconstruction and conventional 2D cephalometric images. Angle Orthod 81:11–16

Yitschaky O, Redlich M, Abed Y, Faerman M, Casap N, Hiller N (2011) Comparison of common hard tissue cephalometric measurements between computed tomography 3D reconstruction and conventional 2D cephalometric images. Angle Orthod 81:11–16

Yoon YJ, Kim KS, Hwang MS, Kim HJ, Choi EH, Kim KW (2001) Effect of head rotation on lateral cephalometric radiographs. Angle Orthod 71:396–403

Yu S-H, Nahm D-S, Baek S-H (2008) Reliability of landmark identification on monitor-displayed lateral cephalometric images. Am J Orthod Dentofac Orthop 133:790.e1–790.e6

Yue W, Yin D, Li C, Wang G, Xu T (2006) Automated 2-D cephalometric analysis on X-ray images by a model-based approach. IEEE Trans Biomed Eng 53:1615–1623

Zamora N, Llamas JM, Cibrian R, Gandia JL, Paredes V (2012) A study on the reproducibility of cephalometric landmarks when undertaking a three-dimensional (3D) cephalometric analysis. Med Oral Patol Oral Cir Bucal 17(4):e678–e688. https://doi.org/10.4317/medoral.17721

Download references

Acknowledgements

Author would like to acknowledge Dr. H. K. Sardana and Dr. Viren Sardana from CSIR-CSIO, Chandigarh, and Dr. Rajiv Balachandran and Dr. O. P. Kharbanda from AIIMS-CDER, New Delhi for providing their insights to write a review.

Author information

Authors and affiliations.

Biomedical Application Division, CSIR-Central Scientific Instruments Organisation, Chandigarh, 160030, India

Abhishek Gupta

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Abhishek Gupta .

Ethics declarations

Conflict of interest.

The author declares that the submitted work is a part of the work submitted to the University for the Award of a degree.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Gupta, A. On imaging modalities for cephalometric analysis: a review. Multimed Tools Appl 82 , 36837–36858 (2023). https://doi.org/10.1007/s11042-023-14971-4

Download citation

Received : 28 March 2021

Revised : 19 February 2023

Accepted : 22 February 2023

Published : 21 March 2023

Issue Date : October 2023

DOI : https://doi.org/10.1007/s11042-023-14971-4

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Cephalograms
• Cephalometric analysis
• Computed tomography (CT)
• Cone beam computed tomography (CBCT)
• Imaging modality
• And computer-aided diagnosis
• Find a journal
• Publish with us
• Track your research

• Open access
• Published: 30 March 2022

Cephalometric measurements performed on CBCT and reconstructed lateral cephalograms: a cross-sectional study providing a quantitative approach of differences and bias

• Benedetta Baldini   ORCID: orcid.org/0000-0003-4365-4998 1   na1 ,
• Davide Cavagnetto   ORCID: orcid.org/0000-0001-8573-8327 2 , 3   na1 ,
• Giuseppe Baselli   ORCID: orcid.org/0000-0003-2978-1704 4 ,
• Chiarella Sforza   ORCID: orcid.org/0000-0001-6532-6464 5 &
• Gianluca Martino Tartaglia   ORCID: orcid.org/0000-0001-7062-5143 1 , 6   na1  

BMC Oral Health volume  22 , Article number:  98 ( 2022 ) Cite this article

2865 Accesses

10 Citations

1 Altmetric

Metrics details

Cephalometric analysis is traditionally performed on skull lateral teleradiographs for orthodontic diagnosis and treatment planning. However, the skull flattened over a 2D film presents projection distortions and superimpositions to various extents depending on landmarks relative position. When a CBCT scan is indicated for mixed reasons, cephalometric assessments can be performed directly on CBCT scans with a distortion free procedure. The aim of the present study is to compare these two methods for orthodontic cephalometry.

114 CBCTs were selected, reconstructed lateral cephalometries were obtained by lateral radiographic projection of the entire volume from the right and left sides. 2D and 3D cephalometric tracings were performed. Since paired t-tests between left and right-side measurements found no statistically significant differences, mean values between sides were considered for both 2D and 3D values. The following measurements were evaluated: PNS-A; S-N; N-Me; N-ANS; ANS-Me; Go-Me; Go-S; Go-Co; SNA, SNB, ANB; BaŜN; S-N^PNS-ANS; PNS-ANS^Go-Me; S-N^Go-Me. Intraclass correlation coefficients, paired t-test, correlation coefficient and Bland–Altman analysis were performed to compare these techniques.

The values of intra- and inter-rater ICC showed excellent repeatability and reliability: the average (± SD) intraobserver ICCs were 0.98 (± 0.01) and 0.97(± 0.01) for CBCT and RLCs, respectively; Inter-rater reliability resulted in an average ICC (± SD) of 0.98 (± 0.01) for CBCT and 0.94 (± 0.03) for RLC. The paired t-tests between CBCT and reconstructed lateral cephalograms revealed that Go-Me, Go-S, PNS-ANS^Go-Me and S-N^Go-Me measurements were statistically different between the two modalities. All the evaluated sets of measurements showed strong positive correlation; the bias and ranges for the 95% Limits of Agreement showed higher levels of agreement between the two modalities for unpaired measurements with respect to bilateral ones.

The cephalometric measurements laying on the mid-sagittal plane can be evaluated on CBCT and used for orthodontic diagnosis as they do not show statistically significant differences with those measured on 2D lateral cephalograms. For measurements that are not in the mid-sagittal plane, the future development of specific algorithms for distortion correction could help clinicians deduct all the information needed for orthodontic diagnosis from the CBCT scan.

Peer Review reports

Cephalometric analysis was defined by Moyers as a crucial tool to improve our understanding of the morphological characteristics of craniomaxillary structures [ 1 ]. It represents the reference standard for evaluating the bony structures for orthodontic purposes in conjunction with orthopantomography [ 1 ]. However, two-dimensional (2D) lateral cephalometric analyses is hindered by several limitations when a three-dimensional (3D) object is to be studied, because a 3D anatomical structure flattened on a sagittal plane presents distortions and superimposition of bony structures [ 2 ]. Projection errors depend on the representation of a 3D object on a 2D image and to the imaging technology, in particular the distance between the focus, the head, and the film. The resulting superimposition of anatomical structures complicates image interpretation and landmark identification. These distortions reduce measurement accuracy, especially for landmarks far from the midsagittal plane (MSP) [ 3 ].

The use of multi-slice computed tomography (MSCT), allowed to get rid of these limitations and to perform a 3D evaluation of bony structures thus solving the problem of overlapped reference areas [ 3 ]. On the other hand, this imaging modality delivers a significantly higher dose of ionizing radiation compared to 2D cephalograms and therefore its application was restricted to specific cases [ 4 , 5 , 6 ]. Recently, the development of cone-beam computed tomography (CBCT) allowed to obtain precise 3D imaging with consistently less radiation compared to conventional CT scans [ 7 , 8 , 9 , 10 ]. However, since it delivers a significantly higher radiation dose compared to lateral teleradiography, the latter remains the method of choice to assess malocclusions and maxillofacial growth; and to evaluate the effect of facial orthopaedics, orthodontics and orthognathic surgery [ 11 , 12 ].

The limitation of conventional cephalometric measurements and their potential impact on orthodontic diagnosis was assessed in some studies focused on the direct comparison between CBCT and lateral cephalograms [ 13 , 14 ]. However, the need to obtain both imaging modalities in each patient made it more complex to collect samples large enough to draw meaningful conclusions. Since Kumar et al. [ 15 ] demonstrated no significant differences between linear distances and angles assessed with reconstructed lateral cephalometric radiographs (RLC a 2D image that can be obtained from a CBCT scan), and lateral skull teleradiography, it is possible to obtain 2D images from CBCT scans taken for specific indications and use them for research purposes.

The papers by the research groups leaded respectively by Baumrind and Frantz [ 11 , 12 ] on the reliability of 2D landmarks reported the envelopes of error for cephalometric points on skull lateral teleradiography. Considering the ease in landmark identification and the lack of distortion, the envelope of error of CBCT cephalometry is expected to be different both in shape and magnitude [ 14 ]. Available literature on 3D landmark reliability, according to the authors’ scoping review, demonstrated a greater reliability for many cephalometric measurements than the one performed on 2D cephalometric analysis [ 16 , 17 , 18 ].

However, the specific envelopes of error of 3D cephalometric points have not been investigated thoroughly and, even more important, we still miss the correspondent set of reference values. The primary aim of this paper is to measure the differences between 2 and 3D cephalometric variables, focusing on the differences between angles and distances coplanar to the mid-sagittal plane and those that lay on a different plane and are therefore more subject to errors. To the scope, the hypothesis that measurements not belonging to the MSP are affected by a mixed amount of distortion depending on their 3D position will be tested. The data gathered in this study can be a valid starting point for the further analysis of correction coefficients for 3D comparison with 2D normal values of the different measurements depending on the distribution of the errors provided by the Bland–Altman analysis.

Materials and methods

Study design and sample selection.

The patients’ records used in the present cross-sectional study were retrieved from the Dental Department of the Fondazione IRCCS Cà Granda Ospedale Maggiore Policlinico, Milan, Italy. The inclusion criteria were:

All subjects presenting full permanent dentition apart from the third molar;

Skeletal Class I according to Steiner (ANB angle between 0° and 4°, measured on the latero-lateral projection) and a maximum difference of 3 mm between the distance of each Gonion and Maxillaris point from the MSP in the postero-anterior projection [ 19 ];

Absence of cross-bite as reported in patients’ records and confirmed on the CBCT scans.

The exclusion criteria were:

Missing molars or premolars;

Previous orthodontic treatment;

Altered bone metabolism;

Skeletal asymmetry between right and left cephalometric variables greater than 2 mm;

Alterations to the maxillofacial skeleton (acquired or congenital).

The purpose of selecting symmetric skeletal Class I subjects is to design a simplified model in which the difference between imaging modalities was as much as possible due to the measuring instruments and not to the differences between patients [ 14 ].

A total of 750 CBCT taken from January 2012 to June 2016 for mixed reasons (impacted and supernumerary teeth; bicuspid tooth implant needs; obstructive sleep disorders breathing and apnea syndrome; orthognathic surgery; trauma not involving mandibular or maxillary position; foreign objects) at the Dental Department of the Fondazione IRCCS Cà Granda Ospedale Maggiore Policlinico were reviewed and 114 patients’ CBCTs were selected. The sample was composed of Caucasian subjects: 56 males aged between 7 and 21 years, and 59 females between 8 and 19.5 years. All the CBCTs were performed with the same scanner, I-CAT FLX (Imaging Sciences International, Hatfield, PA, USA). The scanning protocol involved a 4 mm slice thickness, a 16 × 22 cm field of view, a 20-s scan time, and a 0.49/0.49/0.5 mm voxel size.

Ethical approval and informed consent

The ethical approval of the present study was obtained by the Fondazione IRCCS Cà Granda, Ospedale Maggiore, Milan, Italy (09/03/2016; n. 421). The protocol was designed in accordance with the Helsinki Declaration, including all amendments and revisions. All patients and the appropriated person who has parental authority gave their written informed consent for all the procedures that are described hereafter and for the data gathered from their records to be used for scientific purposes.

Data elaboration

Raw data from the CBCT scan were coded into Digital Imaging and Communications in Medicine (Dicom3) file format. These data were then processed into Mimics software to perform 3D cephalometric tracings (version 20.0, Materialise, Leuven, Belgium; https://www.materialise.com/en/medical/mimics-innovation-suite/mimics ). Two RLCs (right and left) were reconstructed for each CBCT scan by lateral radiographic projection of the entire volume using iCAT Vision software (Imaging Sciences International, Inc., https://ct-dent.co.uk/i-cat-vision/ ) (Fig.  1 ). All 2D cephalograms were then traced using a dedicated software (Dolphin Imaging Cephalometric and Tracing Software, version 11.9, Chatsworth, California, https://www.dolphinimaging.com/Media/DolphinNews?Subcategory_OS_Safe_Name=20160913 ).

Right and left reconstructed lateral cephalograms

Cephalometric points on CBCT scans were firstly identified in one plane (axial, coronal or sagittal) and then checked in the other two and in the 3D volumetric rendering.

• Cephalometric analysis

2D and 3D cephalometric tracings were performed by two researchers with at least 10 years of experience in 2D and in 3D cephalometry (researcher I, CS; researcher II, GMT). These researchers repeated the tracings after 15 days to assess repeatability and reliability. Each patient was identified using a random identification code, and the researchers were blinded to the subjects’ identities when performing 2D and 3D cephalometric analysis.

Three reference planes for the 3D cephalometric analysis were identified as follows:

Midsagittal plane (MSP) passing through Ba (Basion), S (Sella), and N (Nasion);

Axial plane passing through N, S and normal to MSP;

Coronal plane passing through S and normal to the other two planes.

Sella is intersected by the three anatomical planes, and it is the center of the reference system (point 0, 0, 0).

Fourteen cephalometric landmarks, ten unpaired (i.e., on the MSP) and four lateral symmetrical, were identified in CBCT axial, coronal, and sagittal sections: N (Nasion), S (Sella), Ba (Basion), A (Point A), B (Point B), Me (Menton), PNS (Posterior Nasal Spine), ANS (Anterior Nasal Spine), UI (Upper Incisor), LI (Lower Incisor); and paired landmarks: Sor (Supra Orbital), Mx (Maxillar), Co (Condylion) and Go (Gonion). The position of each point was then checked on the 3D volumetric rendering (Fig.  2 ).

Cephalometric tracing of a CBCT on Materialise Mimics. At the bottom right some cephalometric points and all the three reference planes are visible

The cephalometric analysis was performed according to classical Steiner methods [ 20 ]. A total of 15 measurements (20 considering the paired ones, too): 8 linear (unit: mm) of which 3 paired and 7 angular (unit: degrees) measurements in the vertical, sagittal, and transverse planes of which 2 paired were automatically generated by the program. Measurements taken into consideration are listed and explained in Table 1 .

