- Email: [email protected]
Accuracy and reliability of linear measurements using 3-dimensional computed tomographic imaging software for Le Fort I Osteotomy
Available online at www.sciencedirect.com
British Journal of Oral and Maxillofacial Surgery 52 (2014) 258–263
Accuracy and reliability of linear measurements using 3-dimensional computed tomographic imaging software for Le Fort I Osteotomy Bruno Felipe Gaia a,∗ , Lucas Rodrigues Pinheiro a , Otávio Shoite Umetsubo a , Oseas Santos Jr. a , Felipe Ferreira Costa a , Marcelo Gusmão Paraíso Cavalcanti b,∗∗ a
School of Dentistry, University of São Paulo, College of Dentistry, Department of Radiology, Av. Prof. Lineu Prestes, 2227, CEP 05508-900 São Paulo, SP, Brazil b School of Dentistry, University of São Paulo, College of Dentistry, Department of Stomatology, A. Prof. Lineu Prestes, 2227, CEP 05508-900 São Paulo, SP, Brazil Accepted 23 December 2013 Available online 20 January 2014
Abstract Our purpose was to compare the accuracy and reliability of linear measurements for Le Fort I osteotomy using volume rendering software. We studied 11 dried skulls and used cone-beam computed tomography (CT) to generate 3-dimensional images. Linear measurements were based on craniometric anatomical landmarks that were predefined as specifically used for Le Fort I osteotomy, and identified twice each by 2 radiologists, independently, using Dolphin imaging version 11.5.04.35. A third examiner then made physical measurements using digital calipers. There was a significant difference between Dolphin imaging and the gold standard, particularly in the pterygoid process. The largest difference was 1.85 mm (LLpPtg L). The mean differences between the physical and the 3-dimensional linear measurements ranged from −0.01 to 1.12 mm for examiner 1, and 0 to 1.85 mm for examiner 2. Interexaminer analysis ranged from 0.51 to 0.93. Intraexaminer correlation coefficients ranged from 0.81 to 0.96 and 0.57 to 0.92, for examiners 1 and 2, respectively. We conclude that the Dolphin imaging should be used sparingly during Le Fort I osteotomy. © 2013 The British Association of Oral and Maxillofacial Surgeons. Published by Elsevier Ltd. All rights reserved. Keywords: Cone-beam computed tomography; 3D imaging; Tomography; X-ray computed; Osteotomy Le Fort
Introduction The success of orthognathic surgery depends on many factors including precise and accurate facial analysis and imaging examinations, which are essential determinants for
∗
Corresponding author. Tel.: +55 11 83835620; fax: +55 11 3091 7899. Corresponding author at: University of São Paulo, College of Dentistry, Department of Stomatology, Av. Prof. Lineu Prestes, 2227, CEP 05508-900 São Paulo, SP, Brazil. Tel.: +55 11 3091 7807; fax: +55 11 3091 7899/3091 7831. E-mail addresses: [email protected], [email protected] (B.F. Gaia), [email protected] (M.G.P. Cavalcanti). ∗∗
correcting facial deformities, securing stable results, and reducing complications.1–3 Advances in craniofacial imaging and techniques for the acquisition of images such as the introduction of cone-beam computed tomography (CT) have improved the understanding of anatomical structures and possible anatomical differences presented by patients with dentofacial deformities. There are many commercial and open-source 3dimensional software programs currently available to assist in diagnosis, plan treatment, and predict outcomes related to orthognathic surgery.1,2,4,5 The 3-dimensional imaging software currently available presents different methods of
0266-4356/$ – see front matter © 2013 The British Association of Oral and Maxillofacial Surgeons. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bjoms.2013.12.012
B.F. Gaia et al. / British Journal of Oral and Maxillofacial Surgery 52 (2014) 258–263
linear measurement that could influence the calculation of anatomical landmarks and possibly compromise linear measurement and consequent treatment.6,7 The purpose of this study was to test the accuracy and reliability of linear measurements made by imaging volumetric software for Le Fort I osteotomy.
Material and methods We studied 11 intact human skulls (8 male and 3 female, aged 19–56) provided by the Department of Anatomy of our hospital after approval by the Institutional Review Committee/Human Specimens Committee (0105.0.017.000-11). There was no ethnic or sex preference in choosing the sample, and no history of bone disease was identified in any medical records. Before the skulls were scanned, the mandibles were fixed to the skulls in central occlusion using tape. The skulls were placed in a bulk bag with water for attenuation of the beam and to mimic the soft tissues.8,9 They were then placed in the cone-beam CT unit to keep them in a position similar to that in clinical practice: they had cone-beam CT in an i-CAT cone beam 3-dimensional Dental Imaging System (Imaging Sciences International, Hatfield, PA, USA) at 0.25mm voxel size for 40 s to acquire the raw data. The field of view was a cylinder 20 cm high and 16 cm in diameter. The grey-scale range of the acquired images was 14 bits. After the original image had been acquired the resulting data were recorded and stored in DICOM (Digital Imaging Communication in Medicine) format to prevent loss of data. Multiplanar (axial, coronal, and sagittal) and 3dimensional reconstructed images were obtained simultaneously, and generated in Dolphin Imaging® version 11.5.04.35 (Dolphin Imaging & Management Solutions, Patterson Technology, Chatsworth, CA, USA), installed in a Dell 650 Precision (Dell Computer Corp., Round Rock, TX, USA)
259
independent workstation running Windows XP (Microsoft, Redmond, WA, USA) operating system. The linear measurements of 3-dimensional coordinates used for this study were obtained using the anatomical structures described in Table 1. Most of the bony structures were selected specifically for the study instead of using the traditional craniometric points, in so far as analysis of these structures is highly relevant in the planning of Le Fort I osteotomy (Table 1).10–12 The examiners followed the protocol, starting with hard tissue, then translucent default and volume sculpting, and analysed all volumes scanned using axial and multiplanar images. They were allowed to use rotation, translation, zoom, transparency, and tilt software tools. They used the linear measurement tool to point to the first landmark of the anatomical structure using a circle. This first landmark was pictured on the multiplanar image and was confirmed using a multiplanar guide. The volume was then analysed again, and the second point was identified using the steps described above. After that the software automatically calculated the smallest distance between the 2 anatomical landmarks (Fig. 1). The measurements (n = 11) were made electronically by 2 oral and maxillofacial radiologists with extensive experience in interpreting CT and knowledge of the software tools used. They previously had their assessments calibrated (twice each) independently, with a 7-day time interval between the repeated measurements to ensure intraexaminer reliability. The physical data were obtained by a third examiner whose assessments had previously been calibrated, and who marked the same set of landmarks directly on the dry skull with no knowledge of imaging measurements. Linear measurements on the dry skull were made with digital calipers (167 series; Mitutoyo Sul Americana Ltda, Suzano, SP, Brazil), the same device that had been used by Lopes et al.8 and Moreira et al.9 with 0.3 mm the active point. This was considered to be the gold standard.
