Journal of Cranio-Maxillo-Facial Surgery 43 (2015) 1119e1126
Contents lists available at ScienceDirect
Journal of Cranio-Maxillo-Facial Surgery journal homepage: www.jcmfs.com
Intraoperative navigation for single-splint two-jaw orthognathic surgery: From model to actual surgery Hsin-Wen Chang a, Hsiu-Hsia Lin a, Peerasak Chortrakarnkij b, Sun Goo Kim c, Lun-Jou Lo d, * a
Craniofacial Research Center, Chang Gung Memorial Hospital, Taoyuan, Taiwan Division of Plastic Surgery, Siriraj Hospital, Mahidol University, Bangkok, Thailand Seran Plastic Surgery Clinic, Incheon, South Korea d Plastic & Reconstructive Surgery, and Craniofacial Research Center, Chang Gung Memorial Hospital, Chang Gung University, Taoyuan, Taiwan b c
a r t i c l e i n f o
a b s t r a c t
Article history: Paper received 29 January 2015 Accepted 8 June 2015 Available online 18 June 2015
Objective: This study reported an intraoperative navigation system for single-splint two-jaw orthognathic surgery, and assessed the accuracy of transferring the computer assisted surgical simulation. Methods: A skull model was used for validation, and twenty patients receiving such procedure were enrolled. The procedure contained five phases, including virtual surgery on three-dimensional images, fabrication of surgical positioning guides, preparation of registration and validation landmarks, confirmation of bony position during surgery, and postoperative assessment. Target registration error (TRE) and differences between simulation (T0) and postoperative images (T1) were measured from landmarks to Frankfort horizontal plane (FHP), midesagittal plane (MSP), and coronal plane (COP). Results: For the model experiment, mean TRE was lowest using the hard tissue landmarks (0.60 ± 0.27 mm), and the mean difference (T1-T0) was less than 1 mm to all three planes. For the patients, mean TRE was 1.07 ± 0.18 mm from the hard tissue landmarks. The mean difference was 0.96. ± 0.60 mm from MSP, 1.39 ± 1.11 mm from FHP, and 2.12 ± 1.82 mm from COP. The differences were not significant. Both surgeons and patients were satisfied with the surgical outcome. Conclusion: This study showed that the navigation system had acceptable accuracy and was useful for the two-jaw orthognathic surgery using single-splint method. © 2015 European Association for Cranio-Maxillo-Facial Surgery. Published by Elsevier Ltd. All rights reserved.
Keywords: Surgical navigation Orthognathic surgery Single-splint Computer assisted simulation
1. Introduction Computer-assisted surgical planning and simulation has long been applied in craniofacial surgery but less so in maxillofacial surgery (Lo et al., 1994). Because of current improvement in imaging and software, computer-assisted surgery has been increasingly used in orthognathic surgery to improve surgical outcomes. This is particularly helpful for patients with significant facial deformity and asymmetry. However, the accuracy and convenience of transferring the virtual surgery from the image laboratory to the operation theater has yet to be established. Studies have reported
* Corresponding author. Plastic & Reconstructive Surgery, Chang Gung Memorial Hospital, 5, Fu-Shin Street, Kwei Shan, Taoyuan, 333, Taiwan. Tel.: þ886 3 3281200x2105; fax: þ886 3 3271029. E-mail address: [email protected] (L.-J. Lo).
the adoption of intraoperative aids to improve the connection between simulation and actual surgical execution (Zinser et al., 2013b; Mischkowski et al., 2007; Lübbers et al., 2011a). An approach of using a series of surgical splints in conjunction with two-dimensional planning was suggested (Zinser et al., 2012). Computer-aided design/computer-aided manufacture (CAD/CAM) positioning guides were introduced (Polley and Figueroa, 2013). The navigation system was found to be useful for complex two-jaw orthognathic surgery, especially for patients with facial asymmetry (Bell, 2011). A navigation system with preplanned safety margin and registration instrument for accurate surgical implementation was reported (Shim et al., 2013). An augmented reality technique was applied into a waferless navigated orthognathic surgery with an interactive image-guided visualization display (Mischkowski et al., 2006; Zinser et al., 2013a). Navigation system can be used for intraoperative guiding as well as for validation of tissue position. However, inaccuracy of the
http://dx.doi.org/10.1016/j.jcms.2015.06.009 1010-5182/© 2015 European Association for Cranio-Maxillo-Facial Surgery. Published by Elsevier Ltd. All rights reserved.