Statistical analysis

A preliminary analysis was performed on the records of 30 patients to gain information for sample size calculation. Sample size calculation was performed a priori using MedCalc® Statistical Software version 20.013 (MedCalc Software Ltd, Ostend, Belgium; https://www.medcalc.org ; 2021) to estimate the required total number of cases for a method comparison study using the Bland–Altman statistics to obtain a statistical power of the study greater than 0.95 at an α = 0.05. The data used in the analysis were relative to PNS-ANS^Go L-Me as they were that required: Mean Difference between the two methods = 14.76, σ = 2.61 and Maximum Allowed Difference between methods equal to 21.52. Based on these parameters, the required sample size was 105 patients. This measurement was chosen as it was the one that required the largest sample size, thus ensuring a sufficient statistical power for all the variables under evaluation in the present study. In order to ensure robustness of data 114 records were selected.

Before further analyses, paired t-tests were conducted between left and right homologous measurement to check if they were significantly different. As shown in Table 2 , there was not a significant difference between the two sides for all the considered measurements, so bilateral values were averaged. The reason of this choice was to reduce the number of variables, also considering that 2D cephalometric analysis was performed on 2D RLCs, thus on 2D projected images.

The collected data were statistically analysed using IBM SPSS software (IBM Corp. Released 2020. IBM SPSS Statistics for Windows, Version 27.0. Armonk, NY: IBM Corp). Data distribution was assessed using Kolmogorov-Smirnoff test and confirmed data had a Gaussian distribution. In order to evaluate the intra-rater reliability, the variation of data measured by the same rater in two observations under the same conditions, an Intra Class Coefficient (ICC) was calculated. To quantify the inter-rater reliability, the variation of data measured by different raters, the ICC was estimated after a multilevel mixed-effects linear regression. The values of intra- and inter-rater ICCs were interpreted according to Cicchetti and Sparrow [ 21 ]: [0; 0.40) poor, [0.40; 0.60) fair, [0.60; 0.75) good, and [0.75; 1.0] excellent reliability. After this check, averaged values across all operators’ measurements were considered.

The mean value and standard deviation of each measurement were computed separately for 3D and 2D values. Standard Errors of Measurement (SEM) and 95% Confidence Intervals (CI) were calculated. Moreover, agreement and correlation between the two sets of values, 3D and 2D cephalometric measurements, were evaluated using Bland–Altman and scatter plots, respectively. In particular, Bland–Altman analysis evaluates the agreement between two sets of measurements, and it is usually used in a clinical context to compare a new measurement method against a gold standard. In this study, the averaged values associated to linear and angular measurements relative to each patient were employed for Bland–Altman analysis. The diagram was constructed by plotting the differences d i between patient \(i\) 3D vs 2D measurement value on the y-axis against their averaged value m i on the x-axis.

with bias \(\overline{d }\) between the two sets equal to the mean of the differences d i :

and upper and lower levels of agreement ( \(LOA\) ) equal to

where s is the standard deviation across \({d}_{i}\) differences, and \(LOA\) represent the limits at the 95% confidence interval of having normally distributed 3D vs 2D differences.

Table 2 reports the paired t-test between left and right homologous measurements. The test was performed for both 2D and 3D sets of values to check for differences between them. As shown in the table, homologous values were not significantly different for all the considered measurements (p-value > 0.05), so the averaged value between them was considered for each pair.

Table 3 reports the results of intra-rater and inter-rater reliability evaluated for each measurement. In all sets, the values of intra- and inter-rater ICC showed excellent repeatability and reliability for both 2D and 3D measurements (ICC > 0.75) [ 21 ]. The average ± standard deviation (SD) intraobserver ICCs were 0.98 (± 0.01) and 0.97(± 0.01), respectively for CBCT and 2D cephalograms. Inter-rater reliability resulted in an average ICC (± SD) of 0.98 (± 0.01) for CBCT and 0.94 (± 0.03) for 2D cephalometry.

Table 4 presents mean, SD, SEM, lower and upper limits of 95% CIs of cephalometric values obtained using the 3D and 2D analyses and paired t-test between 2 and 3D measurements. Table 5 presents the comparisons by means of correlation coefficients and Bland–Altman analysis for the same variables. In general, all the sets of measurements showed strong positive correlations: 0.995 ± 0.012 for unpaired measurements and 0.924 ± 0.060 for bilateral measurements. For Bland–Altman analysis, the biases and ranges for the 95% Limits Of Agreement of each measurement showed higher levels of agreement between the two modalities for unpaired measurements relative to bilateral ones. For unpaired measurements, bias was in the range [0.01; 0.02] mm for linear and [− 0.02; 0.13]° for angular values, while for bilateral measurements bias was between 0.27 and 12.30 mm for linear and between 11.51 and 14.64° for angular values. Since the Bland Altman plots and the scatter plots relative to the measurements that belong to the same group (i.e. bilateral and unpaired measurements) show similar trends, one linear variable and one angular variable were selected for each group so as to avoid redundancy. The Bland Altman plots and the scatter plots relative to unpaired measurements (N-ANS and SNA) are shown in Fig.  3 and those relative to bilateral measurements (Go-S and PNS-ANS^Go-Me) are reported in Fig.  4 .

Bland–Altman plots (panels A , C ) and scatter plots (panels B , D ) between CBCT measurements (3D values) and RLC measurements (2D values). Results are presented for unpaired measurements: linear measurement N-ANS and angular measurement SNA

Bland–Altman plots (panels A , C ) and scatter plots (panels B , D ) between CBCT measurements (3D values) and RLC measurements (2D values). Results are presented for bilateral measurements: linear measurement Go-S and angular measurement PNS-ANS^Go-Me

Scatterplots (panels B and D of Figs.  3 and 4 ) showed an overall linear correlation between 3 and 2D measurements across all conditions. Unpaired measurements showed better agreement in the Bland–Altman plots and higher correlation values in the scatterplots compared to bilateral measurements (see Table 5 ).

While most of the clinical orthodontic knowledge related to craniofacial analysis derives from 2D cephalometric analysis, the assessment of linear and angular measurements using 2D radiographs raises an important issue as a 3D object is flattened into a 2D image. In fact, the main disadvantages of the 2D conventional cephalometric analysis are represented by projective distortions and rotational errors, which might affect the reliability and reproducibility of the measured values, in particular those not belonging to the MSP [ 3 ]. In order to overcome the limitations associated with traditional cephalometric analysis, cephalometric measurements using CBCT images have been introduced, and found to have a similar performance than 2D cephalometric tracings, with the great advantage to provide reliable 3D information while using a single cephalometric analysis and not two projections (i.e., lateral and posteroanterior) [ 22 , 23 ]. To date, CBCT imaging has a crucial role for the assessment and management of complex cases in orthodontics and in maxillofacial surgery. Since it delivers higher radiation doses compared to 2D radiology, its application is limited to specific cases where a 3D assessment of the jaw is required. In these cases, performing 3D cephalometry directly on the CBCT scan could be useful to avoid delivering further radiation.

The current study pointed out the differences in measuring distances and angles that are coplanar and not to the MSP using CBCT and RLCs with the aim to identify which cephalometric measurements can be measured by both techniques with an acceptable margin of error. To limit the biological variety and focus on inter subject variability, in this first study only symmetrical Class I patients were selected; the paired t-test between sides confirmed the absences of statistically significant differences for each paired measurement and allowed to consider the averaged values between sides in further analysis. The paired t-tests between CBCT and RLC revealed that the values whose differences were statistically significant were Go-Me, Go-S, PNS-ANS^Go-Me and S-N^Go-Me. These results can be explained by the distortion error due to the lateral landmarks projection on the MSP. About that, a recent article compared RLC and CBCT [ 14 ] focusing on the changes between imaging techniques’ outcomes during mandibular growth and reported that the extent of distortion was positively correlated to the angle of incidence of the segment Go-Me. In our study, Go-Me and Go-S were found to suffer a similar distortion for the same reason. Considering the two angular measurements (PNS-ANS^Go-Me and S-N^Go-Me), the statistically significant difference between the methods can be due to the interaction of a line that is coplanar with the MSP and a segment that is instead para-axial to it.

Scatterplots demonstrated the high positive correlation between 2 and 3D measurements, while further information could be obtained by Bland–Altman analysis, the last providing useful information to assess the bias of the measurements to formulate educated guesses about their correction using specific algorithms to be investigated in future studies. For all unpaired measurements the bias was close to 0, meaning that there was not a systematic error between the two sets of 3D and 2D measures. Moreover, the limits of agreement were narrow and there were few outliers with respect to them: the two methods were essentially equivalent. On the other hand, even if the correlation between bilateral measurements obtained by 2D and 3D techniques was strong, Bland–Altman analysis evidenced a bias between the two sets. This bias was not a constant value for all the considered patients and values were normally distributed inside the LOAs. An exception was represented by the linear measurement Go-Co: the difference between the averages of CBCT and RLCs measurements was not significant (p > 0.05), the bias was close to zero (0.27 mm) and the LOAs were quite narrow (-0.17; 0.72 mm).

This was allegedly due to anatomical reasons: this linear measurement (mandibular ramus height) bears almost no distortion when projected into the MSP since it is almost parallel to it. Similar results were previously obtained for the same measurements although with different methods [ 18 ]. In general, for angular measurements, such as PNS-ANS^Go-Me in Fig.  4 , there was a trend in the distribution of the differences against the mean values, i.e., differences between the methods tend to get smaller as the mean increases. Probably, for smaller angles, the error of projection on the MSP causes a relatively greater projective error. Furthermore, high angle patients usually present reduced mandibular width [ 24 ] thus a reduced distortion as the Go-Me segment is less rotated from the MSP. The lowest correlation between 2 and 3D measurements was relative to PNS-ANS^Go-Me. This event could allegedly be due to the unpredictable distortion of the segment Go-Me that affects the angle in an unpredictable way since the other segment of this angle is not coplanar with it and belongs to the MSP.

Both techniques appeared to have excellent reproducibility (0.98–0.96 for CBCT and 0.96–0.84 for RLC) and reliability (0.99–0.97 for CBCT and 0.98–0.96 for RLC) as measured by ICC index and therefore they appeared allegedly comparable in terms of diagnostic accuracy. Available literature agrees on the high reproducibility and reliability of both the imaging methods as assessed in studies using physical measurements and scans made on dry skulls [ 3 , 17 , 25 , 26 ] and in other studies comparing cephalometric measurements on 2D cephalograms and on CBCT scans [ 13 , 27 , 28 ], but little information on which measurements are subjected to greater errors and which of them can be corrected and efficiently used for orthodontic diagnosis are evaluated [ 18 , 29 ].

To the authors’ knowledge, in the first study that applied Bland–Altman analysis to assess biases between conventional 2D measurement system and 3D imaging method [ 3 ], dry skulls were measured using a digital caliper to establish the linear physical measurements as a gold standard to make comparisons between lateral cephalograms and reconstructed 3D images. However, the 3D imaging relied only on the acquisition of three 2D cephalograms (standard lateral (90°), frontal (0°), and oblique (45°)) merged into the same 3D matrix. The evaluated 3D approach showed high precision with a greatly reduced bias to the gold standard and much less variability in its measure compared with the conventional 2D approach.

Regarding the evaluation of the difference between conventional 2D cephalometry and CBCT in vivo on patients, Oz et al. [ 13 ] performed a study on 11 patients with several limitations: despite sample size calculation was declared no information on the exact calculation was provided nor allegedly sufficient sample numerosity was reached to obtain meaningful inferential statistics. Also, no clear description of 3D cephalometric measurements was given and no information regarding bias between the two imaging methods was given. They found no statistical differences among 2D and 3D CBCT-generated cephalogram measurements, except for Go-Me and Condylion-Gnathion (Co-Gn) linear measurements. Li et al. [ 28 ] analyzed the differences between CBCT and RLC cephalometric methods on 40 patients by means of a paired t -test. The results indicated that the two methods showed significant differences in all measurements (SNA, SNB, ANB, MP-FH, SND, U1-NA linear mesurement, U1-NA angle, U1-L1, L1-NB linear measurement, L1-NB angle, L1-MP, L1-FH, OP to SN, Pog to NB, GoGn to SN). These results however are mainly due to the statistics that have been chosen to perform the comparison. In fact a paired t-test detects systematic differences even if they are of low or of no clinical significance. Pittayapat et al. [ 27 ] focused on the comparison of the CBCT vs physical measurements and lateral cephalometry vs physical measurements measured on dry skulls by means of inter- and intra-observer variability expressed as a percentage of coefficients of variability (absolute difference between methods below 1%). No direct comparison was performed between CBCT and lateral cephalogram nor any inferential statistics aiming to compare coefficients of variation, as done in other studies [ 14 ], was performed. The Go-Me segment was the most affected by the distortion of the projection, as found in our study. Gribel et al. [ 18 ] performed 12 craniometric measurements on 25 dry skulls and compared them with cephalometric indexes on lateral teleradiography and on CBCT scans. No meaningful difference was observed between the direct assessments on dry skulls and the on CBCT scans on analysis of variance (ANOVA) (P > 0.05). The comparison between all the cephalometric indexes and the craniometrics measurements retrieved statistically significant results (Tukey test, P < 0.05). Great differences were observed between different variables. Gribel et al. [ 29 ] stated that none of assessed cephalometric variables measured in CBCT were significantly different from the craniometrics assessments if a trigonometric algorithm correction was applied on 2D measurements taken on lateral cephalogram. However, its evidence has low clinical significance since the CBCT data were used to derive individualized correction factors that are applicable for that particular subject only. Measurements on the MSP were calculated simply by reducing the 10% magnification of the cephalometric distortion since the MSP is already parallel to the cephalometric film. However, in all these studies no information regarding bias between the two imaging methods was given.

Limitations of the study

In this study, patients with facial asymmetry in the bilateral cephalometric variables greater than 2 mm were excluded to reduce the errors of 2D RLC tracing in order to obtain a simplified model to test our hypothesis that measurements not laying on the MSP are affected by a certain amount of distortion that could be estimated and corrected.