Table 1 Description of the craniometric anatomical structures and their respective linear measurements. Anatomical structures
Description
1. Length of the maxilla (ANS–PNS)
Distance between the anterior (ANS) and posterior nasal spines (PNS) indicating the total length of the maxilla Distance between the pyriform aperture (15 mm above the ANS) up to the anterior portion of descending palatine artery canal indicating the length of the lateral nasal wall. Distance of the length of bony fusion of the pterygoid process of the sphenoid bone to the tuberosity of the maxilla (lateral view) Distance between lateral and medial point of the bone fusion of the pterygoid process to the tuberosity of the maxilla indicating the length of bony fusion (axial view) Distance between a point in the pterygoid fossa (the shortest distance to the descending palatine artery canal) and the distal portion of the lateral pterygoid plate indicating the length of the lateral plate of the pterygoid process
2. Length of the lateral nasal walla (LNW) 3. Length of the pterygoid processa (LPtg) 4. Thickness of the junction of the pterygoid process and the tuberosity of the maxillaa (TPtg) 5. Length of the lateral plate of the pterygoid processa (LLpPtg)
6. Length of the medial plate of the pterygoid processa (LMpPtg) a
Distance between a point in the pterygoid fossa (the shortest distance to the descending palatine artery canal) and the distal portion of the medial pterygoid plate indicating the length of the medial plate of the pterygoid process
Measurements made on right (R) and left (L) sides of the anatomical structure.
260
B.F. Gaia et al. / British Journal of Oral and Maxillofacial Surgery 52 (2014) 258–263
Fig. 1. Three-dimensional measurement of the length of the maxilla (ANS–PNS) using Dolphin Imaging® software. This linear 3-dimensional measurement using multiplanar reconstruction (52.1 mm) shows a different result from the measurement made over the 3-dimensional reconstructed image (62.4 mm).
Data analysis consisted of comparing measurements on the dry skull (gold standard) with multiplanar-cone-beam CT linear measurements obtained with Dolphin Imaging® software. Reliability was assessed based on the intraexaminer and interexaminer measurements, and accuracy of the difference between the measurements made by the examiners and the corresponding physical measurements. The first set of measurements was used for the purposes of interexaminer agreement. Intraexaminer, interexaminer, and gold standard comparisons, and the correlation between individual measurements taken by examiners 1 and 2, were made using the intraclass correlation coefficient (ICC), with a 95% confidence interval (CI).
Results The results for the linear measurements are shown in Tables 2–4. Table 2 shows the mean (SD) differences (mm) between Dolphin Imaging® and physical measurements. The highest mean difference for each measurement was in the region of the pterygoid process. Examiner 1 found the highest mean differences for TPtg L (1.09 mm and 1.12 mm for the first and second analyses, respectively). Examiner 2 found the highest mean differences for (LLpPtg L) Lappet L for both analyses (1.69 mm and 1.85 mm). The mean differences between the physical and the linear measurements ranged from −0.01 to 1.12 mm for examiner 1, and from 0
Table 2 Mean (SD) comparison of Dolphin imaging and physical measurements. Physical measurementa
Examiner 1 (first)
Examiner 1 (second)
Examiner 2 (first)
Examiner 2 (second)
Gold standard
ANS–PNS LNW R LNW L LPtg R LPtg L TPtg R TPtg L LLpPtg R LLpPtg L LMpPtg R LMpPtg L
52.08 (2.75) 42.83 (2.74) 42.19 (2.16) 10.05 (2.96) 9.37 (2.30) 9.95 (1.46) 9.33 (1.84) 14.96 (1.82) 15.51 (2.60) 7.44 (2.20) 8.61 (2.51)
51.82 (2.86) 43.15 (2.81) 41.93 (2.22) 9.87 (2.98) 9.70 (2.41) 9.70 (1.53) 9.30 (1.84) 15.01 (1.60) 15.64 (2.78) 7.51 (2.67) 8.22 (2.57)
52.02 (2.70) 42.44 (2.51) 42.25 (2.33) 10.03 (3.05) 9.31 (1.98) 10.58 (1.52) 10.42 (2.26) 14.37 (2.42) 14.70 (2.41) 7.75 (2.18) 8.01 (2.73)
51.94 (2.47) 42.56 (2.28) 42.27 (2.37) 10.14 (2.78) 9.67 (1.99) 10.59 (1.47) 10.61 (2.11) 13.82 (2.22) 14.54 (2.25) 7.81 (2.21) 8.40 (2.30)
52.23 (3.05) 43.40 (2.28) 42.95 (2.82) 10.75 (3.16) 10.01 (2.66) 10.56 (1.82) 10.42 (2.15) 15.46 (2.36) 16.39 (3.00) 7.50 (3.34) 8.38 (3.12)
a
Measurements made on right (R) and left (L) sides of the anatomical structure.
