Abstract
Conventional lateral and posteroanterior (PA) cephalometric radiographs are important modalities used in orthodontics and oral maxillofacial surgery to evaluate bone and soft-tissue morphology. However, they often suffer from structural superposition and inadequate head positioning during the scan, which limits their diagnostic utility and affects treatment decisions. In this work, we present a new method for generating digital cephalometric radiographs from three-dimensional (3D) images acquired with a dental CBCT system. With the proposed method, both digital lateral and PA cephalometric radiographs can be generated from 3D images obtained with a single CBCT scan. Our preliminary results reveal that the proposed method can yield cephalo- metric radiographs with improved contrast and landmarks of clinical relevance.
Introduction
In orthodontics and oral maxillofacial surgery, cephalometric radiographs, including lateral and posteroanterior (PA) cephalometric radiographs, are valuable tools for providing information of bone and soft-tissue morphology, which is crucial for guiding treatment plans and evaluating treatment progress and outcomes [1–4]. However, the two-dimensional (2D) nature of radiography can result in undesirable superpo- sition of anatomical structures, potentially obscuring detailed features such as tooth root canals and complicating diagnostic decision-making [5–7]. Additionally, the quality and thus utility of conventional cephalometric radiographs are often compromised by improper patient head positioning [8], [9]. Even slight deviations from the optimal position can introduce errors in angular and linear measurements of landmarks, thus reducing the diagnostic accuracy and effectiveness of treatment decisions.
Cone-beam computed tomography (CBCT) has been adopted widely for yielding three-dimensional (3D) images of dental structures, which are valuable for orthodontic diagnosis [10–15]. A 2D digital lateral or PA cephalometric radiograph can be generated from 3D CBCT images using a clinically pecific cephalometric projection geometry. An important advantage of this method is its flexibility: projec- tion directions can be selected according to clinical needs, and portions of the head image can be selected for generation of its projection for revealing detailed structures within the portion. Equally importantly, it avoids positioning errors and overlap issues that occur in conventional cephalometric radiography. In this work, we propose and evaluate a new method for generating digital lateral and PA cephalometric radiographs from 3D CBCT images. This method allows for the selection of a portion of the head image and for yielding its lateral cephalometric radiographs with enhanced precision and utility.
Materials and Methods
- Dental imaging system
Both dental CBCT and cephalometric scans are performed with Bondream 3D 1030 system (Bondent Technology Co Ltd, China). The setting for CBCT and cephalometric scans are 8 mA, 90 kV, 16 second exposure time and 8 mA, 90 kV, 14 second exposure time, respectively.
- Generation of digital cepkalometric radiograpks from 3D CBCT images
3D CBCT images: From the collected CBCT projection data, 3D images are reconstructed by use of the FDK algorithm [16]. The image array size is 600×600×600 with a pixel size of 0.3 mm×0.3 mm×0.3 mm, resulting a standard head FOV of 18 cm. A threshold is applied to the reconstructed 3D images to remove background noise and artifacts. To further enhance the image quality, a median filter is applied to the 3D images to reduce image noise, followed by an edge enhancement with a canny filter.
Generation of digital cepkalometric radiograpks: From these processed 3D images, key anatomical structures including airways, dental arch, and mandible bones are identified in 3D CBCT images. By analyzing the shapes and spatial locations of these structures, a set of linear and angular parameters is derived to quantitatively assess the head's orientation. This assessment is critical as it directly influences the selection of the optimal projection direction for generating cephalometric radiographs, which varies from patient to patient due to the positioning variation of each CBCT scan. Once the optimal projection direction is determined, a parallel-beam projection is used to generate the digital cephalometric radiographs. This method is chosen over cone-beam projection because it can provide better accuracy for the cephalometric measurements. With the parallel-beam projection, the projection paths with various lengths can be selected to yield digital cephalometric radiographs, which enables the generation of lateral cephalo- metric radiographs from left, right, and whole volumes as shown in top row of Figure 1; and of the PA cephalometric radiographs from front and whole volumes as shown in bottom.row of Figure 1. Finally, a contrast-limited adaptive histogram equalization (CLAHE) procedure is applied to the generated cephalometric radiographs to further enhance the image contrast.
Figure 1: Top row: parallel projection for generating lateral cephalometric radiographs from left volume (a), right volume (b), and whole volume (c). Bottom row: parallel projection for generating PA cephalometric radiographs from front volume (d) and whole volume (e). The red line segments indicate the projection paths used in the generation of cephalometric radiographs.