Also, the study investigates only skeletal Class I patients; further investigations should focus on the morphological cases that occur on skeletal Class 2 and 3 patients. For these patients with more complex morphologies, a CBCT is usually available to assess the facial structure and investigate the asymmetry between the sides of the face. However, the study design would be much more complex. Moreover, a possible development of the present study could include the estimation of 2D measurements from 3D ones. Measuring 2D from 3D linear and angular distances can be useful in cases where you have a patient's CBCT and want to trace 2D data without exposing the patient to further radiation.

The analysis of the effect of the facial orthopaedic treatment of the mandibular shape could also be of some interest. Evaluation of patients with different skeletal classes and those with varying degrees of asymmetry would be useful. Such data would allow conclusions to be drawn that could be more easily extended to the entire population, since the results of the present study only fit people with a satisfactory degree of facial symmetry and a normal anteroposterior maxilla-mandibular relationship. It may be useful to widen the sample size by a multicentre approach so to increase the robustness of our findings on this debated topic.

It appears that most cephalometric indexes investigated in this study can be measured on CBCT and used for orthodontic diagnosis since they do not present statistically significant differences with the ones measured on 2D lateral cephalograms. However, the measurements where one or more landmarks lay far from the MSP bear distortions that could allegedly be overcome using specific formulas converting 3D values into 2D ones as the performed Bland–Altman and correlation analyses seem to suggest. The future development of specific algorithms for this purpose could help clinicians avoid exposing patients to unnecessary 2D cephalometric lateral radiographs where CBCT is indicated and to deduct all information needed for orthodontic diagnosis from the CBCT scan.

Availability of data and materials

The data analysed during the current study are available on request due to restrictions, e.g., privacy or ethical.

Abbreviations

Cone beam computed tomography

Three-dimensional

Two-dimensional

Reconstructed lateral cephalogram

A of Down’s

B of Down’s

Anterior nasal spine point

Posterior nasal spine point

Menton point

Basion point

Sella point

Condylion point

Gonion point

Upper incisor point

Lower incisor point

Supra orbital point

Maxillary point

Intraclass correlation coefficient

Multi slice computed tomography

Midsagittal plane

Nasion point

Confidence interval

Standard errors of measurement

Levels of agreement

Moyers RE, Bookstein FL, Guire KE. The concept of pattern in craniofacial growth. Am J Orthod. 1979;76:136–48.

Article   Google Scholar  

Broadbent B. A new X-ray technique and its application to orthodontia. Angle Orthod. 1931;1:45–66.

Google Scholar  

Adams GL, Gansky SA, Miller AJ, Harrell WE, Hatcher DC. Comparison between traditional 2-dimensional cephalometry and a 3-dimensional approach on human dry skulls. Am J Orthod Dentofac Orthop. 2004;126:397–409.

Pearce MS, Salotti JA, Little MP, McHugh K, Lee C, Kim KP, et al. Radiation exposure from CT scans in childhood and subsequent risk of leukaemia and brain tumours: a retrospective cohort study. Lancet. 2012;380:499–505.

Oenning AC, Jacobs R, Pauwels R, Stratis A, Hedesiu M, Salmon B. Cone-beam CT in paediatric dentistry: DIMITRA project position statement. Pediatr Radiol. 2018;48:308–16.

Pauwels R, Beinsberger J, Collaert B, Theodorakou C, Rogers J, Walker A, et al. Effective dose range for dental cone beam computed tomography scanners. Eur J Radiol. 2012;81:267–71.

Gateno J, Xia JJ, Teichgraeber JF. New 3-dimensional cephalometric analysis for orthognathic surgery. J Oral Maxillofac Surg. 2011;69:606–22.

Palomo JM, Kau CH, Palomo LB, Hans MG. Three-dimensional cone beam computerized tomography in dentistry. Dent Today. 2006;25:130–5.

PubMed   Google Scholar  

Scarfe WC, Farman AG, Levin MD, Gane D. Essentials of maxillofacial cone beam computed tomography. Alpha Omegan. 2010;103:62–7.

Kapila SD, Nervina JM. CBCT in orthodontics: assessment of treatment outcomes and indications for its use. Dentomaxillofac Radiol. 2015;44:20140282.

Baumrind S, Frantz RC. The reliability of head film measurements: 2. Conventional angular and linear measures. Am J Orthod. 1971;60:505–17.

Baumrind S, Frantz RC. The reliability of head film measurements: 1. Landmark identification. Am J Orthod. 1971;60:111–27.

Oz U, Orhan K, Abe N. Comparison of linear and angular measurements using twodimensional conventional methods and three-dimensional cone beam CT images reconstructed from a volumetric rendering program in vivo. Dentomaxillofac Radiol. 2011;40:492–500.

Maspero C, Farronato M, Bellincioni F, Cavagnetto D, Abate A. Assessing mandibular body changes in growing subjects: a comparison of CBCT and reconstructed lateral cephalogram measurements. Sci Rep. 2020;10:1–12.

Kumar V, Ludlow J, Mol A, Cevidanes L. Comparison of conventional and cone beam CT synthesized cephalograms. Dentomaxillofac Radiol. 2007;36:263–9.

Brown AA, Scarfe WC, Scheetz JP, Silveira AM, Farman AG. Linear accuracy of cone beam CT derived 3D images. Angle Orthod. 2009;79:150–7.

Moreira CR, Sales MAO, Lopes PML, Cavalcanti MGP. Assessment of linear and angular measurements on three-dimensional cone-beam computed tomographic images. Oral Surg Oral Med Oral Pathol Oral Radiol Endodontol. 2009;108:430–6.

Gribel BF, Gribel MN, Frazão DC, McNamara JA, Manzi FR. Accuracy and reliability of craniometric measurements on lateral cephalometry and 3D measurements on CBCT scans. Angle Orthod. 2011;81:28–37.

Hwang H-S, Youn I-S, Lee K-H, Lim H-J. Classification of facial asymmetry by cluster analysis. Am J Orthod Dentofac Orthop. 2007;132:279.e1-279.e6.

Steiner CC. Cephalometrics in clinical practice. Angle Orthod. 1959;29:8–29.

Cicchetti DV, Sparrow SA. Developing criteria for establishing interrater reliability of specific items: applications to assessment of adaptive behavior. Am J Ment Defic. 1981;86:127–37.

Leonardi R. Cone-beam computed tomography and three-dimensional orthodontics. Where we are and future perspectives. J Orthod. 2019;46(1_suppl):45–8.

Maspero C, Abate A, Bellincioni F, Cavagnetto D, Lanteri V, Costa A, et al. Comparison of a tridimensional cephalometric analysis performed on 3T-MRI compared with CBCT: a pilot study in adults. Prog Orthod. 2019;20:40. https://doi.org/10.1186/s40510-019-0293-x .

Article   PubMed   PubMed Central   Google Scholar  

Enlow DH, Moyers RE. Handbook of facial growth. WB Saunders Company; 1982.

van Vlijmen OJC, Bergé SJ, Swennen GRJ, Bronkhorst EM, Katsaros C, Kuijpers-Jagtman AM. Comparison of cephalometric radiographs obtained from cone-beam computed tomography scans and conventional radiographs. J Oral Maxillofac Surg. 2009;67:92–7.

Lascala C, Panella J, Marques M. Analysis of the accuracy of linear measurements obtained by cone beam computed tomography (CBCT-NewTom). Dentomaxillofac Radiol. 2004;33:291–4.

Pittayapat P, Bornstein MM, Imada TSN, Coucke W, Lambrichts I, Jacobs R. Accuracy of linear measurements using three imaging modalities: two lateral cephalograms and one 3D model from CBCT data. Eur J Orthod. 2014;37:202–8.

Li NA, Hu BO, Mi F, Song J. Preliminary evaluation of cone beam computed tomography in three-dimensional cephalometry for clinical application. Exp Ther Med. 2017;13:2451–5.

Gribel BF, Gribel MN, Manzi FR, Brooks SL, McNamara JA. From 2D to 3D: an algorithm to derive normal values for 3-dimensional computerized assessment. Angle Orthod. 2011;81:3–10.

Download references

Acknowledgements

Not applicable.

This study received no external funding.

Author information

Benedetta Baldini and Davide Cavagnetto contributed equally to this work and are therefore to be considered co-first authors

Authors and Affiliations

Facial Surgery and Dentistry Fondazione IRCCS Cà Granda, UOC Maxillo, Ospedale Maggiore Policlinico, 20142, Milan, Italy

Benedetta Baldini & Gianluca Martino Tartaglia

Department of Oral and Maxillofacial Surgery, Amsterdam University Medical Center (Amsterdam UMC), Location AMC, Meibergdreef 9, 1105 AZ, Amsterdam, The Netherlands

Davide Cavagnetto

Department of Surgical Sciences, University of Torino, Via Nizza 230, 10126, Torino, Italy

Department of Electronics, Information and Bioengineering, Politecnico Di Milano, 20133, Milano, Italy

Giuseppe Baselli

Faculty of Medicine and Surgery, Department of Biomedical Sciences for Health, Functional Anatomy Research Center (FARC), Università Degli Studi Di Milano, Milan, Italy

Chiarella Sforza

Department of Biomedical, Surgical and Dental Sciences, School of Dentistry, University of Milan, 20100, Milan, Italy

Gianluca Martino Tartaglia

You can also search for this author in PubMed   Google Scholar

Contributions

Conception and design of study: DC; Acquisition of clinical data: CS, GMT; Analysis and interpretation of data collected: BB; Drafting of article and/or critical revision: BB, DC, GB; All authors read and approved the final manuscript.

Corresponding author

Correspondence to Benedetta Baldini .

Ethics declarations

Ethics approval and consent to participate.

The ethical approval of the present study was obtained by the Fondazione IRCCS Cà Granda, Ospedale Maggiore, Milan, Italy (09/03/2016; n. 421). The protocol was designed in accordance with the Helsinki Declaration, including all amendments and revisions. All patients and the appropriated person who has parental authority gave their written informed consent to all the procedures that are described in the manuscript.

Consent for publication

All patients and the appropriated person who has parental authority gave consent for publication of data gathered in this study.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Baldini, B., Cavagnetto, D., Baselli, G. et al. Cephalometric measurements performed on CBCT and reconstructed lateral cephalograms: a cross-sectional study providing a quantitative approach of differences and bias. BMC Oral Health 22 , 98 (2022). https://doi.org/10.1186/s12903-022-02131-3

Download citation

Received : 14 December 2021

Accepted : 15 March 2022

Published : 30 March 2022

DOI : https://doi.org/10.1186/s12903-022-02131-3

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Three-dimensional imaging
• Orthodontics
• Maxillofacial
• Bland–Altman analysis

BMC Oral Health

ISSN: 1472-6831

• Submission enquiries: [email protected]
• General enquiries: [email protected]

• Open access
• Published: 31 May 2021

Comparison of cephalometric measurements between conventional and automatic cephalometric analysis using convolutional neural network

• Sangmin Jeon 1 &
• Kyungmin Clara Lee   ORCID: orcid.org/0000-0003-3102-4550 2  

Progress in Orthodontics volume  22 , Article number:  14 ( 2021 ) Cite this article

5894 Accesses

16 Citations

Metrics details

The rapid development of artificial intelligence technologies for medical imaging has recently enabled automatic identification of anatomical landmarks on radiographs. The purpose of this study was to compare the results of an automatic cephalometric analysis using convolutional neural network with those obtained by a conventional cephalometric approach.

Material and methods

Cephalometric measurements of lateral cephalograms from 35 patients were obtained using an automatic program and a conventional program. Fifteen skeletal cephalometric measurements, nine dental cephalometric measurements, and two soft tissue cephalometric measurements obtained by the two methods were compared using paired t test and Bland-Altman plots.

A comparison between the measurements from the automatic and conventional cephalometric analyses in terms of the paired t test confirmed that the saddle angle, linear measurements of maxillary incisor to NA line, and mandibular incisor to NB line showed statistically significant differences. All measurements were within the limits of agreement based on the Bland-Altman plots. The widths of limits of agreement were wider in dental measurements than those in the skeletal measurements.

Conclusions

Automatic cephalometric analyses based on convolutional neural network may offer clinically acceptable diagnostic performance. Careful consideration and additional manual adjustment are needed for dental measurements regarding tooth structures for higher accuracy and better performance.

Introduction

Cephalometric analysis is an essential diagnostic tool for the treatment planning and evaluation of orthodontic patients. Accurate identification of the anatomical landmarks on cephalograms is critical for a reliable cephalometric analysis [ 1 ]. Lateral cephalometric radiographs have been employed as an essential tool in orthodontics. However, to analyze such radiographs, the important anatomical structures need to be identified by a landmark identification and manual tracing process. However, this analysis requires a skilled orthodontist, and the process is time-consuming.

In computer science, artificial intelligence (AI) refers to the study of systems that perform tasks that require human intelligence using different computerized algorithms [ 2 , 3 ]. Machine learning is a method of data analysis that allows computer programs to automatically improve through cognitive content. It is a branch of technology that allows systems to learn from data, identify patterns, and make decisions with minimal human intervention [ 4 ]. These programs make decisions by examining large amounts of input data and with known outputs, subsequently, drawing conclusions on the input data with unknown outputs based on the initial “training” process.

In recent years, the use of AI in medicine and healthcare for the diagnosis and treatment of patients has been a topic of significant interest [ 5 ]. This has resulted in the application of AI and machine learning technologies to dental processes including the classification of temporomandibular joint osteoarthritis and osteoporosis, prediction of the debonding probability of computer-aided design/computer-aided manufacturing (CAD/CAM) crowns, automatic detection and classification of jaw lesions and periodontal bone loss, survival prediction of oral cancer patients, tooth labeling, detection and diagnosis of dental caries, and detection of osteoporosis [ 6 , 7 , 8 , 9 , 10 , 11 , 12 , 13 ]. Furthermore, programs have been developed to automatically digitize the anatomical structures on lateral cephalometric radiographs. With these programs, automatic cephalometric analysis including diagnostic and analytical imaging tasks can be performed by AI and machine learning technologies. However, to the best of our knowledge, few recent studies about AI performance of cephalometric analysis which is useful for clinicians are available. Previous studies about deep learning algorithm reported that AI accurately detected cephalometric landmarks [ 14 , 15 ]. In order to further explore the application of these technologies to clinical orthodontics, the results of clinical performance of cephalometric analysis are necessary. The purpose of the present study was to evaluate the accuracy of automatic cephalometric analysis by comparing with that of conventional cephalometric measurements.