B.F. Gaia et al. / British Journal of Oral and Maxillofacial Surgery 52 (2014) 258–263
261
Table 3 Intra-class correlation (ICC) for each measurement made by examiners 1 and 2 using Dolphin imaging compared with physical measurements. Physical measurements ANS–PNS LNW R LNW L LPtg R LPtg L TPtg R TPtg L LLpPtg R LLpPtg L LMpPtg R LMpPtg L
Examiner 1
Examiner 2
ICC
(95% CI)
ICC
(95% CI)
0.96 0.81 0.82 0.92 0.84 0.81 0.82 0.81 0.81 0.83 0.94
(0.86–0.99) (0.39–0.94) (0.45–0.95) (0.73–0.98) (0.48–0.95) (0.42–0.95) (0.44–0.95) (0.41–0.95) (0.42–0.95) (0.48–0.95) (0.81–0.98)
0.87 0.66 0.89 0.85 0.84 0.92 0.85 0.73 0.57 0.63 0.66
(0.57 to 0.96) (0.11 to 0.90) (0.62 to 0.97) (0.53 to 0.96) (0.48 to 0.95) (0.73 to 0.98) (0.52 to 0.96) (0.23 to 0.92) (−0.03 to 0.87) (0.06 to 0.89) (0.11 to 0.90)
Table 4 Inter-examiner correlation (ICC) for each measurement made by examiners 1 and 2 using Dolphin imaging compared with physical measurements. Physical measurements
ICC
(95% CI)
ANS–PNS LNW R LNW L LPtg R LPtg L TPtg R TPtg L LLpPtg R LLpPtg L LMpPtg R LMpPtg L
0.93 0.51 0.76 0.94 0.74 0.73 0.75 0.70 0.84 0.73 0.68
(0.77 to 0.98) (−0.12 to 0.85) (0.30 to 0.93) (0.78 to 0.98) (0.25 to 0.92) (0.24 to 0.92) (0.28 to 0.93) (0.18 to 0.91) (0.48 to 0.95) (0.24 to 0.92) (0.14 to 0.91)
to 1.85 mm for examiner 2 (Table 2). Table 3 shows that all measurements made by the examiners were in excellent to moderate agreement. The intraexaminer analysis correlation coefficient ranged from 0.80 to 0.96 and from 0.57 to 0.92 for examiners 1 and 2, respectively, indicating almost perfect agreement for examiner 1, and moderate to almost perfect agreement for examiner 2. The interexaminer analysis correlation coefficient ranged from 0.51 to 0.94 using Dolphin Imaging® software, indicating moderate to almost perfect agreement between examiners (Table 4). Both the highest and the lowest deviations from the gold standard for each measurement were found for LNW R and LPtg R, respectively. Fig. 2 (box plot) indicates general similarity between the examiners, proving that the measurements made using Dolphin did not differ significantly from each other. These results show that differences between examiners and the gold standard were caused by Dolphin’s methods of 3-dimensional linear measurement.
Discussion The introduction of 3-dimensional tomographic images for orthognathic planning and surgical simulation, together with
Fig. 2. Box plots show the mean and interquartile ranges for 3-dimensional measurements of linear distance using Dolphin from 3-dimensional conebeam computed tomography. (1) Examiner 1; (2) Examiner 1 second analysis; (3) Examiner 2; (4) Examiner 2 second analysis.
the rapidly emerging availability of this technology, has broadened the use and application of 3-dimensional imaging, which compensate for the drawbacks of 2-dimensional measurements.13–17 The criteria for 3-dimensional analyses are essential not only to ensure the accuracy of planning of treatment, but also to evaluate anatomical structures and plan ahead for possible alterations, which may lead to operative and postoperative complications.4,18,19 New software programs, such as Dolphin Imaging® , have been developed to improve anatomical analysis before, during, and after operation using a single software program.2,3 Currently available commercial software includes similar tools, such as endocranial navigation, rotation and translation of images, adjustment of images, multiplanar and 3-dimensional reconstructions, and linear and angular measurements. However, these programs use different reconstruction algorithms, or different methods of measurement, or both, and different tools to manipulate the images, and are ultimately intended to aid professionals to formulate more accurate maxillofacial diagnoses and plans of treatment.6 Lagravère et al.20 evaluated the accuracy of linear and angular measurements in 3-dimensional cone-beam CT images and found that the mean measurement error was less than 1 mm and 1◦ from the gold standard. They emphasised that there was a need to use quantitative analyses to obtain this degree of high-quality image, and assumed that this error factor was an acceptable result; however, we think that 1 mm is a considerable distance in the posterior osteotomy of the maxilla, particularly near the descending palatine artery. These quantitative analyses are important in planning operations, to prevent serious complications and reduce morbidity. Power et al.21 evaluated the reliability and reproducibility of 2-dimensional and 3-dimensional cephalometric tracings made with Dolphin Imaging® version 8.0, and concluded that this program was not as reliable as traditional techniques for planning orthognathic movements. Periago et al.22 evaluated the accuracy of linear measurements made on Dolphin 3-dimensional (version 2.3) and concluded that many linear measurements may deviate significantly from the anatomical dimensions. Some of these measurements were considered sufficiently accurate from a clinical standpoint to conduct a proper craniofacial analysis. Our results are agreement with these statements, in that they showed differences ranging
262
B.F. Gaia et al. / British Journal of Oral and Maxillofacial Surgery 52 (2014) 258–263
from 0 to 1.95 mm from the physical measurement, which could either improve or compromise the anatomical evaluation and treatment planning. Most of the related publications did not describe the method of measurement. In a recent publication from our department6 we illustrated 2 different methods of linear measurement: linear analysis over the 3-dimensional reconstructed images and 3-dimensional linear analyses (considered the real 3dimensional measurement), and we concluded that these different methods do influence the accuracy of linear measurement. Imaging software that automatically calculates the smallest distance between the initial and the final landmarks produces true 3-dimensional linear measurements and presents better accuracy and reproducibility compared with physical measurements. The Dolphin program version 11.5.04.35 that was tested in this study automatically calculated the measurements made according to a true method of 3-dimensional linear measurement using multiplanar images, although our results show intraexaminer agreement varying from excellent to moderate. This variability may be explained by the fact that even though the anatomical landmarks may be identified by a multiplanar guide, the selection of the initial and final points must be made on the same spatial plane, which compromises the identification and measurement of anatomical landmarks. Another limitation of the program was related to the measurements made over the 3-dimensional reconstructed images, as the process of locating the anatomical landmarks may vary according to how the landmarks are seen, and the 3-dimensional renderings (that are projected images and not actual surfaces) compromise the linear measurements. This implies that the measurements were obtained over a single view – considered to be 2-dimensional – of 3-dimensional reconstructed images, even with the software tool for correction of tilting. This finding was confirmed when the 3-dimensional reconstructed images were rotated, or translated, or both, which modified the spatial position of the linear measurement, and proved the inadequacy of this method for obtaining 3-dimensional linear measurements. This finding justifies our preference for measurements obtained by multiplanar imaging over those obtained by 3-dimensional reconstructed images. The difficulty in locating the anatomical landmarks, particularly those in the pterygoid region, was described by Ueki et al.11 They stated that the position of the measurement affects the thickness of the pterygomaxillary region, which suggests that the measurements could induce unexpected fractures in that region. De Oliveira et al.7 evaluated the reliability of identification of anatomical landmarks using Dolphin Imaging® , and concluded that the intraobserver and interobserver reliability were excellent, particularly when the multiplanar guide was used, whereas the location of landmarks on the 3-dimensional reconstruction may lead to errors. Nagasaka et al.23 showed that the more closely located 2 landmarks are, the greater the measurement error.