Results
In this study, we evaluate a total of 30 cases with the age range of 14-39 years and present two examples here. For patient 1, the lateral cephalometric radiographs generated from left, right, and whole volumes with the projection path shown in top row of Figure 1 are displayed in Figures 2a-2c, respective- ly. For comparison, the conventional lateral cephalometric radiograph is shown in Figure 2d. It can be observed that the lateral cephalometric radiography generated from whole 3D CBCT images (Figure 2c) resembles the conventional lateral cephalometric radiography (Figure 2d). For both images, key anatomical structures such as the soft tissue boundary, nasal bone, maxilla, mandible, and the sella turcica are clearly visible. However, there is a noticeable misalignment between the inferior borders of the left and right mandibles in the conventional lateral cephalometric radiography, indicating a potential positioning error. Conversely, the synthetic lateral cephalometric radiographs derived from whole CBCT images show proper alignment, suggesting that our method can project CBCT images along the optimal direction to minimize positioning errors, thereby providing more reliable and accurate depictions of anatomical measurements. Additional- ly, the frontonasal suture, highlighted by the red arrow in synthetic cephalometric radiographs, is more clearly discern- ible than that in the conventional lateral cephalometric radiograph.
Figure 2. Lateral cephalometric radiographs generated from left volume (a), right volume (b), and whole volume (c) formed by CBCT images and the corresponding conventional cephalometric radiograph (d) for patient 1. Row 1: Full-view cephalo- metric radiographs. Row 2: Zoomed-in view of the corresponding cephalometric radiographs in Row 1.
Furthermore, we select a region-of-interest (ROI) containing all teeth, indicated by a red rectangle in the top-left panel of Figure 2, and display the corresponding zoomed-in views in row 2 of Figure 2. Due to the removal of the teeth overlap in the Figures 2a and 2b, the contrast of the teeth to the surround- ing maxillary and mandibular bones is enhanced compared to those in Figures 2c and 2d. The root canals can also be identi- fied in Figures 2a and 2b, whereas they are hardly visible in Figures 2c and 2d.
Similarly, for patient 2, we display lateral cephalometric radiographs derived from left, right, and whole CBCT volumes in Figures 3a-3c, respectively, along with the conventional lateral cephalometric radiograph in Figure 3d. Due to the positioning error during the acquisition of conventional lateral cephalometric radiograph, the alignment and contrast of the teeth and their root canals are suboptimal compared to those in the synthetic cephalometric radiographs. Additional- ly, the frontonasal suture, highlighted by the red arrow in Fig. 3c, is more clearly discernible than that in Figure 3d. Again, we select an ROI containing all teeth indicated by a red rectan- gle in topleft panel of Figure 3. The zoomed-in views of the corresponding cephalometric radiographs are shown in row 2 of Figure 3. It can be observed that teeth and root canals are more clearly discernible in Figures 3a and 3b than those in the conventional lateral cephalometric radiograph because of the reduced structure overlap. Notably, the generated cephalomet- ric radiographs in Figures 3a and 3b show that the missing tooth is on the left side.
Figure 3. Lateral cephalometric radiographs generated from left volume (a), right volume (b), and whole volume (c) formed by CBCT images and the corresponding conventional cephalometric radiograph (d) for patient 2. Row 1: Full-view cephalometric radiographs. Row 2: Zoomed-in view of the corresponding cephalometric radiographs in Row 1.
Additionally, we generate the PA cephalometric radiographs from front and whole CBCT volumes with the projection path shown in bottom row of Figure 1. The resulting images, along with zoomed-in views of an ROI highlighted by the red rectangle in the left panel, are displayed in Figures 4 and 5. These images clearly demonstrate the symmetry of the patient's head about the vertical middle line, indicating that the projection directions used in generating the PA cephalo- metric radiographs are optimally selected. As a result, the key anatomical structures, such as the soft tissue bound- ary, nasal cavity, mandible, and menton, are clearly visible in the generated PA cephalometric radiographs. Furthermore, in the PA cephalometric radiographs generated from the front CBCT volumes only, by removing the spine structures from the 3D CBCT images, the contrast of the teeth and their canals is notably enhanced, with greater clarity observed in Figures 4a and 5a compared to Figures 4b and 5b.
Figure 4. PA cephalometric radiographs generated from front volume (a) and whole volume (b) formed by CBCT images and their corresponding zoomed-in views (c)-(d) for patient 1.