This retrospective study was approved by the Institutional Review Board of the Chonnam National University Dental Hospital, Gwangju, Korea (CNUDH-EXP-2019-023). The inclusion criteria were (1) a fully erupted permanent dentition, and (2) no broad prosthetic restorations such as metal crowns or bridges, on the molars. The exclusion criteria were (1) multiple missing tooth and broad prosthetic restorations such as metal crowns or bridges, on the molars and (2) history of orthodontic treatment or orthognathic surgery. Conventional lateral cephalograms of 35 orthodontic patients (20 men, 15 women; mean age = 23.8 years) were obtained using OrthoCeph ® OC100 (Instrumentarium Imaging Co., Tuusula, Finland). The cephalograms were imported to the V-ceph TM (version 8.0, Cybermed Inc., Seoul, Korea) for the conventional cephalometric analysis and to the CephX TM (ORCA Dental AI Inc., Herzliya, Israel) for the AI analysis (Fig. 1 ). Sixteen anatomical landmarks were chosen (Table 1 ), and 15 skeletal cephalometric measurements, 9 dental cephalometric measurements, and 2 soft tissue cephalometric measurements were obtained by an experienced single examiner with over 7 years of experience in orthodontic treatment.

Cephalometric analysis using conventional ( a ) and AI ( b ) methods

Statistical analysis

The sample size calculation was performed according to the result of previous study of Hwang et al. [ 15 ]. The effect size was calculated to 0.49. A statistical power of 80 percent and a type I error of 5 percent was assumed by the G*power program (version 3.1.9.2, Heinrich-Heine-University, Dusseldorf, Germany). The calculation indicated that 35 individuals were required in the study.

All data were revealed to be normally distributed. Paired t test was then performed to determine the differences between the AI and conventional programs. For the purpose of comparing the two measurements obtained from each two methods graphically, the differences between the two methods were plotted using Bland-Altman analysis [ 16 ]. Shapiro-Wilk test and paired t test were conducted using SPSS software package (version 23.0; IBM, Armonk, NY) and Bland-Altman plots were made by MedCalc (Ostend, Belgium). Significance level was set of 5%. To assess the errors of each method, the process of acquiring measurements using the conventional program was repeated after 2 weeks, and the measurement errors were calculated using Dahlberg’s method [ 17 ]. The range of error was 0.1 to 0.3 mm for the linear measurements and from 0.1 to 0.3° for the angular measurements. For inter-examiner reproducibility, the second examiner performed the process of acquiring measurements using the conventional program, and the measurements were compared with first examiner’s measurements using the intraclass correlation coefficient (ICC). The ICC values were found to be statistically insignificant showing a mean of 0.91 (ICC 0.88-0.94), indicated excellent reliability.

Table 2 summarizes the differences between the measurements obtained by the conventional and AI methods. Statistically significant differences were found in saddle angle, linear measurements of maxillary incisor to NA line, and mandibular incisor to NB line. The soft tissue measurements did not show any significant difference between the two methods.

All measurements were within the limits of agreement based on the Bland-Altman plots. The measurements that showed significance in the paired t test were within the limits of agreement (Figs. 2 , 3 , and 4 ). The widths of limits of agreement were wider in dental measurements than those in the skeletal measurements (Table 3 ).

Bland-Altman plots for the skeletal measurements in each conventional and AI methods. For each plot, the x -axis represents the mean of the compared measurements, and the y -axis represents the difference between the compared measurements. The blue line represents the bias, and the red-hashed lines represent the upper and lower limits of agreement

Bland-Altman plots for the dental measurements in each conventional and AI methods

Bland-Altman plots for the soft tissue measurements in each conventional and AI methods

In orthodontics, cephalometric analysis is commonly performed by computerized method, which includes manual identification of the landmarks on a monitor. The software automatically calculates the distances and angles which are necessary for the cephalometric analysis. Otherwise, direct tracing of the radiograph is transferred to a computer. These computerized cephalometric analyses may cause some errors, such as transferring and measurement errors, even though the manual landmark identification is performed by a clinician [ 18 , 19 ]. Leonardi et al. [ 20 ] reported that the accuracy of a cephalometric analysis varies between 60 and 80% for a computerized analysis compared with the fully manual process, where the total errors should be no more than 0.59 mm in the x direction and 0.56 mm in the y direction to be considered acceptable. Recent studies showed that despite this, cephalometric analysis performed by computerized systems appear to be considered reliable [ 21 , 22 , 23 ]. However, the process of manually identifying cephalometric landmarks on cephalograms requires a lot of time and has possibility of errors regardless of the experience of the clinician. Since the first study on automatic identification of cephalometric landmarks by Levy-Mandel et al. [ 24 ] in 1986, several researchers have tried to automate landmark identification using knowledge-based techniques or image matching methods and learning systems. However, only a few clinical studies have been conducted on automatic landmark identification [ 25 , 26 , 27 , 28 ].

The program used in this study was Ceph-X. The program is based on the machine learning; automatic landmark localization algorithm is based on convolutional neural network. The program requires the confirmation of landmark position before calculating measurements. Full automation of all steps is challenging due to overlaying structures and inhomogeneous intensity values in the cephalometric radiographs. Thus, calculating measurements process may not be performed by AI. This study is conducted to provide a clear picture about the possibility of replacing the traditional cephalometric process with the digital one. The study focused mainly to evaluate its usability for cephalometric analysis and measurements using automated program.

A previous study reported that this system shows an accuracy of 96.6% when compared with manual cephalometric approaches, with an acceptable variation of less than approximately 0.5 mm and 1° [ 29 ]. Our results showed that three measurements, including the saddle angle, linear measurements of maxillary incisor to NA line and mandibular incisor to NB line exhibit statistically significant differences between the conventional and AI methods. The landmark identification of tooth structures can be affected by the surrounding superimposing anatomical structures, and clinicians also make this error. Particularly, identifying the mandibular incisor is difficult because it is generally located below the maxillary incisor due to overjet and overbite. Moreover, the widths of limits of agreement in the Bland-Altman plots were wider in dental measurements than those in the skeletal measurements. AI may have lower accuracy of performance in detecting tooth structures. The soft tissue measurements did not show any significant difference between the conventional and AI methods.

Based on the Bland-Altman plots, the measurements are in sufficiently good agreement. In the plots, the measurements that showed significant differences in the paired t test were within the limit of agreement (Figs. 2 , 3 , and 4 ). The wide limits of agreement in the Bland-Altman plots were defined clinically. Although there were statistically significant differences in some measurements and wide limits of agreement in the Bland-Altman plots between the two methods, the cephalometric analysis can be performed faster with the AI technique. In the present study, no manual adjustment after automatic landmark digitization was performed in order to exclusively evaluate the AI performance. With some manual adjustment made to landmark identification, the AI technique for cephalometric analysis may provide good performance. Considering that AI technologies will continue to improve in terms of the accuracy of measurement analysis with additional data and increasing use, the accuracy of cephalometric analysis based on AI techniques applied to clinical orthodontics will only further improve. Previous study by Hwang et al. [ 15 ] using recently proposed deep-learning method has reported that the mean error in landmark detection between AI and human was 1.46 ± 2.97 mm. In the present study, the mean error in all cephalometric measurements between conventional method and AI was 0.6 ± 3.1 mm. Although the errors in landmark identification cannot be compared directly with cephalometric measurements, the error using AI may be acceptable in clinics. In the study, the time needed for automatic tracing was within 5 s. In the conventional method, the mean time for tracing was about 6 min. Correcting lines requires lots of time. Considering this, automatic cephalometric analyses could help clinicians with manual adjustment.

The limitation of this study is that the sample size is smaller than that employed in previous studies on AI and machine learning technologies [ 30 , 31 ]. In addition, one kind of radiographic machine was used to take cephalometric radiographs in the present study. Since the software used in this study is a commercially available cephalometric analysis program, it is believed that the performance of the software may be same with the images taken by various radiographic machines.

With the limitation of this study, the results indicate that automatic cephalometric analyses based on convolutional neural network may offer clinically acceptable diagnostic performance. Careful consideration and additional manual adjustment are needed for dental measurements regarding tooth structures for higher accuracy and better performance.

Availability of data and materials

The data and materials obtained in this study belong to the authors, and are therefore available only upon request, after approval by the authors.

Abbreviations

• Artificial intelligence

Computer-aided design/computer-aided manufacturing

Tng TT, Chan TC, Hagg U, Cooke MS. Validity of cephalometric landmarks. An experimental study on human skulls. Eur J Orthod. 1994;16(2):110–20. https://doi.org/10.1093/ejo/16.2.110 .

Article   PubMed   Google Scholar  

Chartrand G, Cheng PM, Vorontsov E, Drozdzal M, Turcotte S, Pal CJ, et al. Deep learning: a primer for radiologists. Radiographics. 2017;37(7):2113–31. https://doi.org/10.1148/rg.2017170077 .

Russell S, Norvig P. Artificial intelligence: a modern approach. 3rd ed. Upper saddle river: Pearson; 2009.

Google Scholar  

Mitchell T. Machine learning. 1st ed. 2 Pennsylvania plaza: McGraw-Hill Education; 1997.

Liew C. The future of radiology augmented with artificial intelligence: a strategy for success. Eur J Radiol. 2018;102:152–6. https://doi.org/10.1016/j.ejrad.2018.03.019 .

de Dumast P, Mirabel C, Cevidanes L, Ruellas A, Yatabe M, Ioshida M, et al. A web-based system for neural network based classification in temporomandibular joint osteoarthritis. Comput Med Imaging Graph. 2018;67:45–54. https://doi.org/10.1016/j.compmedimag.2018.04.009 .

Article   PubMed   PubMed Central   Google Scholar  

Yamaguchi S, Lee C, Karaer O, Ban S, Mine A, Imazato S. Predicting the debonding of CAD/CAM composite resin crowns with AI. J Dent Res. 2019;98(11):1234–8. https://doi.org/10.1177/0022034519867641 .

Ariji Y, Yanashita Y, Kutsuna S, Muramatsu C, Fukuda M, Kise Y, et al. Automatic detection and classification of radiolucent lesions in the mandible on panoramic radiographs using a deep learning object detection technique. Oral Surg Oral Med Oral Pathol Oral Radiol. 2019;128(4):424–30. https://doi.org/10.1016/j.oooo.2019.05.014 .

Krois J, Ekert T, Meinhold L, Golla T, Kharbot B, Wittemeier A, et al. Deep learning for the radiographic detection of periodontal bone loss. Sci Rep. 2019;9(1):8495. https://doi.org/10.1038/s41598-019-44839-3 .

Kim DW, Lee S, Kwon S, Nam W, Cha IH, Kim HJ. Deep learning-based survival prediction of oral cancer patients. Sci Rep. 2019;9(1):6994. https://doi.org/10.1038/s41598-019-43372-7 .

Chen H, Zhang K, Lyu P, Li H, Zhang L, Wu J, et al. A deep learning approach to automatic teeth detection and numbering based on object detection in dental periapical films. Sci Rep. 2019;9(1):3840. https://doi.org/10.1038/s41598-019-40414-y .

Lee JH, Kim DH, Jeong SN, Choi SH. Detection and diagnosis of dental caries using a deep learning-based convolutional neural network algorithm. J Dent. 2018;77:106–11. https://doi.org/10.1016/j.jdent.2018.07.015 .

Cruz AS, Lins HC, Medeiros RVA, Filho JMF, da Silva SG. Artificial intelligence on the identification of risk groups for osteoporosis, a general review. Biomed Eng Online. 2018;17(1):12. https://doi.org/10.1186/s12938-018-0436-1 .

Park JH, Hwang HW, Moon JH, Yu Y, Kim H, Her SB, et al. Automated identification of cephalometric landmarks: Part 1-Comparisons between the latest deep-learning methods YOLOV3 and SSD. Angle Orthod. 2019;89(6):903–9. https://doi.org/10.2319/022019-127.1 .

Hwang HW, Park JH, Moon JH, Yu Y, Kim H, Her SB, et al. Automated identification of cephalometric landmarks: Part 2-Might it be better than human? Angle Orthod. 2020;90(1):69–76. https://doi.org/10.2319/022019-129.1 .

Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1(8476):307–10.

Dahlberg G. Statistical methods for medical and biological students. London: George Allen & Unwin Ltd; 1940. p. 122–32.

Richardson A. A comparison of traditional and computerized methods of cephalometric analysis. Eur J Orthod. 1981;3(1):15–20. https://doi.org/10.1093/ejo/3.1.15 .

Turner PJ, Weerakone S. An evaluation of the reproducibility of landmark identification using scanned cephalometric images. J Orthod. 2001;28(3):221–9. https://doi.org/10.1093/ortho/28.3.221 .

Leonardi R, Giordano D, Maiorana F, Spampinato C. Automatic cephalometric analysis: a systematic review. Angle Orthod. 2008;78(1):145–51. https://doi.org/10.2319/120506-491.1 .

Paixão MB, Sobral MC, Vogel CJ, Araujo TM. Comparative study between manual and digital cephalometric tracing using Dolphin Imaging software with lateral radiographs. Dent Press J Orthod. 2010;15(6):123–30. https://doi.org/10.1590/S2176-94512010000600016 .

Article   Google Scholar  

Erkan M, Gurel HG, Nur M, Demirel B. Reliability of four different computerized cephalometric analysis programs. Eur J Orthod. 2012;34(3):318–21. https://doi.org/10.1093/ejo/cjr008 .