Weissheimer et al.24 compared the precision and accuracy of 6 imaging software programs, including the Dolphin program version 11, for measuring upper airway volume in cone-beam CT data, and concluded that all the programs were reliable but had some limitations. The same methods were used by El and Palomo,25 who concluded that the 3dimensional imaging software programs that were studied gave highly reliable calculations of volume, but were not accurate. Based on these methods we concluded that 3-dimensional linear measurements obtained from multiplanar cone-beam CT using Dolphin Imaging® should be used sparingly before Le Fort I osteotomy as some marginal errors, particularly in the pterygoid region, may be critical for linear measurements.
Authors contribution None.
Conflict of interest None.
Ethical approval None.
Acknowledgement We acknowledge Claudio Mendes Pannuti, Ph.D., of the Department of Periodontology, School of Dentistry, University of São Paulo, for his review of the statistical analysis.
References 1. Tucker S, Cevidanes LH, Styner M, et al. Comparison of actual surgical outcomes and 3-dimensional surgical simulations. J Oral Maxillofac Surg 2010;68:2412–21. 2. Swennen GR, Mollemans W, Schutyser F. Three-dimensional treatment planning of orthognathic surgery in the era of virtual imaging. J Oral Maxillofac Surg 2009;67:2080–92 [Erratum in: J Oral Maxillofac Surg 2009;67:2703]. 3. Pi˜neiro-Aguilar A, Somoza-Martín M, Gandara-Rey JM, et al. Blood loss in orthognathic surgery: a systematic review. J Oral Maxillofac Surg 2011;69:885–92. 4. Popat H, Richmond S, Drage NA. New developments in: threedimensional planning for orthognathic surgery. J Orthod 2010;37:62–71 [Erratum in: J Orthod 2010;37:319]. 5. Quevedo LA, Ruiz JV, Quevedo CA. Using a clinical protocol for orthognathic surgery and assessing a 3-dimensional virtual approach: current therapy. J Oral Maxillofac Surg 2011;69:623–37. 6. Gaia BF, Pinheiro LR, Umetsubo OS, et al. Comparison of precision and accuracy of linear measurements performed by two different imaging
B.F. Gaia et al. / British Journal of Oral and Maxillofacial Surgery 52 (2014) 258–263
7.
8.
9.
10.
11.
12.
13.
14.
15.
software programs and obtained from 3D-CBCT images for Le Fort I osteotomy. Dentomaxillofac Radiol 2013;42:20120178. de Oliveira AE, Cevidanes LH, Phillips C, et al. Observer reliability of three-dimensional cephalometric landmark identification on cone-beam computerized tomography. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2009;107:256–65. Lopes PM, Moreira CR, Perrella A, et al. 3D volume rendering maxillofacial analysis of angular measurements by multislice CT. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2008;105:224–30. Moreira CR, Sales MA, Lopes PM, et al. Assessment of linear and angular measurements on three-dimensional cone-beam computed tomographic images. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2009;108:430–6. O’Regan B, Bharadwaj G. Prospective study of the incidence of serious posterior maxillary haemorrhage during a tuberosity osteotomy in low level Le Fort I operations. Br J Oral Maxillofac Surg 2007;45: 538–42. Ueki K, Hashiba Y, Marukawa K, et al. Assessment of pterygomaxillary separation in Le Fort I Osteotomy in class III patients. J Oral Maxillofac Surg 2009;67:833–9. Ueki K, Hashiba Y, Marukawa K, et al. Determining the anatomy of the descending palatine artery and pterygoid plates with computed tomography in Class III patients. J Craniomaxillofac Surg 2009;37: 469–73. Kang SH, Kim MK, Park SY, et al. Early orthognathic surgery with threedimensional image simulation during presurgical orthodontics in adults. J Craniofac Surg 2011;22:473–81. Cevidanes LH, Styner MA, Proffit WR. Image analysis and superimposition of 3-dimensional cone-beam computed tomography models. Am J Orthod Dentofacial Orthop 2006;129:611–8. Albuquerque MA, Gaia BF, Cavalcanti MG. Comparison between multislice and cone-beam computerized tomography in the volumetric assessment of cleft palate. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2011;112:249–57.
263
16. Park SH, Yu HS, Kim KD, et al. A proposal for a new analysis of craniofacial morphology by 3-dimensional computed tomography. Am J Orthod Dentofacial Orthop 2006;129(600):e23–34. 17. Cavalcanti MG, Haller JW, Vannier MW. Three-dimensional computed tomography landmark measurement in craniofacial surgical planning: experimental validation in vitro. J Oral Maxillofac Surg 1999;57: 690–4. 18. Kramer FJ, Baethge C, Swennen G, et al. Intra- and perioperative complications of the LeFort I osteotomy: a prospective evaluation of 1000 patients. J Craniofac Surg 2004;15:971–9. 19. Santos DT, Costa e Silva AP, Vanier MW, et al. Validity of multislice computerized tomography for diagnosis of maxillofacial fractures using an independent workstation. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2004;98:715–20. 20. Lagravère MO, Carey J, Toogood RW, et al. Three-dimensional accuracy of measurements made with software on cone-beam computed tomography images. Am J Orthod Dentofacial Orthop 2008;134: 112–6. 21. Power G, Breckon J, Sherriff M, et al. Dolphin Imaging Software: an analysis of the accuracy of cephalometric digitization and orthognathic prediction. Int J Oral Maxillofac Surg 2005;34:619–26. 22. Periago DR, Scarfe WC, Moshiri M, et al. Linear accuracy and reliability of cone beam CT derived 3-dimensional images constructed using an orthodontic volumetric rendering program. Angle Orthod 2008;78:387–95. 23. Nagasaka S, Fujimura T, Segoshi K. Development of a non-radiographic cephalometric system. Eur J Orthod 2003;25:77–85. 24. Weissheimer A, Menezes LM, Sameshima GT, et al. Imaging software accuracy for 3-dimensional analysis of the upper airway. Am J Orthod Dentofacial Orthop 2012;142:801–13. 25. El H, Palomo JM. Measuring the airway in 3 dimensions: a reliability and accuracy study. Am J Orthod Dentofacial Orthop 2010;137(Suppl. 4). S50-2,e1–9.