1120
H.-W. Chang et al. / Journal of Cranio-Maxillo-Facial Surgery 43 (2015) 1119e1126
navigation system could occur, and the registration error has been reported as an important factor. Luebbers et al. reported low accuracy from laser facial skin-based registration method, which was around 2 mm (Sun et al., 2013a, 2013b; Lübbers et al., 2011b). This error was caused largely by distortion of facial tissues from nasotracheal intubation (Sun et al., 2013a, 2013b), shifting of facial skin surface between supine and upright position, which could increase mean target registration error (TRE) from 1.1 to 1.7 mm (Sun et al., 2013b; Marmulla et al., 2006), and the higher noise of cone beam computed tomography (CBCT) (Sun et al., 2013a). Therefore, the facial skin-based registration method was not adequate for imageguided bimaxillary surgery (Sun et al., 2013a, 2013b). A landmarkbased registration method was reported as being more reliable, and resulted in better accuracy in the clinical practice of orthognathic surgery; however, it depends on selection or creation of landmarks. Increase in TRE was found because of less definable and reliable landmarks on the facial skeleton (Sun et al., 2013a, 2013b). Previous studies also indicated that longer distance between the registration points induced higher precision in the navigation system. Additionally, TRE was found to be positively related to the closeness from the registered landmarks when checking the accuracy during the operation (Sun et al., 2013b; Bettschart et al., 2012; Hoffmann et al., 2005). To address this issue, the noticeable dental landmarks were used as points for registration (Sun et al., 2013a, 2013b). Other noninvasive methods such as template-based registration techniques were shown to be useful, which was reported to be 1e2 mm of TRE but required more preoperative effort (Sun et al., 2013a, 2013b; Hoffmann et al., 2005). We have found that an intraoperative navigation system is helpful for single-splint, two-jaw orthognathic surgery, and our experience is reported herein. 2. Material and methods The experimental work and clinical application were approved by the institutional review board of Chang Gung Memorial Hospital, Taoyuan, Taiwan (numbers 101e4244C and 103e1929B). Informed consent was obtained from all patients. Surgical planning and simulation were performed using single-splint method for two-jaw orthognathic surgery (Yu et al., 2009). The maxillary and mandibular segments were fixed into the final occlusal splint and moved as a unit of maxillomandibular complex (MMC) to the planned position. No intermediate occlusal splint was applied. 2.1. Validation test on plastic skull model A synthetic skull model (type: A20, 3B Scientific GmbH, Hamburg, Germany) was used for testing the registration accuracy and navigation method. For fiducial marker registration, four 7-mm plaster markers were attached to the palatal plate fabricated with hygienic orthodontic resin. This resin plate was attached to the palatal surface of the skull model (Fig. 1A). The skull model with fiducial markers underwent CBCT scanning using an i-CATTM scanner (Imaging Sciences International, Hatfield, PA, USA) (voxel resolution: 0.4 mm). Data were stored in the Digital Image Communications in Medicine (DICOM) format. The image data were subsequently registered to the navigation system. DICOM files were imported into a virtual surgery planning software, Simplant Pro (Materialize Dental, Leuven, Belgium). Dental casts were fabricated, a two-jaw surgical plan was made, and the final occlusion splint was created for the skull model. The dental casts with the splint for the planned final occlusion were scanned by a three-dimensional (3D) laser scanner (3shape R700, Denmark). Data were stored in stereolithographic (STL) format for maxillary, mandibular, and the final occlusion of dentition.
Fig. 1. Validation study on a plastic skull model. (A) Palatal plate with four fiducial markers; (B) simulation surgery on the skull model.