Tikku T, Khanna R, Maurya R, Srivastava K, Bhushan R. Comparative evaluation of cephalometric measurements of monitor-displayed images by Nemoceph software and its hard copy by manual tracing. J Oral Biol Craniofacial Res. 2014;4(1):35–41. https://doi.org/10.1016/j.jobcr.2013.11.002 .

Lévy-Mandel AD, Venetsanopoulos AN, Tsotsos JK. Knowledge-based landmarking of cephalograms. Comput Biomed Res. 1986;19(3):282–309. https://doi.org/10.1016/0010-4809(86)90023-6 .

Yan C, Venetsanopoulos A, Fillery E. An expert system for landmarking cephalograms. 6th Internation Workshop on Expert Systems & Their Applications, Agence de l’Informatique; 1987. p. 337–56.

Parthasarathy S, Nugent ST, Gregson PG, Fay DF. Automatic landmarking of cephalograms. Comput Biomed Res. 1989;22(3):248–69. https://doi.org/10.1016/0010-4809(89)90005-0 .

Cohen JD, Dunbar K, McClelland JL. On the control of automatic processes: a parallel distributed processing account of the stroop effect. Psychological review. 1990;97(3):332–61. https://doi.org/10.1037/0033-295X.97.3.332 .

Davis DN, Forsyth D. Knowledge-based cephalometric analysis: a comparison with clinicians using interactive computer methods. Computers and Biomedical research. 1994;27(3):210–28. https://doi.org/10.1006/cbmr.1994.1018 .

Mosleh MA, Baba MS, Malek S, Almaktari RA. Ceph-X: development and evaluation of 2D cephalometric system. BMC Bioinformatics. 2016;17:S499.

Wang CW, Huang CT, Hsieh MC, Li CH, Chang SW, Li WC, et al. Evaluation and comparison of anatomical landmark detection methods for cephalometric x-ray images: a grand challenge. IEEE Trans Med Imaging. 2015;34(9):1890–900. https://doi.org/10.1109/TMI.2015.2412951 .

Nishimoto S, Sotsuka Y, Kawai K, Ishise H, Kakibuchi M. Personal computer-based cephalometric landmark detection with deep learning using cephalograms on the internet. J Craniofac Surg. 2019;30(1):91–5. https://doi.org/10.1097/SCS.0000000000004901 .

Download references

Acknowledgements

Not applicable

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (Ministry of Science and ICT) (No. 2020R1F1A1070617 and NRF-2017R1D1A1B03032132). All materials used here belonged to the authors and nothing was provided by third-parts or private companies; therefore, the authors have no conflict of interest related to the present work.

Author information

Authors and affiliations.

Gwangju, Republic of Korea

Sangmin Jeon

Department of Orthodontics, School of Dentistry, Chonnam National University, 33 Yongbong-ro, Buk-gu, Gwangju, 61186, Republic of Korea

Kyungmin Clara Lee

You can also search for this author in PubMed   Google Scholar

Contributions

SJ acquired all data and KCL analyzed and interpreted the data. The author(s) read and approved the final manuscript.

Corresponding author

Correspondence to Kyungmin Clara Lee .

Ethics declarations

Ethics approval and consent to participate.

All participants provided informed consents. The present study was approved by Chonnam National University Dental Hospital Institutional Review Board (CNUDH-EXP-2019-023).

Consent for publication

Not applicable.

Competing interests

The authors declare that we have no competing interests in relation to the present work.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Jeon, S., Lee, K.C. Comparison of cephalometric measurements between conventional and automatic cephalometric analysis using convolutional neural network. Prog Orthod. 22 , 14 (2021). https://doi.org/10.1186/s40510-021-00358-4

Download citation

Received : 04 November 2020

Accepted : 27 April 2021

Published : 31 May 2021

DOI : https://doi.org/10.1186/s40510-021-00358-4

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Cephalometric analysis
• Machine learning
• Convolutional neural network

Academia.edu no longer supports Internet Explorer.

To browse Academia.edu and the wider internet faster and more securely, please take a few seconds to  upgrade your browser .

Enter the email address you signed up with and we'll email you a reset link.

• We're Hiring!
• Help Center

The role of cephalometrics in orthodontic case analysis and diagnosis

1952, American Journal of Orthodontics

Related Papers

American Journal of Orthodontics and Dentofacial Orthopedics

Markus Bernhard

Forensic Science Medicine and Pathology

Dr Ravi bhaskar R

Dental Montes

European Scientific Journal

Mohammed Amayeri , fayez saleh

European journal of orthodontics

Herman Duterloo

The Professional Medical Journal

Faris Altaf

Background: Morphological characteristics of various races and populationgroups usually appear in geographical knots and clusters. Standardized cephalometricrecordsare immensely valuable for clinical and forensic purposes.In spite of its clinical significance nodata is available about the cephalic indices and prevalence of cephalic phenotypes in SouthernPunjab. Objectives: This study was undertaken to record baseline cephalometric data, cephalicindices and craniotypes. Study design: A cross-sectional population study. Place and durationof study: The study was carried out at the Multan Medical and Dental College, Multan and tookabout fourteen months to complete. Material and methods: The study was conducted on 672adult individuals (430 males and 242 females) from in and around the city of Multan in SouthernPunjab. Linear measurements of the head including maximum cranial length (glabella-inionlength), maximum cranial breadth (maximum bi-parietal diameter) and maximum auricularhead he...

Stomatologija / issued by public institution "Odontologijos studija" ... [et al.]

Rolandas Pletkus , Linas Zaleckas

OBJECTIVES. To evaluate the perpendicularity of the line connecting external auditive meatus to the midsagital plane and the palatal suture as a midsagittal symmetry reference line. Setting and Sample Population - 62 randomly chosen human skulls from osteological collection (Vilnius University). MATERIAL AND METHODS. The skulls were photographed (Nikon 40 D, Nikkor lens 50 mm) from basal, frontobasal and frontal views. Photos were analysed with Adobe Photoshop CS5 (Adobe). The first line connected frontal points of external auditive meatus and the angle to the midsagittal plane was measured. The second line (the palatal suture) was compared to the median sagittal plane. Data was analysed with SPSS 17 (IBM). RESULTS. The mean value for the angles of the line between the external auditive meatus and the midsagittal plane in basal views was 90.12° (SD=1.48°) and in frontobasal 90.36° (SD=2.25°). No statistically significant differences were found between groups of age and sex. The inte...

Robert McCarthy

This study tests several hypotheses of integration between the cranial base and face in primates. After reviewing the definition and anatomical basis for the posterior maxillary (PM) plane, which demarcates the back of the midface at its junction with the sphenoid, we demonstrate how the PM plane can be identified accurately on radiographs, and confirm that it maintains a 90° angle relative to the Neutral Horizontal Axis of the orbits in all primates. In addition, we use the PM plane to test Dabelow’s (1929) hypothesis that the orbits and anterior cranial base are more highly integrated in anthropoids than in strepsirrhines, and we test the hypothesis that the midline anterior cranial base (planum sphenoideum) and anterior cranial floor (planum sphenoideum plus cribriform plate) in primates are highly correlated with each other relative to the PM plane. The mean angle between the anterior cranial base and the PM plane does not differ significantly from 90° in anthropoids, but differs significantly in strepsirrhines. The anterior cranial base and anterior cranial floor, however, correlate well with each other relative to the PM plane in both suborders of primates, independent of orbital orientation and configuration. The PM plane, anterior cranial base, and anterior cranial floor, therefore, form an integrated structural complex, a “facial block,” whose orientation relative to the posterior cranial base influences craniofacial shape among anthropoids in which orbital orientation influences the orientation of the anterior cranial base. One such effect is that increases in cranial base flexion shorten the antero-posterior length of the nasopharynx.

Comparison of cephalometric measurements of living subjects and ancient skulls in Anatolia

Serkan Sahin

Aim: This study aims to provide insights into the evolutionary adaptation of human, via comparing the craniofacial characteristics of living subjects and ancient skulls from Anatolia. Material and Methods: Cone-beam computed tomography (CBCT) generated 2D cephalometric projections of 32 ancient skulls and well matched lateral cephalometric images of 32 patients were evaluated. Sixteen widely used cephalometric measurements were performed. Intra-class correlation coefficients were used to examine intra-observer reliability. Mann-Whitney tests and chi-square tests were used to compare cephalometric measurements of the groups. Results: : The linear measurements of living subjects were smaller than the linear measurements of ancient skulls (p<0.05). Significant differences were found between the groups in the cranial base lengths, maxillary and mandibular dimensions (p<0.05). The maxilla and mandible were found more prognathic in ancient men and women (p<0.001). Conclusion: Environmental factors and genetic changes lead to a reduction in the sagittal and vertical dimensions of the human craniofacial complex. Dental practitioners should consider these evolutionary changes during the treatment planning process.

Oral surgery, oral medicine, and oral pathology

Stephen Schendel

The architectural and structural craniofacial analysis is based upon mutual balance of the cranial and facial bony structures. With this technique, the bases and calvaria of the cranium and then the face can be studied successively in relation to the cranium and craniospinal articulation. Statistical averages are avoided, and individual proportions influenced by the unique features of each skeleton are relied upon. The dentition is placed within the cephalic context, and therapy etiologic factors of dentofacial dysmorphoses which would not otherwise be demonstrated by conventional analysis are made obvious. This technique is particularly useful to the maxillofacial surgeon, as it clearly demonstrates all of the maxillofacial deformities and the pathologic balances that need to be corrected. In severe craniofacial malformations, it offers better possibilities than other cephalometric analysis methods of detecting the various cranial and facial anomalies which characterize these condi...

Lateral cephalometric analysis for treatment planning in orthodontics based on MRI compared with radiographs: A feasibility study in children and adolescents

* E-mail: [email protected]

Affiliation Department of Neuroradiology, Heidelberg University Hospital, Heidelberg, Germany

Affiliation Division of Experimental Radiology, Department of Neuroradiology, Heidelberg University Hospital, Heidelberg, Germany

Affiliation Institute for Medical Biometry and Informatics, Heidelberg University Hospital, Heidelberg, Germany

Affiliation Department of Orthodontics and Dentofacial Orthopaedics, Heidelberg University Hospital, Germany

• Alexander Heil, 
• Eduardo Lazo Gonzalez, 
• Tim Hilgenfeld, 
• Philipp Kickingereder, 
• Martin Bendszus, 
• Sabine Heiland, 
• Ann-Kathrin Ozga, 
• Andreas Sommer, 
• Christopher J. Lux, 
• Sebastian Zingler

• Published: March 23, 2017
• https://doi.org/10.1371/journal.pone.0174524
• Reader Comments

The objective of this prospective study was to evaluate whether magnetic resonance imaging (MRI) is equivalent to lateral cephalometric radiographs (LCR, “gold standard”) in cephalometric analysis.

The applied MRI technique was optimized for short scanning time, high resolution, high contrast and geometric accuracy. Prior to orthodontic treatment, 20 patients (mean age ± SD, 13.95 years ± 5.34) received MRI and LCR. MRI datasets were postprocessed into lateral cephalograms. Cephalometric analysis was performed twice by two independent observers for both modalities with an interval of 4 weeks. Eight bilateral and 10 midsagittal landmarks were identified, and 24 widely used measurements (14 angles, 10 distances) were calculated. Statistical analysis was performed by using intraclass correlation coefficient (ICC), Bland-Altman analysis and two one-sided tests (TOST) within the predefined equivalence margin of ± 2°/mm.

Geometric accuracy of the MRI technique was confirmed by phantom measurements. Mean intraobserver ICC were 0.977/0.975 for MRI and 0.975/0.961 for LCR. Average interobserver ICC were 0.980 for MRI and 0.929 for LCR. Bland-Altman analysis showed high levels of agreement between the two modalities, bias range (mean ± SD) was -0.66 to 0.61 mm (0.06 ± 0.44) for distances and -1.33 to 1.14° (0.06 ± 0.71) for angles. Except for the interincisal angle ( p = 0.17) all measurements were statistically equivalent ( p < 0.05).

Conclusions

This study demonstrates feasibility of orthodontic treatment planning without radiation exposure based on MRI. High-resolution isotropic MRI datasets can be transformed into lateral cephalograms allowing reliable measurements as applied in orthodontic routine with high concordance to the corresponding measurements on LCR.

Citation: Heil A, Lazo Gonzalez E, Hilgenfeld T, Kickingereder P, Bendszus M, Heiland S, et al. (2017) Lateral cephalometric analysis for treatment planning in orthodontics based on MRI compared with radiographs: A feasibility study in children and adolescents. PLoS ONE 12(3): e0174524. https://doi.org/10.1371/journal.pone.0174524

Editor: Christoph Kleinschnitz, Julius-Maximilians-Universität Würzburg, GERMANY

Received: January 24, 2017; Accepted: March 10, 2017; Published: March 23, 2017

Copyright: © 2017 Heil et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data regarding the results are within the paper. The authors confirm that restrictions apply to the data underlying the findings because public availability could compromise patient confidentiality or participant privacy. Anonymized data that support the findings of this study are available from the corresponding author ( [email protected] ) upon reasonable request and with permission of the Heidelberg University Hospital Ethics Committee ( [email protected] ).

Funding: AH, SH, CJL and SZ receive a research grant from the Dietmar Hopp foundation (grant number: 23011228; http://dietmar-hopp-stiftung.de/ ). PK is supported by a postdoctoral fellowship of the University of Heidelberg. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Angular and linear measurements performed on lateral cephalometric radiographs (LCR) play a pivotal role in orthodontic routine diagnostics. Introduced in the 1930s [ 1 ] and further developed over many decades, lateral cephalometric analysis on LCR has remained the standard method in clinical routine until today. By assessing skeletal and dental relationships, it allows diagnosis and monitoring of various growth and development abnormalities [ 2 ]. For example, lateral cephalometric analysis is important for the evaluation of severe skeletal malocclusions and for the planning of orthodontic appliances or orthognathic surgery [ 2 , 3 ]. Radiation protection is of major importance in orthodontics, as the vast majority of patients are children or adolescents and as in most cases a series of radiographs is taken in the course of treatment. Because of the increased lifetime risk for stochastic radiation effects [ 4 – 6 ], it would be desirable to perform imaging in complete absence of ionizing radiation.