British Journal of Oral and Maxillofacial Surgery 52 (2014) 258–263
Accuracy and reliability of linear measurements using 3-dimensional computed tomographic imaging software for Le Fort I Osteotomy Bruno Felipe Gaia a,∗ , Lucas Rodrigues Pinheiro a , Otávio Shoite Umetsubo a , Oseas Santos Jr. a , Felipe Ferreira Costa a , Marcelo Gusmão Paraíso Cavalcanti b,∗∗ a
School of Dentistry, University of São Paulo, College of Dentistry, Department of Radiology, Av. Prof. Lineu Prestes, 2227, CEP 05508-900 São Paulo, SP, Brazil b School of Dentistry, University of São Paulo, College of Dentistry, Department of Stomatology, A. Prof. Lineu Prestes, 2227, CEP 05508-900 São Paulo, SP, Brazil Accepted 23 December 2013 Available online 20 January 2014
Abstract Our purpose was to compare the accuracy and reliability of linear measurements for Le Fort I osteotomy using volume rendering software. We studied 11 dried skulls and used cone-beam computed tomography (CT) to generate 3-dimensional images. Linear measurements were based on craniometric anatomical landmarks that were predefined as specifically used for Le Fort I osteotomy, and identified twice each by 2 radiologists, independently, using Dolphin imaging version 11.5.04.35. A third examiner then made physical measurements using digital calipers. There was a significant difference between Dolphin imaging and the gold standard, particularly in the pterygoid process. The largest difference was 1.85 mm (LLpPtg L). The mean differences between the physical and the 3-dimensional linear measurements ranged from −0.01 to 1.12 mm for examiner 1, and 0 to 1.85 mm for examiner 2. Interexaminer analysis ranged from 0.51 to 0.93. Intraexaminer correlation coefficients ranged from 0.81 to 0.96 and 0.57 to 0.92, for examiners 1 and 2, respectively. We conclude that the Dolphin imaging should be used sparingly during Le Fort I osteotomy. © 2013 The British Association of Oral and Maxillofacial Surgeons. Published by Elsevier Ltd. All rights reserved. Keywords: Cone-beam computed tomography; 3D imaging; Tomography; X-ray computed; Osteotomy Le Fort
Introduction The success of orthognathic surgery depends on many factors including precise and accurate facial analysis and imaging examinations, which are essential determinants for
∗
Corresponding author. Tel.: +55 11 83835620; fax: +55 11 3091 7899. Corresponding author at: University of São Paulo, College of Dentistry, Department of Stomatology, Av. Prof. Lineu Prestes, 2227, CEP 05508-900 São Paulo, SP, Brazil. Tel.: +55 11 3091 7807; fax: +55 11 3091 7899/3091 7831. E-mail addresses: [email protected], [email protected] (B.F. Gaia), [email protected] (M.G.P. Cavalcanti). ∗∗
correcting facial deformities, securing stable results, and reducing complications.1–3 Advances in craniofacial imaging and techniques for the acquisition of images such as the introduction of cone-beam computed tomography (CT) have improved the understanding of anatomical structures and possible anatomical differences presented by patients with dentofacial deformities. There are many commercial and open-source 3dimensional software programs currently available to assist in diagnosis, plan treatment, and predict outcomes related to orthognathic surgery.1,2,4,5 The 3-dimensional imaging software currently available presents different methods of
0266-4356/$ – see front matter © 2013 The British Association of Oral and Maxillofacial Surgeons. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bjoms.2013.12.012
B.F. Gaia et al. / British Journal of Oral and Maxillofacial Surgery 52 (2014) 258–263
linear measurement that could influence the calculation of anatomical landmarks and possibly compromise linear measurement and consequent treatment.6,7 The purpose of this study was to test the accuracy and reliability of linear measurements made by imaging volumetric software for Le Fort I osteotomy.
Material and methods We studied 11 intact human skulls (8 male and 3 female, aged 19–56) provided by the Department of Anatomy of our hospital after approval by the Institutional Review Committee/Human Specimens Committee (0105.0.017.000-11). There was no ethnic or sex preference in choosing the sample, and no history of bone disease was identified in any medical records. Before the skulls were scanned, the mandibles were fixed to the skulls in central occlusion using tape. The skulls were placed in a bulk bag with water for attenuation of the beam and to mimic the soft tissues.8,9 They were then placed in the cone-beam CT unit to keep them in a position similar to that in clinical practice: they had cone-beam CT in an i-CAT cone beam 3-dimensional Dental Imaging System (Imaging Sciences International, Hatfield, PA, USA) at 0.25mm voxel size for 40 s to acquire the raw data. The field of view was a cylinder 20 cm high and 16 cm in diameter. The grey-scale range of the acquired images was 14 bits. After the original image had been acquired the resulting data were recorded and stored in DICOM (Digital Imaging Communication in Medicine) format to prevent loss of data. Multiplanar (axial, coronal, and sagittal) and 3dimensional reconstructed images were obtained simultaneously, and generated in Dolphin Imaging® version 11.5.04.35 (Dolphin Imaging & Management Solutions, Patterson Technology, Chatsworth, CA, USA), installed in a Dell 650 Precision (Dell Computer Corp., Round Rock, TX, USA)
259
independent workstation running Windows XP (Microsoft, Redmond, WA, USA) operating system. The linear measurements of 3-dimensional coordinates used for this study were obtained using the anatomical structures described in Table 1. Most of the bony structures were selected specifically for the study instead of using the traditional craniometric points, in so far as analysis of these structures is highly relevant in the planning of Le Fort I osteotomy (Table 1).10–12 The examiners followed the protocol, starting with hard tissue, then translucent default and volume sculpting, and analysed all volumes scanned using axial and multiplanar images. They were allowed to use rotation, translation, zoom, transparency, and tilt software tools. They used the linear measurement tool to point to the first landmark of the anatomical structure using a circle. This first landmark was pictured on the multiplanar image and was confirmed using a multiplanar guide. The volume was then analysed again, and the second point was identified using the steps described above. After that the software automatically calculated the smallest distance between the 2 anatomical landmarks (Fig. 1). The measurements (n = 11) were made electronically by 2 oral and maxillofacial radiologists with extensive experience in interpreting CT and knowledge of the software tools used. They previously had their assessments calibrated (twice each) independently, with a 7-day time interval between the repeated measurements to ensure intraexaminer reliability. The physical data were obtained by a third examiner whose assessments had previously been calibrated, and who marked the same set of landmarks directly on the dry skull with no knowledge of imaging measurements. Linear measurements on the dry skull were made with digital calipers (167 series; Mitutoyo Sul Americana Ltda, Suzano, SP, Brazil), the same device that had been used by Lopes et al.8 and Moreira et al.9 with 0.3 mm the active point. This was considered to be the gold standard.