The study processes consisted of five phases. The first phase was performing virtual surgical planning on 3D CBCT. The virtual surgery included LeFort I and bilateral ramus sagittal split osteotomy. Repositioning of the bony segments was carried out according to the treatment plan (Fig. 1B). During translation and rotation of the bony segments, parameters such as the facial symmetry, facial profile, occlusal cant and inclination, bony collision, or gap from the segments in the maxillary and mandibular regions were taken into consideration. The second phase was designing positioning guides for the final position of MMC in virtual planning (Fig. 2). Two virtual “blank objects” were created and placed over the anterior surface of the maxilla crossing the osteotomy line and covering the piriform rim. Boolean operation was performed to subtract the overlapping part with the skull image and obtained the 3D image for the positioning guide. The positioning guides were manufactured from the CAD images by a 3D printer using biocompatible materials. The third phase was preoperative preparation for the navigation system using the software iPlan CMF 3.0 (BrainLab, Germany) by identifying registration and validation points and performing virtual osteotomies. The surgical simulation was imported into this iPlan software. Works performed in the imaging laboratory were identification of eight registration points (Fig. 3A), the LeFort I osteotomy lines (Fig. 3B), the same as in the surgical simulation,
Fig. 2. Two positioning guides were designed at LeFort I level for the final position of maxillomandibular complex.
H.-W. Chang et al. / Journal of Cranio-Maxillo-Facial Surgery 43 (2015) 1119e1126
1121
Fig. 3. Preoperative preparation for the navigation. The original cone beam computed tomography (CBCT) image (gray) is superimposed with the surgical simulation image (blue) in B and C. (A) Localization of registration points; (B) marking the LeFort I osteotomy line; (C) localization of validation points.
Fig. 4. Intraoperative registration using hard tissue landmark registration method.
and several validation points for intraoperative checking of MMC position (Fig. 3C). The fourth phase was to register the skull model to the navigation system in the operation theater. There were two steps to prepare for this navigation system (Kolibri 2.0®, BrainLab, Germany). The first step was to apply the required instrument for the navigation-guided surgery. The reference tracker was fixed to upper lateral frontal bone, where it was easily detected by the infrared camera of the navigation system. After the camera detected the tracker, the second step began with the registration process. Three
registration methods were performed, including surface matching, bone landmark registration (Fig. 4), and fiducial marker registration. For surface matching, Z-touch® (BrainLab, Germany) laser scanner was used for this marker-free registration. The system camera acquired signals on the facial surface by detecting the reflection from the laser beam projection. Approximately 200 points were required to complete the registration, mainly from the forehead, nose, and cheek areas. For landmark registration, eight bony or dental landmarks were selected before the operation, including the anterior nasal spine, both infraorbital foramens, the lateral corner of one central incisor, the left and right canine tips, and the mesiobuccal cusp of first maxillary molars (Table 1). The spikes on the palatal fiducial markers were used for registration. While the registration was performed, the TRE was calculated by the distance between the actual position and the target point in the virtual procedure (Fig. 5). Operation and navigation were started after all registration procedures were completed. Osteotomies were performed according to plan in the virtual surgery. The fabricated final occlusal splint was used to wire the maxillary and mandibular segments together, forming the MMC. This MMC was moved to the planned position, assisted by positioning guides. Subsequently, the MMC position was checked by the navigation system and fixed with plates and screws. The final phase was postoperative 3D assessment. CBCT was performed postoperatively. The DICOM files were imported into simulation software and superimposed with the preoperative
Table 1 Definition of hard tissue landmarks and reference planes. Hard tissue landmarks Landmark
Abbreviation
Definition
A point Upper incisor Maxillary canine Maxillary first molar Orbitale Porion Basion Nasion
A U1 U3R/U3L U6R/U6L Or Po Ba N
Maximum concavity in midline of alveolar process of maxilla Most medial and inferior corner of left central incisor Cusp of maxillary canine (right and left) Mesiobuccal cusp of first maxillary molar (right and left) Most inferior point of each infraorbital rim Most superior point of each external acoustic meatus Most anterior point of the foramen magnum Midpoint of the frontonasal suture
Plane
Abbreviation
Definition
Frankfort horizontal plane Midesagittal plane Coronal plane
FHP MSP COP
A plane through or left, or right and the middle point of two Po A plane perpendicular to FHP and through N and Ba A plane perpendicular to FHP and MSP and through Ba
Reference planes
1122
H.-W. Chang et al. / Journal of Cranio-Maxillo-Facial Surgery 43 (2015) 1119e1126
Fig. 5. Performing the intraoperative registration on the skull model and checking the target registration error.
virtual surgical planning. A color map showed the differences between virtual surgery (T0) and postoperative image (T1). The color scale ranged from 0 to 3 mm, with the minimum difference coded as green and the maximum difference coded as purple. The differences between T0 and T1 were measured by the distance from hard tissue landmarks to Frankfort horizontal plane (FHP), midesagittal plane (MSP), and coronal plane (COP) (Fig. 6 and Table 1).