As magnetic resonance imaging (MRI) is not associated with radiation exposure and capable to generate geometrically accurate datasets, it may evolve as a promising modality for cephalometric analysis as applied in orthodontics or related disciplines such as orthognathic surgery. Along with recent technical milestones, MRI is moving into focus in dental imaging [ 7 ]. Modern MRI methods can visualize dental and periodontal structures excellently due to increased field strength [ 8 ], parallel imaging techniques [ 9 ] and dedicated coil systems [ 10 – 12 ]. Reasons for the lack of MRI studies in orthodontics might be linked to specific requirements that have to be fulfilled to enable comprehensive and differentiated lateral cephalometric analysis. From the young patients’ perspective, examination time should be as short as possible and the procedure needs to be well-tolerated. Simultaneously, a large field of view is necessary to cover all relevant anatomic landmarks and the generated images must enable clear identification of dental as well as skeletal structures. Finally, image postprocessing should allow the performance of all established measurements required for treatment planning in correspondence to the measurements taken on LCR.

Here, we present an application-optimized, isotropic MRI technique that meets these criteria and a postprocessing algorithm that allows to transform the acquired MRI datasets into lateral cephalograms including the relevant midsagittal and bilateral landmarks. Based on this approach, a prospective in vivo study was performed to compare a series of well-established angular and linear measurements on LCR to those on corresponding MRI derived lateral cephalograms. The null hypothesis of non-equivalence was rejected if the measurements on LCR and MRI were within a low and clinically acceptable tolerance level of ± 2 mm and ± 2°, respectively. The purpose of the study was to evaluate whether MRI can be equivalent to LCR (“gold standard”) in cephalometric analysis.

Materials and methods

Ethics and funding.

This prospective study was approved by the local research ethics committee of the University of Heidelberg (approval number: S-294/2014). Written informed consent was obtained from the patients, in case of minority from their parents as well.

Twenty-one patients with various orthodontic disorders were enrolled in the study before treatment. Exclusion criteria were fixed orthodontic appliances, metal restorations, severe facial asymmetries, missing permanent incisors, no occlusion of either first premolars or second deciduous molars, contraindications to MRI and insufficient image quality of LCR or MRI. One patient had to be excluded because of head rotation around the vertical axis on LCR. Accordingly, 20 patients (8 females) were available for analysis. Mean age ± standard deviation was 13.95 years ± 5.34 (range, 8–26 years).

Lateral cephalometric radiographs

All LCR were acquired using the imaging system Orthopos XG 3D ready Ceph with a CCD line sensor (Sirona Dental Systems, Bensheim, Germany) at 72 kV, 15 mA, an exposure time of 9.4 s and a source-midsagittal plane distance of 1.5 m. Pixel size was 0.027 mm 2 . A 50 mm calibration ruler for magnification correction was integrated in the vertically aligned nose support of the device.

MRI examinations

All MRI examinations were performed at a 3T MRI system (MAGNETOM Trio TIM; Siemens Healthcare, Erlangen, Germany) with a 16-channel multipurpose coil (Variety; Noras MRI products, Hoechberg, Germany). Apart from standard localizer sequences, a T1 weighted, isotropic SPACE ( sampling perfection with application optimized contrasts using different flip angle evolution ) sequence with an examination time of 6:59 min was conducted. This sequence included GRAPPA ( generalized autocalibrating partially parallel acquisitions ) for parallel imaging with an acceleration factor of 2, effective resolution was 0.68 mm 3 . Detailed sequence parameters are shown in Table 1 . The field of view covered all relevant midsagittal and bilateral cephalometric landmarks. Prior to examination of study participants, the applied MRI technique was tested for geometric accuracy using the large ACR MRI Accreditation Phantom. According to the Phantom Test Guidance [ 13 ], seven measurements of known values were taken (1 end-to-end measurement with a known value of 148 mm, 6 diameter measurements each with a known value of 190 mm).

• PPT PowerPoint slide
• PNG larger image
• TIFF original image

https://doi.org/10.1371/journal.pone.0174524.t001

Postprocessing of MRI datasets

Postprocessing of in vivo measurements was performed by two radiologists (ELG and AH, both radiology residents with 3 and 4 years of experience in dental imaging and image postprocessing, respectively). Multiplanar reconstructions (MPR) along the anatomic sagittal plane were acquired from primary MRI datasets. Sagittal MPR were transformed into lateral cephalograms covering the predefined landmarks ( Fig 1 ) with dedicated software (AMIRA-3D v5.4.1; Zuse Institute, Berlin, Germany) as shown in Fig 2 . Total time of postprocessing was approximately 15 minutes per patient.

A total of 10 midsagittal (blue marked) and 8 bilateral (red marked) landmarks were included in cephalometric analysis: S = Sella; N = Nasion; ANS = Anterior nasal spine; PNS = Posterior nasal spine; A = Point A (most concave point of anterior maxilla); B = Point B (most concave point of mandibular symphysis); Is = Incision superius; Ii = Incision inferius; As = Apex superius; Ai = Apex inferius; Pg = Pogonion (most anterior point of mandibular symphysis); Gn = Gnathion (midpoint between Pg and Me); Me = Menton (most inferior point of mandibular symphysis); D = Point D (geometric center of the symphysis); Go = Gonion; tGo = Gonion tangent point (intersection between the mandibular line and the ramus line); Ar = (junction between inferior surface of the cranial base and the posterior border of the ascending rami of the mandible); ppOcc = posterior point of occlusion.

https://doi.org/10.1371/journal.pone.0174524.g001

I = A multiplanar reconstruction (MPR) along the anatomic sagittal plane was acquired from primary magnetic resonance imaging (MRI) datasets. II = The midsagittal plane is coloured in red for better visualization of the workflow. Nine slices containing the landmarks necessary for cephalometric analysis were selected (1). The paired lateral slices were cropped preserving the relevant landmarks on the left (2a) and right (2b) side. The midsagittal plane and the 8 cropped lateral slices were merged into a lateral MRI cephalogram (3). III = Lateral cephalometric analysis was performed on lateral MRI cephalograms and corresponding lateral cephalometric radiographs (LCR) with dedicated software. For each modality two observers placed 10 midsagittal and 8 bilateral landmarks from which 14 angles and 10 distances were calculated automatically by software. Measurements were taken twice with an interval of 4 weeks.

https://doi.org/10.1371/journal.pone.0174524.g002

Cephalometric analysis of LCR and MRI

Lateral cephalometric analysis was performed on LCR and MRI cephalograms in DICOM format using dedicated software for cephalometry (Romexis v4.0.0; Planmeca, Helsinki, Finland). A customized analysis protocol with measurements widely used in orthodontic routine was predefined including Steiner’s analysis [ 14 ], the analysis module of the European board of Orthodontics [ 15 ] and Wits appraisal [ 16 ]. After calibration to the protocol, two independent observers (observer I: AH; observer II: SZ, an orthodontist with 8 years of experience in dental imaging) performed cephalometric analysis twice on each patient for both modalities with an interval of 4 weeks. Observers were blinded to the patients’ identities. All LCR were corrected for magnification with a known 50 mm distance on the calibration ruler. Eight bilateral and 10 midsagittal landmarks were traced ( Fig 1 ). From these landmarks 14 angular and 10 linear measurements ( Table 2 ) were performed automatically.

https://doi.org/10.1371/journal.pone.0174524.t002

Statistical analysis

Statistical analysis was performed with software (R version 3.3.1; R Foundation for Statistical Computing, Vienna, Austria). For all measurements, intra- and interobserver agreement was analyzed by intraclass correlation coefficient (ICC). Bland-Altman analysis was used to assess the agreement between the two modalities [ 17 ] for each type of measurement with average values of the two time points and two investigators. Statistical analysis aimed to test for equivalence between the corresponding measurements on LCR and MRI. In this approach, equivalence can be claimed when the confidence interval of the difference in outcome between the compared groups is within a predetermined equivalence margin that can be justified clinically and scientifically [ 18 ]. Equivalence testing between LCR and MRI was carried by two one-sided tests (TOST) with α = 0.05 and a 1─2α confidence interval [ 19 ], also using average values of the two time points and two investigators. Prior to testing, equivalence margins (± θ) of ± 2 mm and ± 2° were defined, referring to clinically acceptable levels of variance for lateral cephalometric analysis as published before [ 20 , 21 ]. Null hypothesis of TOST was that the two mean values were not equivalent. If the 1–2α confidence interval was completely contained within the ± θ interval, the null hypothesis was rejected and the two datasets were considered equivalent ( p- value < 0.05).

According to the ACR Phantom Test Guidance [ 13 ], all seven measurements performed with the MRI sequence used in the study ( Table 1 ) were congruent with the known values of the ACR Phantom.

Both observers showed very high intraobserver agreement for MRI measurements, average (± SD, range) intraobserver ICC were 0.977 (± 0.019, 0.926–0.996) for observer I and 0.975 (± 0.017, 0.937–0.992) for observer II. Similar intraobserver ICC were observed for the LCR counterparts with mean values (± SD, range) of 0.975 (± 0.016, 0.935–0.997) for observer I and 0.961 (± 0.065, 0.692–0.998) for observer II.

Interobserver agreement was excellent for MRI with an average (± SD, range) ICC of 0.980 (± 0.014, 0.938–0.997). In comparison, interobserver agreement for LCR was also excellent, but moderately lower compared to MRI with an average (± SD, range) ICC of 0.929 (± 0.106, 0.467–0.996). Intraobserver and interobserver ICC for all measurements are shown in Table 3 .

https://doi.org/10.1371/journal.pone.0174524.t003

Bland-Altman analysis revealed high of levels agreement between the two modalities for all measurements, bias range (mean ± SD) was -0.66 to 0.61 mm (0.06 ± 0.44) for linear and -1.33 to 1.14° (0.06 ± 0.71) for angular measurements ( Table 4 ). Exemplary Bland-Altman plots according to Steiner’s analysis [ 14 ] are shown in Fig 3 . At the predefined equivalence margins (± θ) of ± 2 mm / ± 2° statistical equivalence between MRI and LCR was observed in 23 out of 24 measurements ( p < 0.05), only for the interincisal angle (Ui/Li) the null hypothesis of non-equivalence could not be rejected (p = 0.17) ( Table 4 ). This result is in line with the corresponding Bland-Altman analysis, where Ui/Li showed the highest bias (-1.33°) and the widest 95% limits of agreement (-7.22°, 4.56°) of all measurements ( Table 4 , Fig 3 ).

Solid lines represent the mean of all differences (bias), dashed lines represent the 95% limits of agreement. Exemplary measurements according to Steiner’s analysis [ 14 ] are shown in this figure: (a) SNA-angle, (b) SNB-angle, (c) ANB-angle, (d) SND-angle, (e) Ui/NA-angle, (f) Is/NA-distance, (g) Li/NB-angle, (h) Ii/NB-distance, (i) Pg/NB-distance, (j) Ui/Li-angle, (k) SN/OcP-angle and (l) SN/GoGn-angle.

https://doi.org/10.1371/journal.pone.0174524.g003

https://doi.org/10.1371/journal.pone.0174524.t004

In particular in children and adolescents, avoidance of radiation exposure is crucial. In this study, we aimed to show equivalence of MRI to radiographs in lateral cephalometry as a basis for orthodontic treatment planning. To our knowledge, MRI based standardized lateral cephalometric analysis including midsagittal as well as bilateral landmarks has not been evaluated before. An isotropic T1-weighted sequence with excellent contrast, high spatial resolution and short scanning time formed the basis for our new approach. Images yielded from this MRI technique allowed a clear depiction of the dental and skeletal cephalometric landmarks. The subsequent postprocessing algorithm enabled the transformation of the isotropic MRI datasets into lateral cephalograms covering the midsagittal and bilateral landmarks necessary for diagnostics and treatment planning in orthodontics. Based on these generated lateral MRI cephalograms it was possible to perform a detailed cephalometric analysis with a broad spectrum of measurements as used in orthodontic routine. Linear and angular cephalometric measurements taken on lateral MRI cephalograms turned out to be highly reliable as interobserver and intraobserver agreement was excellent. As a principal finding, we found high levels of agreement between the measurements on lateral MRI cephalograms and the corresponding measurements on LCR in a clinical environment by examining young patients with various orthodontic abnormalities. Statistical equivalence between the two modalities was shown for 23 out of 24 measured distances and angles within a strict predefined equivalence margin of ± 2 mm / ± 2°. The only measurement without statistical equivalence was the interincisal angle, which also showed a slightly higher bias level in Bland-Altman analysis in comparison to the other cephalometric measurements. This, however, was not an unexpected finding, as the interincisal angle is prone to measurement errors when performed on LCR [ 22 , 23 ]. Nonetheless, the mean difference of -1.33° in Bland-Altman analysis still indicated a low and clinically tolerable bias for the interincisal angle. Considering the overall high concordance with LCR (“gold standard”) and the absence of radiation exposure, lateral cephalometric analysis for the assessment and monitoring of orthodontic conditions could be performed by MRI in the future to keep radiation dose in young patients as low as possible.

Even though mean differences between LCR and MRI were generally low, they should be analyzed thoroughly. As intra- and interobserver reliability were consistently high for both modalities, the slight differences were presumably due to systematic errors. Like all radiographic techniques, LCR are accompanied by distortion and magnification [ 22 , 24 – 26 ]. As we proved geometric accuracy for the applied MRI technique by standardized phantom measurements, it is most likely that the slight differences for angular and linear measurements predominantly derived from LCR. Considering that studies comparing conventional computed tomography (CT) or cone-beam computed tomography (CBCT) to LCR showed very similar differences in lateral cephalometric measurements [ 21 , 27 , 28 ] and that CT-techniques are geometrically accurate under normal conditions [ 29 ], it is legitimate to compare these results to ours. The hypothesis that intrinsic limitations of LCR were the main error source in the present study is strongly supported by ex vivo studies, which showed very high concordance between measurements on MRI and CT [ 30 ] or MRI and CBCT [ 31 ] [ 32 ].