Table 1 Description of the craniometric anatomical structures and their respective linear measurements. Anatomical structures
Description
1. Length of the maxilla (ANS–PNS)
Distance between the anterior (ANS) and posterior nasal spines (PNS) indicating the total length of the maxilla Distance between the pyriform aperture (15 mm above the ANS) up to the anterior portion of descending palatine artery canal indicating the length of the lateral nasal wall. Distance of the length of bony fusion of the pterygoid process of the sphenoid bone to the tuberosity of the maxilla (lateral view) Distance between lateral and medial point of the bone fusion of the pterygoid process to the tuberosity of the maxilla indicating the length of bony fusion (axial view) Distance between a point in the pterygoid fossa (the shortest distance to the descending palatine artery canal) and the distal portion of the lateral pterygoid plate indicating the length of the lateral plate of the pterygoid process
2. Length of the lateral nasal walla (LNW) 3. Length of the pterygoid processa (LPtg) 4. Thickness of the junction of the pterygoid process and the tuberosity of the maxillaa (TPtg) 5. Length of the lateral plate of the pterygoid processa (LLpPtg)
6. Length of the medial plate of the pterygoid processa (LMpPtg) a
Distance between a point in the pterygoid fossa (the shortest distance to the descending palatine artery canal) and the distal portion of the medial pterygoid plate indicating the length of the medial plate of the pterygoid process
Measurements made on right (R) and left (L) sides of the anatomical structure.
260
B.F. Gaia et al. / British Journal of Oral and Maxillofacial Surgery 52 (2014) 258–263
Fig. 1. Three-dimensional measurement of the length of the maxilla (ANS–PNS) using Dolphin Imaging® software. This linear 3-dimensional measurement using multiplanar reconstruction (52.1 mm) shows a different result from the measurement made over the 3-dimensional reconstructed image (62.4 mm).
Data analysis consisted of comparing measurements on the dry skull (gold standard) with multiplanar-cone-beam CT linear measurements obtained with Dolphin Imaging® software. Reliability was assessed based on the intraexaminer and interexaminer measurements, and accuracy of the difference between the measurements made by the examiners and the corresponding physical measurements. The first set of measurements was used for the purposes of interexaminer agreement. Intraexaminer, interexaminer, and gold standard comparisons, and the correlation between individual measurements taken by examiners 1 and 2, were made using the intraclass correlation coefficient (ICC), with a 95% confidence interval (CI).
Results The results for the linear measurements are shown in Tables 2–4. Table 2 shows the mean (SD) differences (mm) between Dolphin Imaging® and physical measurements. The highest mean difference for each measurement was in the region of the pterygoid process. Examiner 1 found the highest mean differences for TPtg L (1.09 mm and 1.12 mm for the first and second analyses, respectively). Examiner 2 found the highest mean differences for (LLpPtg L) Lappet L for both analyses (1.69 mm and 1.85 mm). The mean differences between the physical and the linear measurements ranged from −0.01 to 1.12 mm for examiner 1, and from 0
Table 2 Mean (SD) comparison of Dolphin imaging and physical measurements. Physical measurementa
Examiner 1 (first)
Examiner 1 (second)
Examiner 2 (first)
Examiner 2 (second)
Gold standard
ANS–PNS LNW R LNW L LPtg R LPtg L TPtg R TPtg L LLpPtg R LLpPtg L LMpPtg R LMpPtg L
52.08 (2.75) 42.83 (2.74) 42.19 (2.16) 10.05 (2.96) 9.37 (2.30) 9.95 (1.46) 9.33 (1.84) 14.96 (1.82) 15.51 (2.60) 7.44 (2.20) 8.61 (2.51)
51.82 (2.86) 43.15 (2.81) 41.93 (2.22) 9.87 (2.98) 9.70 (2.41) 9.70 (1.53) 9.30 (1.84) 15.01 (1.60) 15.64 (2.78) 7.51 (2.67) 8.22 (2.57)
52.02 (2.70) 42.44 (2.51) 42.25 (2.33) 10.03 (3.05) 9.31 (1.98) 10.58 (1.52) 10.42 (2.26) 14.37 (2.42) 14.70 (2.41) 7.75 (2.18) 8.01 (2.73)
51.94 (2.47) 42.56 (2.28) 42.27 (2.37) 10.14 (2.78) 9.67 (1.99) 10.59 (1.47) 10.61 (2.11) 13.82 (2.22) 14.54 (2.25) 7.81 (2.21) 8.40 (2.30)
52.23 (3.05) 43.40 (2.28) 42.95 (2.82) 10.75 (3.16) 10.01 (2.66) 10.56 (1.82) 10.42 (2.15) 15.46 (2.36) 16.39 (3.00) 7.50 (3.34) 8.38 (3.12)
a
Measurements made on right (R) and left (L) sides of the anatomical structure.