Figure 5. PA cephalometric radiographs generated from front volume (a) and whole volume (b) formed by CBCT images and their corresponding zoomed-in views (c)-(d) for patient 2.
Discussion and Conclusion
In this study, we propose a new method for generating both lateral and PA cephalometric radiographs from 3D CBCT images acquired using the Bondream 3D 1030 system. Results from studies suggest that 1) the proposed method can yield both lateral and PA cephalometric radiographs with improved contrast by automatically selecting the optimal projection direction to minimize positioning errors, which occur virtually in every conventional cephalometric radiograph acquisition; and 2) the flexibility of using varying projection paths, enablingfor generation of the cephalometric radiographs from left, right, front, and whole volumes formed by the 3D CBCT images, further enhancing cephalometric radiograph contrast and yielding accurate spatial location information. The studies presented in this work are limited to the image quality of 3D CBCT images. In some experiments, the physical factors, such as the spatial resolution and artifacts of the 3D CBCT images may affect the quality of the generated lateral and PA cephalometric radiographs.
References
- EK Pae, GA McKenna, T Joseph Sheehan, R Garcia, A Kuhlberg, R Nanda (2001) “Role of lateral cephalograms in assessing severity and difficulty of orthodontic cases,” Am. J. Orthod. Dentofac, 120: 254-62.
- AR Durao, P Pittayapat, MIB Rockenbach, R Olszewski, S Ng, AP Ferreira, R Jacobs (2013) “Validity of 2d lateral cepha- lometry in orthodontics: a systematic review,” Int. J. Oral. Maxillofac. Surg., 14: 1-11.
- NM Helal, OA Basri, BA Hosam (2019) “Significance of cephalometric radiograph in orthodontic treatment plan decision,” J. Contemp. Dent. Pract., 7: 789-93.
- F Savoldi, X Gou, CP McGrath, Y Yang, SC Chow, JK Tsoi, M Gu (2020) “Reliability of lateral cephalometric radiographs in the assessment of the upper airway in children: A retrospec- tive study,” Angle Orthod., 1: 47-55.
- J Heinz, K Stewart, A Ghoneima (2009) “Evaluation of two-dimensional lateral cephalogram and three-dimensional cone beam computed tomography superimpositions: a comparative study,” 48: 519-25.
- RU Hi, Nalcaci FATH Ozturk, O Sokucu (2010) “A compari- son of two-dimensional radiography and three-dimensional computed tomography in angular cephalometric measure- ments,” Dentomaxillofac. Radiol, 39: 100-6.
- L Paula, P Solon-de Mello, C Mattos, A Ruellas, E Sant'Anna (2015) “Influence of magnification and superimposition of structures on cephalometric diagnosis,” Dent. Press J. Orthod, 20: 29-34.
- YJ Yoon, KS Kim, MS Hwang, HJ Kim, EH Choi, KW Kim (2001) “Effect of head rotation on lateral cephalometric radiographs,” Angle Orthod. 71: 396-403.
- OT David, RA Tuce, O Munteanu, A Neagu, I Panainte (2017) “Evaluation of the influence of patient positioning on the reliability of lateral cephalometry,” La radiologia medica, 122: 520-9.
- J Mah, J Huang, H Choo (2010) “Practical applications of cone-beam computed tomography in orthodontics,” J. Am. Dent. Assoc, 141: 7S-13S.
- S Kapila, R Conley, W Harrell Jr (2011) “The current status of cone beam computed tomography imaging in ortho- dontics,” Dentomaxillofac. Radiol., 40: 24-34.
- R Hodges, K Atchison, S White (2013) “Impact of cone-beam computed tomography on orthodontic diagnosis and treatment planning,” Am. J. Orthod. Dentofac, 143: 665-74.
- SD Kapila, JM Nervina (2015) “CBCT in orthodontics: assessment of treatment outcomes and indications for its use,” Dentomaxillofac. Radiol, 44: 20140282.
- A Tadinada, A Marczak, S Yadav, P Mukherjee (2016) “Applications of cone beam computed tomography in ortho- dontics: a review,” Turk. J. Orthod., 29: 73.
- WC Scarfe, B Azevedo, S Toghyani, AG Farman (2017) “Cone beam computed tomographic imaging in orthodon- tics,” Aust. Dent. J, 62: 33-50.
- LA Feldkamp, LC.Davis, JW Kress (1984) “Practical cone-beam algorithm,” J. Opt. Soc. Am. A, 1: 612-9.