An essential element of our feasibility study was a MRI technique with the potential to become a routine application for orthodontic treatment planning. It should be highlighted that we were able to provide a short protocol which was well-tolerated by the children and adolescents who participated in the study. Including patient positioning and planning on standard localizer sequences, the MRI examinations were performed within a total time of about 10 minutes leading to high compliance without relevant motion artifacts.

Our study aimed to compare MRI with LCR due to high relevance in orthodontic routine. However, potential capabilities of the applied MRI technique are not restricted to lateral cephalometry. The second important radiographic image tool in orthodontics are panoramic radiographs (PR), typically used for evaluation of dental development, unerupted or supernumerary teeth and alveolar bone morphology [ 33 ]. As of principle, such analyses are also feasible on MRI datasets as acquired in our study. If future studies showed equivalence between MRI and PR, the latter could be avoided providing the possibility of orthodontic imaging without any radiation exposure. Furthermore, isotropic MRI datasets have the potential to perform three-dimensional (3D) cephalometric analysis, which might lead to more differentiated and conclusive diagnoses compared to two-dimensional radiographs. Several approaches for 3D cephalometry have been made based on CT and CBCT, but reliable procedures could not be established due to the lack of comparative norms [ 34 ]. By contrast, non-ionizing MRI provides the possibility to establish proper standards of 3D cephalometry, as the whole spectrum of orthodontic conditions including normal collectives and patients with slight malconditions could be analyzed. Another advantage of MRI over X-ray methods is the visualization of soft tissues. This a key point for future studies, as there are no objective methods to monitor changes in soft tissues under therapy [ 35 ].

A limitation of the present study was that the true values of the cephalometric measurements were not known. Even though lateral cephalometry on LCR is the “gold standard”, it is prone to measurement errors as described above and therefore should not be used as a reference standard in a diagnostic accuracy study. Thus, accuracy for MRI can only be claimed for the phantom measurements, but not for in vivo data.

A further limitation was that MRI datasets had to be postprocessed to generate the lateral cephalograms necessary for data analysis. Specific postprocessing software was required and the algorithm could only be performed with sufficient user experience. However, this limitation is not surprising regarding the framework of a feasibility study aiming at introducing this new approach of MRI based cephalometric analysis. As a next step, we suggest the implementation of software solutions allowing user-friendly and time-efficient postprocessing of primary MRI datasets into lateral cephalograms. Ideally, only sagittal MPR and selection of slices with the relevant landmarks will have to be performed by the user in such applications. All subsequent steps to the final lateral cephalogram could then be computed fully automated without user interaction. Furthermore, future software for MRI based cephalometric analysis should be integrated into existing standard software to facilitate broad application in clinical routine.

In conclusion, this study shows that full lateral cephalometric analysis as applied in orthodontics is feasible based on postprocessed MRI datasets. There was a high concordance with equivalent measurements taken on LCR, which is the standard method in clinical routine. Our MRI based approach for the first time enables the assessment of orthodontic conditions by using clinically standardized analysis methods in absence of radiation exposure to the mostly young patients. The short and well-tolerated examination protocol applied in our feasibility study could be integrated into clinical routine. Further studies with large patient populations using different MRI systems should be conducted to support our findings and to evaluate whether MRI and LCR are equivalent in lateral cephalometric analysis under the most diverse clinical and technical conditions. Moreover, our MRI technique has the potential to overcome the limitations of projection radiography in the future.

Acknowledgments

AH, SZ, SH and CJL wish to acknowledge the support by the Dietmar Hopp Foundation. Furthermore, the authors kindly thank NORAS MRI products GmbH (Höchberg, Germany) for providing the 16-channel multipurpose coil.

Author Contributions

• Conceptualization: AH MB SH CJL SZ.
• Data curation: AH ELG PK AO.
• Formal analysis: AH PK AO.
• Funding acquisition: AH SH CJL SZ.
• Investigation: AH ELG SH AS SZ.
• Methodology: AH ELG MB SH CJL SZ.
• Project administration: AH ELG SH AS SZ.
• Resources: AH ELG TH SH AS SZ.
• Software: AH ELG TH.
• Supervision: MB SH CJL.
• Validation: PK AO MB CJL.
• Visualization: AH.
• Writing – original draft: AH SZ.
• Writing – review & editing: AH MB SH CJL TH PK SZ.
• View Article
• Google Scholar
• 2. American Association of Orthodontists. Clinical Practice Guidelines for Orthodontics and Dentofacial Orthopedics 2014. Available from: https://www.aaoinfo.org/system/files/media/documents/2014 Cllinical Practice Guidelines.pdf .
• 3. British Orthodontic Society. Guidelines for the Use of Radiographs in Clinical Orthodontics 2015. Available from: http://www.bos.org.uk/Portals/0/Public/docs/General Guidance/Orthodontic Radiographs 2016—2.pdf .
• PubMed/NCBI
• 13. The American College of Radiology. Phantom Test Guidance for the ACR MRI Accreditation Program 2005. Available from: http://www.acraccreditation.org/~/media/ACRAccreditation/Documents/MRI/LargePhantomGuidance.pdf?la=en .
• 15. The European Board of Orthodontists. An illustrated guide to prepare for the examination 2005. Available from: http://www.eoseurope.org/documents/instructions/EBO05.pdf .
• 33. American Dental Association. Dental radiographic examinations: recommendations for patient selection and limiting radiation exposure 2012. Available from: http://www.fda.gov/Radiation-EmittingProducts/RadiationEmittingProductsandProcedures/MedicalImaging/MedicalX-Rays/ucm116504.htm .

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List

Lateral cephalometric analysis for treatment planning in orthodontics based on MRI compared with radiographs: A feasibility study in children and adolescents

Alexander heil.

1 Department of Neuroradiology, Heidelberg University Hospital, Heidelberg, Germany

Eduardo Lazo Gonzalez

Tim hilgenfeld, philipp kickingereder, martin bendszus, sabine heiland.

2 Division of Experimental Radiology, Department of Neuroradiology, Heidelberg University Hospital, Heidelberg, Germany

Ann-Kathrin Ozga

3 Institute for Medical Biometry and Informatics, Heidelberg University Hospital, Heidelberg, Germany

Andreas Sommer

4 Department of Orthodontics and Dentofacial Orthopaedics, Heidelberg University Hospital, Germany

Christopher J. Lux

Sebastian zingler.

• Conceptualization: AH MB SH CJL SZ.
• Data curation: AH ELG PK AO.
• Formal analysis: AH PK AO.
• Funding acquisition: AH SH CJL SZ.
• Investigation: AH ELG SH AS SZ.
• Methodology: AH ELG MB SH CJL SZ.
• Project administration: AH ELG SH AS SZ.
• Resources: AH ELG TH SH AS SZ.
• Software: AH ELG TH.
• Supervision: MB SH CJL.
• Validation: PK AO MB CJL.
• Visualization: AH.
• Writing – original draft: AH SZ.
• Writing – review & editing: AH MB SH CJL TH PK SZ.

Associated Data

All relevant data regarding the results are within the paper. The authors confirm that restrictions apply to the data underlying the findings because public availability could compromise patient confidentiality or participant privacy. Anonymized data that support the findings of this study are available from the corresponding author ( [email protected] ) upon reasonable request and with permission of the Heidelberg University Hospital Ethics Committee ( ed.grebledieh-inu.dem@I-noissimmokkihte ).

The objective of this prospective study was to evaluate whether magnetic resonance imaging (MRI) is equivalent to lateral cephalometric radiographs (LCR, “gold standard”) in cephalometric analysis.

The applied MRI technique was optimized for short scanning time, high resolution, high contrast and geometric accuracy. Prior to orthodontic treatment, 20 patients (mean age ± SD, 13.95 years ± 5.34) received MRI and LCR. MRI datasets were postprocessed into lateral cephalograms. Cephalometric analysis was performed twice by two independent observers for both modalities with an interval of 4 weeks. Eight bilateral and 10 midsagittal landmarks were identified, and 24 widely used measurements (14 angles, 10 distances) were calculated. Statistical analysis was performed by using intraclass correlation coefficient (ICC), Bland-Altman analysis and two one-sided tests (TOST) within the predefined equivalence margin of ± 2°/mm.

Geometric accuracy of the MRI technique was confirmed by phantom measurements. Mean intraobserver ICC were 0.977/0.975 for MRI and 0.975/0.961 for LCR. Average interobserver ICC were 0.980 for MRI and 0.929 for LCR. Bland-Altman analysis showed high levels of agreement between the two modalities, bias range (mean ± SD) was -0.66 to 0.61 mm (0.06 ± 0.44) for distances and -1.33 to 1.14° (0.06 ± 0.71) for angles. Except for the interincisal angle ( p = 0.17) all measurements were statistically equivalent ( p < 0.05).

Conclusions

This study demonstrates feasibility of orthodontic treatment planning without radiation exposure based on MRI. High-resolution isotropic MRI datasets can be transformed into lateral cephalograms allowing reliable measurements as applied in orthodontic routine with high concordance to the corresponding measurements on LCR.

Introduction

Angular and linear measurements performed on lateral cephalometric radiographs (LCR) play a pivotal role in orthodontic routine diagnostics. Introduced in the 1930s [ 1 ] and further developed over many decades, lateral cephalometric analysis on LCR has remained the standard method in clinical routine until today. By assessing skeletal and dental relationships, it allows diagnosis and monitoring of various growth and development abnormalities [ 2 ]. For example, lateral cephalometric analysis is important for the evaluation of severe skeletal malocclusions and for the planning of orthodontic appliances or orthognathic surgery [ 2 , 3 ]. Radiation protection is of major importance in orthodontics, as the vast majority of patients are children or adolescents and as in most cases a series of radiographs is taken in the course of treatment. Because of the increased lifetime risk for stochastic radiation effects [ 4 – 6 ], it would be desirable to perform imaging in complete absence of ionizing radiation.

As magnetic resonance imaging (MRI) is not associated with radiation exposure and capable to generate geometrically accurate datasets, it may evolve as a promising modality for cephalometric analysis as applied in orthodontics or related disciplines such as orthognathic surgery. Along with recent technical milestones, MRI is moving into focus in dental imaging [ 7 ]. Modern MRI methods can visualize dental and periodontal structures excellently due to increased field strength [ 8 ], parallel imaging techniques [ 9 ] and dedicated coil systems [ 10 – 12 ]. Reasons for the lack of MRI studies in orthodontics might be linked to specific requirements that have to be fulfilled to enable comprehensive and differentiated lateral cephalometric analysis. From the young patients’ perspective, examination time should be as short as possible and the procedure needs to be well-tolerated. Simultaneously, a large field of view is necessary to cover all relevant anatomic landmarks and the generated images must enable clear identification of dental as well as skeletal structures. Finally, image postprocessing should allow the performance of all established measurements required for treatment planning in correspondence to the measurements taken on LCR.

Here, we present an application-optimized, isotropic MRI technique that meets these criteria and a postprocessing algorithm that allows to transform the acquired MRI datasets into lateral cephalograms including the relevant midsagittal and bilateral landmarks. Based on this approach, a prospective in vivo study was performed to compare a series of well-established angular and linear measurements on LCR to those on corresponding MRI derived lateral cephalograms. The null hypothesis of non-equivalence was rejected if the measurements on LCR and MRI were within a low and clinically acceptable tolerance level of ± 2 mm and ± 2°, respectively. The purpose of the study was to evaluate whether MRI can be equivalent to LCR (“gold standard”) in cephalometric analysis.

Materials and methods

Ethics and funding.

This prospective study was approved by the local research ethics committee of the University of Heidelberg (approval number: S-294/2014). Written informed consent was obtained from the patients, in case of minority from their parents as well.

Twenty-one patients with various orthodontic disorders were enrolled in the study before treatment. Exclusion criteria were fixed orthodontic appliances, metal restorations, severe facial asymmetries, missing permanent incisors, no occlusion of either first premolars or second deciduous molars, contraindications to MRI and insufficient image quality of LCR or MRI. One patient had to be excluded because of head rotation around the vertical axis on LCR. Accordingly, 20 patients (8 females) were available for analysis. Mean age ± standard deviation was 13.95 years ± 5.34 (range, 8–26 years).

Lateral cephalometric radiographs

All LCR were acquired using the imaging system Orthopos XG 3D ready Ceph with a CCD line sensor (Sirona Dental Systems, Bensheim, Germany) at 72 kV, 15 mA, an exposure time of 9.4 s and a source-midsagittal plane distance of 1.5 m. Pixel size was 0.027 mm 2 . A 50 mm calibration ruler for magnification correction was integrated in the vertically aligned nose support of the device.

MRI examinations

All MRI examinations were performed at a 3T MRI system (MAGNETOM Trio TIM; Siemens Healthcare, Erlangen, Germany) with a 16-channel multipurpose coil (Variety; Noras MRI products, Hoechberg, Germany). Apart from standard localizer sequences, a T1 weighted, isotropic SPACE ( sampling perfection with application optimized contrasts using different flip angle evolution ) sequence with an examination time of 6:59 min was conducted. This sequence included GRAPPA ( generalized autocalibrating partially parallel acquisitions ) for parallel imaging with an acceleration factor of 2, effective resolution was 0.68 mm 3 . Detailed sequence parameters are shown in Table 1 . The field of view covered all relevant midsagittal and bilateral cephalometric landmarks. Prior to examination of study participants, the applied MRI technique was tested for geometric accuracy using the large ACR MRI Accreditation Phantom. According to the Phantom Test Guidance [ 13 ], seven measurements of known values were taken (1 end-to-end measurement with a known value of 148 mm, 6 diameter measurements each with a known value of 190 mm).

SPACE = sampling perfection with application optimized contrasts using different flip angle evolution, GRAPPA = generalized autocalibrating partially parallel acquisitions.