B.F. Gaia et al. / British Journal of Oral and Maxillofacial Surgery 52 (2014) 258–263
261
Table 3 Intra-class correlation (ICC) for each measurement made by examiners 1 and 2 using Dolphin imaging compared with physical measurements. Physical measurements ANS–PNS LNW R LNW L LPtg R LPtg L TPtg R TPtg L LLpPtg R LLpPtg L LMpPtg R LMpPtg L
Examiner 1
Examiner 2
ICC
(95% CI)
ICC
(95% CI)
0.96 0.81 0.82 0.92 0.84 0.81 0.82 0.81 0.81 0.83 0.94
(0.86–0.99) (0.39–0.94) (0.45–0.95) (0.73–0.98) (0.48–0.95) (0.42–0.95) (0.44–0.95) (0.41–0.95) (0.42–0.95) (0.48–0.95) (0.81–0.98)
0.87 0.66 0.89 0.85 0.84 0.92 0.85 0.73 0.57 0.63 0.66
(0.57 to 0.96) (0.11 to 0.90) (0.62 to 0.97) (0.53 to 0.96) (0.48 to 0.95) (0.73 to 0.98) (0.52 to 0.96) (0.23 to 0.92) (−0.03 to 0.87) (0.06 to 0.89) (0.11 to 0.90)
Table 4 Inter-examiner correlation (ICC) for each measurement made by examiners 1 and 2 using Dolphin imaging compared with physical measurements. Physical measurements
ICC
(95% CI)
ANS–PNS LNW R LNW L LPtg R LPtg L TPtg R TPtg L LLpPtg R LLpPtg L LMpPtg R LMpPtg L
0.93 0.51 0.76 0.94 0.74 0.73 0.75 0.70 0.84 0.73 0.68
(0.77 to 0.98) (−0.12 to 0.85) (0.30 to 0.93) (0.78 to 0.98) (0.25 to 0.92) (0.24 to 0.92) (0.28 to 0.93) (0.18 to 0.91) (0.48 to 0.95) (0.24 to 0.92) (0.14 to 0.91)
to 1.85 mm for examiner 2 (Table 2). Table 3 shows that all measurements made by the examiners were in excellent to moderate agreement. The intraexaminer analysis correlation coefficient ranged from 0.80 to 0.96 and from 0.57 to 0.92 for examiners 1 and 2, respectively, indicating almost perfect agreement for examiner 1, and moderate to almost perfect agreement for examiner 2. The interexaminer analysis correlation coefficient ranged from 0.51 to 0.94 using Dolphin Imaging® software, indicating moderate to almost perfect agreement between examiners (Table 4). Both the highest and the lowest deviations from the gold standard for each measurement were found for LNW R and LPtg R, respectively. Fig. 2 (box plot) indicates general similarity between the examiners, proving that the measurements made using Dolphin did not differ significantly from each other. These results show that differences between examiners and the gold standard were caused by Dolphin’s methods of 3-dimensional linear measurement.
Discussion The introduction of 3-dimensional tomographic images for orthognathic planning and surgical simulation, together with
Fig. 2. Box plots show the mean and interquartile ranges for 3-dimensional measurements of linear distance using Dolphin from 3-dimensional conebeam computed tomography. (1) Examiner 1; (2) Examiner 1 second analysis; (3) Examiner 2; (4) Examiner 2 second analysis.
the rapidly emerging availability of this technology, has broadened the use and application of 3-dimensional imaging, which compensate for the drawbacks of 2-dimensional measurements.13–17 The criteria for 3-dimensional analyses are essential not only to ensure the accuracy of planning of treatment, but also to evaluate anatomical structures and plan ahead for possible alterations, which may lead to operative and postoperative complications.4,18,19 New software programs, such as Dolphin Imaging® , have been developed to improve anatomical analysis before, during, and after operation using a single software program.2,3 Currently available commercial software includes similar tools, such as endocranial navigation, rotation and translation of images, adjustment of images, multiplanar and 3-dimensional reconstructions, and linear and angular measurements. However, these programs use different reconstruction algorithms, or different methods of measurement, or both, and different tools to manipulate the images, and are ultimately intended to aid professionals to formulate more accurate maxillofacial diagnoses and plans of treatment.6 Lagravère et al.20 evaluated the accuracy of linear and angular measurements in 3-dimensional cone-beam CT images and found that the mean measurement error was less than 1 mm and 1◦ from the gold standard. They emphasised that there was a need to use quantitative analyses to obtain this degree of high-quality image, and assumed that this error factor was an acceptable result; however, we think that 1 mm is a considerable distance in the posterior osteotomy of the maxilla, particularly near the descending palatine artery. These quantitative analyses are important in planning operations, to prevent serious complications and reduce morbidity. Power et al.21 evaluated the reliability and reproducibility of 2-dimensional and 3-dimensional cephalometric tracings made with Dolphin Imaging® version 8.0, and concluded that this program was not as reliable as traditional techniques for planning orthognathic movements. Periago et al.22 evaluated the accuracy of linear measurements made on Dolphin 3-dimensional (version 2.3) and concluded that many linear measurements may deviate significantly from the anatomical dimensions. Some of these measurements were considered sufficiently accurate from a clinical standpoint to conduct a proper craniofacial analysis. Our results are agreement with these statements, in that they showed differences ranging
262
B.F. Gaia et al. / British Journal of Oral and Maxillofacial Surgery 52 (2014) 258–263
from 0 to 1.95 mm from the physical measurement, which could either improve or compromise the anatomical evaluation and treatment planning. Most of the related publications did not describe the method of measurement. In a recent publication from our department6 we illustrated 2 different methods of linear measurement: linear analysis over the 3-dimensional reconstructed images and 3-dimensional linear analyses (considered the real 3dimensional measurement), and we concluded that these different methods do influence the accuracy of linear measurement. Imaging software that automatically calculates the smallest distance between the initial and the final landmarks produces true 3-dimensional linear measurements and presents better accuracy and reproducibility compared with physical measurements. The Dolphin program version 11.5.04.35 that was tested in this study automatically calculated the measurements made according to a true method of 3-dimensional linear measurement using multiplanar images, although our results show intraexaminer agreement varying from excellent to moderate. This variability may be explained by the fact that even though the anatomical landmarks may be identified by a multiplanar guide, the selection of the initial and final points must be made on the same spatial plane, which compromises the identification and measurement of anatomical landmarks. Another limitation of the program was related to the measurements made over the 3-dimensional reconstructed images, as the process of locating the anatomical landmarks may vary according to how the landmarks are seen, and the 3-dimensional renderings (that are projected images and not actual surfaces) compromise the linear measurements. This implies that the measurements were obtained over a single view – considered to be 2-dimensional – of 3-dimensional reconstructed images, even with the software tool for correction of tilting. This finding was confirmed when the 3-dimensional reconstructed images were rotated, or translated, or both, which modified the spatial position of the linear measurement, and proved the inadequacy of this method for obtaining 3-dimensional linear measurements. This finding justifies our preference for measurements obtained by multiplanar imaging over those obtained by 3-dimensional reconstructed images. The difficulty in locating the anatomical landmarks, particularly those in the pterygoid region, was described by Ueki et al.11 They stated that the position of the measurement affects the thickness of the pterygomaxillary region, which suggests that the measurements could induce unexpected fractures in that region. De Oliveira et al.7 evaluated the reliability of identification of anatomical landmarks using Dolphin Imaging® , and concluded that the intraobserver and interobserver reliability were excellent, particularly when the multiplanar guide was used, whereas the location of landmarks on the 3-dimensional reconstruction may lead to errors. Nagasaka et al.23 showed that the more closely located 2 landmarks are, the greater the measurement error.