Postprocessing of MRI datasets

Postprocessing of in vivo measurements was performed by two radiologists (ELG and AH, both radiology residents with 3 and 4 years of experience in dental imaging and image postprocessing, respectively). Multiplanar reconstructions (MPR) along the anatomic sagittal plane were acquired from primary MRI datasets. Sagittal MPR were transformed into lateral cephalograms covering the predefined landmarks ( Fig 1 ) with dedicated software (AMIRA-3D v5.4.1; Zuse Institute, Berlin, Germany) as shown in Fig 2 . Total time of postprocessing was approximately 15 minutes per patient.

A total of 10 midsagittal (blue marked) and 8 bilateral (red marked) landmarks were included in cephalometric analysis: S = Sella; N = Nasion; ANS = Anterior nasal spine; PNS = Posterior nasal spine; A = Point A (most concave point of anterior maxilla); B = Point B (most concave point of mandibular symphysis); Is = Incision superius; Ii = Incision inferius; As = Apex superius; Ai = Apex inferius; Pg = Pogonion (most anterior point of mandibular symphysis); Gn = Gnathion (midpoint between Pg and Me); Me = Menton (most inferior point of mandibular symphysis); D = Point D (geometric center of the symphysis); Go = Gonion; tGo = Gonion tangent point (intersection between the mandibular line and the ramus line); Ar = (junction between inferior surface of the cranial base and the posterior border of the ascending rami of the mandible); ppOcc = posterior point of occlusion.

I = A multiplanar reconstruction (MPR) along the anatomic sagittal plane was acquired from primary magnetic resonance imaging (MRI) datasets. II = The midsagittal plane is coloured in red for better visualization of the workflow. Nine slices containing the landmarks necessary for cephalometric analysis were selected (1). The paired lateral slices were cropped preserving the relevant landmarks on the left (2a) and right (2b) side. The midsagittal plane and the 8 cropped lateral slices were merged into a lateral MRI cephalogram (3). III = Lateral cephalometric analysis was performed on lateral MRI cephalograms and corresponding lateral cephalometric radiographs (LCR) with dedicated software. For each modality two observers placed 10 midsagittal and 8 bilateral landmarks from which 14 angles and 10 distances were calculated automatically by software. Measurements were taken twice with an interval of 4 weeks.

Cephalometric analysis of LCR and MRI

Lateral cephalometric analysis was performed on LCR and MRI cephalograms in DICOM format using dedicated software for cephalometry (Romexis v4.0.0; Planmeca, Helsinki, Finland). A customized analysis protocol with measurements widely used in orthodontic routine was predefined including Steiner’s analysis [ 14 ], the analysis module of the European board of Orthodontics [ 15 ] and Wits appraisal [ 16 ]. After calibration to the protocol, two independent observers (observer I: AH; observer II: SZ, an orthodontist with 8 years of experience in dental imaging) performed cephalometric analysis twice on each patient for both modalities with an interval of 4 weeks. Observers were blinded to the patients’ identities. All LCR were corrected for magnification with a known 50 mm distance on the calibration ruler. Eight bilateral and 10 midsagittal landmarks were traced ( Fig 1 ). From these landmarks 14 angular and 10 linear measurements ( Table 2 ) were performed automatically.

Abbreviations for cephalometric landmarks are explained in the footnote of Fig 1 .

* OcP was defined as the line passing through the midpoint between the incisal edges (anterior) and the most distal point of contact of either the first permanent or second deciduous molars (posterior).

Statistical analysis

Statistical analysis was performed with software (R version 3.3.1; R Foundation for Statistical Computing, Vienna, Austria). For all measurements, intra- and interobserver agreement was analyzed by intraclass correlation coefficient (ICC). Bland-Altman analysis was used to assess the agreement between the two modalities [ 17 ] for each type of measurement with average values of the two time points and two investigators. Statistical analysis aimed to test for equivalence between the corresponding measurements on LCR and MRI. In this approach, equivalence can be claimed when the confidence interval of the difference in outcome between the compared groups is within a predetermined equivalence margin that can be justified clinically and scientifically [ 18 ]. Equivalence testing between LCR and MRI was carried by two one-sided tests (TOST) with α = 0.05 and a 1─2α confidence interval [ 19 ], also using average values of the two time points and two investigators. Prior to testing, equivalence margins (± θ) of ± 2 mm and ± 2° were defined, referring to clinically acceptable levels of variance for lateral cephalometric analysis as published before [ 20 , 21 ]. Null hypothesis of TOST was that the two mean values were not equivalent. If the 1–2α confidence interval was completely contained within the ± θ interval, the null hypothesis was rejected and the two datasets were considered equivalent ( p- value < 0.05).

According to the ACR Phantom Test Guidance [ 13 ], all seven measurements performed with the MRI sequence used in the study ( Table 1 ) were congruent with the known values of the ACR Phantom.

Both observers showed very high intraobserver agreement for MRI measurements, average (± SD, range) intraobserver ICC were 0.977 (± 0.019, 0.926–0.996) for observer I and 0.975 (± 0.017, 0.937–0.992) for observer II. Similar intraobserver ICC were observed for the LCR counterparts with mean values (± SD, range) of 0.975 (± 0.016, 0.935–0.997) for observer I and 0.961 (± 0.065, 0.692–0.998) for observer II.

Interobserver agreement was excellent for MRI with an average (± SD, range) ICC of 0.980 (± 0.014, 0.938–0.997). In comparison, interobserver agreement for LCR was also excellent, but moderately lower compared to MRI with an average (± SD, range) ICC of 0.929 (± 0.106, 0.467–0.996). Intraobserver and interobserver ICC for all measurements are shown in Table 3 .

ICC = Intraclass correlation coefficient; Obs. = Observer

Bland-Altman analysis revealed high of levels agreement between the two modalities for all measurements, bias range (mean ± SD) was -0.66 to 0.61 mm (0.06 ± 0.44) for linear and -1.33 to 1.14° (0.06 ± 0.71) for angular measurements ( Table 4 ). Exemplary Bland-Altman plots according to Steiner’s analysis [ 14 ] are shown in Fig 3 . At the predefined equivalence margins (± θ) of ± 2 mm / ± 2° statistical equivalence between MRI and LCR was observed in 23 out of 24 measurements ( p < 0.05), only for the interincisal angle (Ui/Li) the null hypothesis of non-equivalence could not be rejected (p = 0.17) ( Table 4 ). This result is in line with the corresponding Bland-Altman analysis, where Ui/Li showed the highest bias (-1.33°) and the widest 95% limits of agreement (-7.22°, 4.56°) of all measurements ( Table 4 , Fig 3 ).

Solid lines represent the mean of all differences (bias), dashed lines represent the 95% limits of agreement. Exemplary measurements according to Steiner’s analysis [ 14 ] are shown in this figure: (a) SNA-angle, (b) SNB-angle, (c) ANB-angle, (d) SND-angle, (e) Ui/NA-angle, (f) Is/NA-distance, (g) Li/NB-angle, (h) Ii/NB-distance, (i) Pg/NB-distance, (j) Ui/Li-angle, (k) SN/OcP-angle and (l) SN/GoGn-angle.

a Values are means ± standard deviations of measurements of two time points and two investigators.

b All p -values refer to two one-sided tests (TOST) of equivalence with predefined equivalence margins (± θ) of ± 2° and ± 2 mm, respectively. p- values < 0.05 indicate statistical equivalence.

In particular in children and adolescents, avoidance of radiation exposure is crucial. In this study, we aimed to show equivalence of MRI to radiographs in lateral cephalometry as a basis for orthodontic treatment planning. To our knowledge, MRI based standardized lateral cephalometric analysis including midsagittal as well as bilateral landmarks has not been evaluated before. An isotropic T1-weighted sequence with excellent contrast, high spatial resolution and short scanning time formed the basis for our new approach. Images yielded from this MRI technique allowed a clear depiction of the dental and skeletal cephalometric landmarks. The subsequent postprocessing algorithm enabled the transformation of the isotropic MRI datasets into lateral cephalograms covering the midsagittal and bilateral landmarks necessary for diagnostics and treatment planning in orthodontics. Based on these generated lateral MRI cephalograms it was possible to perform a detailed cephalometric analysis with a broad spectrum of measurements as used in orthodontic routine. Linear and angular cephalometric measurements taken on lateral MRI cephalograms turned out to be highly reliable as interobserver and intraobserver agreement was excellent. As a principal finding, we found high levels of agreement between the measurements on lateral MRI cephalograms and the corresponding measurements on LCR in a clinical environment by examining young patients with various orthodontic abnormalities. Statistical equivalence between the two modalities was shown for 23 out of 24 measured distances and angles within a strict predefined equivalence margin of ± 2 mm / ± 2°. The only measurement without statistical equivalence was the interincisal angle, which also showed a slightly higher bias level in Bland-Altman analysis in comparison to the other cephalometric measurements. This, however, was not an unexpected finding, as the interincisal angle is prone to measurement errors when performed on LCR [ 22 , 23 ]. Nonetheless, the mean difference of -1.33° in Bland-Altman analysis still indicated a low and clinically tolerable bias for the interincisal angle. Considering the overall high concordance with LCR (“gold standard”) and the absence of radiation exposure, lateral cephalometric analysis for the assessment and monitoring of orthodontic conditions could be performed by MRI in the future to keep radiation dose in young patients as low as possible.

Even though mean differences between LCR and MRI were generally low, they should be analyzed thoroughly. As intra- and interobserver reliability were consistently high for both modalities, the slight differences were presumably due to systematic errors. Like all radiographic techniques, LCR are accompanied by distortion and magnification [ 22 , 24 – 26 ]. As we proved geometric accuracy for the applied MRI technique by standardized phantom measurements, it is most likely that the slight differences for angular and linear measurements predominantly derived from LCR. Considering that studies comparing conventional computed tomography (CT) or cone-beam computed tomography (CBCT) to LCR showed very similar differences in lateral cephalometric measurements [ 21 , 27 , 28 ] and that CT-techniques are geometrically accurate under normal conditions [ 29 ], it is legitimate to compare these results to ours. The hypothesis that intrinsic limitations of LCR were the main error source in the present study is strongly supported by ex vivo studies, which showed very high concordance between measurements on MRI and CT [ 30 ] or MRI and CBCT [ 31 ] [ 32 ].

An essential element of our feasibility study was a MRI technique with the potential to become a routine application for orthodontic treatment planning. It should be highlighted that we were able to provide a short protocol which was well-tolerated by the children and adolescents who participated in the study. Including patient positioning and planning on standard localizer sequences, the MRI examinations were performed within a total time of about 10 minutes leading to high compliance without relevant motion artifacts.

Our study aimed to compare MRI with LCR due to high relevance in orthodontic routine. However, potential capabilities of the applied MRI technique are not restricted to lateral cephalometry. The second important radiographic image tool in orthodontics are panoramic radiographs (PR), typically used for evaluation of dental development, unerupted or supernumerary teeth and alveolar bone morphology [ 33 ]. As of principle, such analyses are also feasible on MRI datasets as acquired in our study. If future studies showed equivalence between MRI and PR, the latter could be avoided providing the possibility of orthodontic imaging without any radiation exposure. Furthermore, isotropic MRI datasets have the potential to perform three-dimensional (3D) cephalometric analysis, which might lead to more differentiated and conclusive diagnoses compared to two-dimensional radiographs. Several approaches for 3D cephalometry have been made based on CT and CBCT, but reliable procedures could not be established due to the lack of comparative norms [ 34 ]. By contrast, non-ionizing MRI provides the possibility to establish proper standards of 3D cephalometry, as the whole spectrum of orthodontic conditions including normal collectives and patients with slight malconditions could be analyzed. Another advantage of MRI over X-ray methods is the visualization of soft tissues. This a key point for future studies, as there are no objective methods to monitor changes in soft tissues under therapy [ 35 ].

A limitation of the present study was that the true values of the cephalometric measurements were not known. Even though lateral cephalometry on LCR is the “gold standard”, it is prone to measurement errors as described above and therefore should not be used as a reference standard in a diagnostic accuracy study. Thus, accuracy for MRI can only be claimed for the phantom measurements, but not for in vivo data.

A further limitation was that MRI datasets had to be postprocessed to generate the lateral cephalograms necessary for data analysis. Specific postprocessing software was required and the algorithm could only be performed with sufficient user experience. However, this limitation is not surprising regarding the framework of a feasibility study aiming at introducing this new approach of MRI based cephalometric analysis. As a next step, we suggest the implementation of software solutions allowing user-friendly and time-efficient postprocessing of primary MRI datasets into lateral cephalograms. Ideally, only sagittal MPR and selection of slices with the relevant landmarks will have to be performed by the user in such applications. All subsequent steps to the final lateral cephalogram could then be computed fully automated without user interaction. Furthermore, future software for MRI based cephalometric analysis should be integrated into existing standard software to facilitate broad application in clinical routine.

In conclusion, this study shows that full lateral cephalometric analysis as applied in orthodontics is feasible based on postprocessed MRI datasets. There was a high concordance with equivalent measurements taken on LCR, which is the standard method in clinical routine. Our MRI based approach for the first time enables the assessment of orthodontic conditions by using clinically standardized analysis methods in absence of radiation exposure to the mostly young patients. The short and well-tolerated examination protocol applied in our feasibility study could be integrated into clinical routine. Further studies with large patient populations using different MRI systems should be conducted to support our findings and to evaluate whether MRI and LCR are equivalent in lateral cephalometric analysis under the most diverse clinical and technical conditions. Moreover, our MRI technique has the potential to overcome the limitations of projection radiography in the future.

Acknowledgments

AH, SZ, SH and CJL wish to acknowledge the support by the Dietmar Hopp Foundation. Furthermore, the authors kindly thank NORAS MRI products GmbH (Höchberg, Germany) for providing the 16-channel multipurpose coil.

Funding Statement

AH, SH, CJL and SZ receive a research grant from the Dietmar Hopp foundation (grant number: 23011228; http://dietmar-hopp-stiftung.de/ ). PK is supported by a postdoctoral fellowship of the University of Heidelberg. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data Availability