Weissheimer et al.24 compared the precision and accuracy of 6 imaging software programs, including the Dolphin program version 11, for measuring upper airway volume in cone-beam CT data, and concluded that all the programs were reliable but had some limitations. The same methods were used by El and Palomo,25 who concluded that the 3dimensional imaging software programs that were studied gave highly reliable calculations of volume, but were not accurate. Based on these methods we concluded that 3-dimensional linear measurements obtained from multiplanar cone-beam CT using Dolphin Imaging® should be used sparingly before Le Fort I osteotomy as some marginal errors, particularly in the pterygoid region, may be critical for linear measurements.
Authors contribution None.
Conflict of interest None.
Ethical approval None.
Acknowledgement We acknowledge Claudio Mendes Pannuti, Ph.D., of the Department of Periodontology, School of Dentistry, University of São Paulo, for his review of the statistical analysis.
References 1. Tucker S, Cevidanes LH, Styner M, et al. Comparison of actual surgical outcomes and 3-dimensional surgical simulations. J Oral Maxillofac Surg 2010;68:2412–21. 2. Swennen GR, Mollemans W, Schutyser F. Three-dimensional treatment planning of orthognathic surgery in the era of virtual imaging. J Oral Maxillofac Surg 2009;67:2080–92 [Erratum in: J Oral Maxillofac Surg 2009;67:2703]. 3. Pi˜neiro-Aguilar A, Somoza-Martín M, Gandara-Rey JM, et al. Blood loss in orthognathic surgery: a systematic review. J Oral Maxillofac Surg 2011;69:885–92. 4. Popat H, Richmond S, Drage NA. New developments in: threedimensional planning for orthognathic surgery. J Orthod 2010;37:62–71 [Erratum in: J Orthod 2010;37:319]. 5. Quevedo LA, Ruiz JV, Quevedo CA. Using a clinical protocol for orthognathic surgery and assessing a 3-dimensional virtual approach: current therapy. J Oral Maxillofac Surg 2011;69:623–37. 6. Gaia BF, Pinheiro LR, Umetsubo OS, et al. Comparison of precision and accuracy of linear measurements performed by two different imaging
B.F. Gaia et al. / British Journal of Oral and Maxillofacial Surgery 52 (2014) 258–263
7.
8.
9.
10.
11.
12.
13.
14.
15.
software programs and obtained from 3D-CBCT images for Le Fort I osteotomy. Dentomaxillofac Radiol 2013;42:20120178. de Oliveira AE, Cevidanes LH, Phillips C, et al. Observer reliability of three-dimensional cephalometric landmark identification on cone-beam computerized tomography. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2009;107:256–65. Lopes PM, Moreira CR, Perrella A, et al. 3D volume rendering maxillofacial analysis of angular measurements by multislice CT. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2008;105:224–30. Moreira CR, Sales MA, Lopes PM, et al. Assessment of linear and angular measurements on three-dimensional cone-beam computed tomographic images. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2009;108:430–6. O’Regan B, Bharadwaj G. Prospective study of the incidence of serious posterior maxillary haemorrhage during a tuberosity osteotomy in low level Le Fort I operations. Br J Oral Maxillofac Surg 2007;45: 538–42. Ueki K, Hashiba Y, Marukawa K, et al. Assessment of pterygomaxillary separation in Le Fort I Osteotomy in class III patients. J Oral Maxillofac Surg 2009;67:833–9. Ueki K, Hashiba Y, Marukawa K, et al. Determining the anatomy of the descending palatine artery and pterygoid plates with computed tomography in Class III patients. J Craniomaxillofac Surg 2009;37: 469–73. Kang SH, Kim MK, Park SY, et al. Early orthognathic surgery with threedimensional image simulation during presurgical orthodontics in adults. J Craniofac Surg 2011;22:473–81. Cevidanes LH, Styner MA, Proffit WR. Image analysis and superimposition of 3-dimensional cone-beam computed tomography models. Am J Orthod Dentofacial Orthop 2006;129:611–8. Albuquerque MA, Gaia BF, Cavalcanti MG. Comparison between multislice and cone-beam computerized tomography in the volumetric assessment of cleft palate. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2011;112:249–57.
263
16. Park SH, Yu HS, Kim KD, et al. A proposal for a new analysis of craniofacial morphology by 3-dimensional computed tomography. Am J Orthod Dentofacial Orthop 2006;129(600):e23–34. 17. Cavalcanti MG, Haller JW, Vannier MW. Three-dimensional computed tomography landmark measurement in craniofacial surgical planning: experimental validation in vitro. J Oral Maxillofac Surg 1999;57: 690–4. 18. Kramer FJ, Baethge C, Swennen G, et al. Intra- and perioperative complications of the LeFort I osteotomy: a prospective evaluation of 1000 patients. J Craniofac Surg 2004;15:971–9. 19. Santos DT, Costa e Silva AP, Vanier MW, et al. Validity of multislice computerized tomography for diagnosis of maxillofacial fractures using an independent workstation. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2004;98:715–20. 20. Lagravère MO, Carey J, Toogood RW, et al. Three-dimensional accuracy of measurements made with software on cone-beam computed tomography images. Am J Orthod Dentofacial Orthop 2008;134: 112–6. 21. Power G, Breckon J, Sherriff M, et al. Dolphin Imaging Software: an analysis of the accuracy of cephalometric digitization and orthognathic prediction. Int J Oral Maxillofac Surg 2005;34:619–26. 22. Periago DR, Scarfe WC, Moshiri M, et al. Linear accuracy and reliability of cone beam CT derived 3-dimensional images constructed using an orthodontic volumetric rendering program. Angle Orthod 2008;78:387–95. 23. Nagasaka S, Fujimura T, Segoshi K. Development of a non-radiographic cephalometric system. Eur J Orthod 2003;25:77–85. 24. Weissheimer A, Menezes LM, Sameshima GT, et al. Imaging software accuracy for 3-dimensional analysis of the upper airway. Am J Orthod Dentofacial Orthop 2012;142:801–13. 25. El H, Palomo JM. Measuring the airway in 3 dimensions: a reliability and accuracy study. Am J Orthod Dentofacial Orthop 2010;137(Suppl. 4). S50-2,e1–9.