Fig. 6. Measurement of landmark distance to Frankfort horizontal plane, midesagittal plane, and coronal plane.
2.2. Clinical experience Twenty patients receiving single-splint, two-jaw orthognathic surgery between March 2013 and May 2014 were enrolled (Table 2). Of the patients, 12 were female and 8 were male, with a mean age of 19.7 ± 3.6 years. Preoperative CBCT was taken by i-CAT scanner within 1 month before surgery. The DICOM data were imported for surgical simulation using Simplant Pro software. Collaborative orthodontists designed the surgical plan and fabricated the final occlusal splint. Dental casts with the final occlusal splint were scanned by the 3D laser scanner and stored as STL files individually for the maxillary, mandibular, and both dentitions in the final occlusal splint. The five phases were carried out same as in the skull model experiment. The third phase was carried out in the imaging laboratory; it consisted of preoperative preparation including identification of registration points, osteotomy lines, and validation points. At least four hard tissue landmarks were used as registration points, i.e., anterior nasal spine and dental cusps or tips. The fourth phase was in the operating theater to register the patient to the navigation system before performing surgery and navigation. The reference tracker was fixed to the frontal bone behind the hairline. After it was detected, registration was started. Both facial surface matching and hard tissue landmark registration methods were used. Orthognathic surgery and navigation were performed after the registration. Osteotomies were carried out. The maxillary and mandibular segments were wired together in the final occlusal splint. The MMC was moved to the planned position with the aid of positioning guides, which were used in 16 of the 20 patients. The MMC was temporarily fixed and the position checked by navigation system before final fixation was achieved (Fig. 7). LeFort I fixation was first carried out using miniplates and screws, followed by fixation of the mandibular ramus segments using transcutaneous bicortical titanium screws. Positioning of the ascending ramus and condyle was achieved by placing this segment in its original relaxed
Table 2 Patient information and procedures of orthognathic surgery. Gender/age (year)
Diagnosis
1 2 3 4 5 6 7 8 9 10 11 12 13
F/17 F/18 F/17 F/18 F/20 M/18 F/16 F/18 M/19 M/20 F/17 F/25 M/19
14
M/19
Cleft palate Right CLP Left CLP Left CLP Bilateral CLP Right CLP Right CLP Cleft palate Bilateral CLP Left CLP Left CLP Right CLP Prognathism and facial asymmetry Left CLP
15 16 17 18 19 20
F/20 F/24 M/20 M/19 M/18 F/32
Bilateral CLP Cleft palate Left CLP Left CLP Right CLP Cleft palate
Surgical procedures LeFort I osteotomy
BSSO
Genioplasty
Impaction 7 mm, adv R 9 mm L 11 mm Adv R 7 mm L 2 mm, post impaction 3 mm, shift to R 3 mm Post impaction L 4.5 mm R 2.5 mm, yaw rotation L forward, shift to L 3 mm Adv 5 mm, extrude 3 mm at U1, extrude 5 mm at L molar, shift to L 3 mm Adv ANS 6 mm, post impaction R 3.5 mm L 0.5 mm, shift to L 1 mm Adv 8 mm LFI level, U1 adv 4 mm, clockwise rotation, post impaction 2 mm Intrude 3 mm, adv 5 mm Adv 5 mm, impaction R 1 mm L 3 mm, shift to L 1.5 mm Anterior down 2 mm, posterior impaction 2 mm, adv 1.5 mm Adv R 3 mm L 7 mm, Intrusion R post 2 mm, L post 1 mm Adv 6 mm Yaw rotation, R U6 adv 3 mm, L U6 setback 2 mm, shift to R 2 mm Adv R 8.5 mm L 6 mm, yaw rotation, post impaction 4 mm
Autorotation, setback R 10 mm L 6 mm Counterclockwise rotation, setback 5 mm Setback R 7 mm L 3 mm Setback 3 mm Setback R 7 mm L 2 mm Setback R 6 mm L 3 mm Setback R 5 mm L 3 mm Setback L 6.5 mm R 5.5 mm Setback 6 mm Setback R 17 mm, L 14 mm Setback 8 mm Setback 5 mm Setback R 11 mm L 4 mm
Adv 7 mm Adv 6 mm Lengthening R 3 mm, adv L 3 mm Narrowing L side 4 mm Shift to L 2 mm, adv 5 mm Nil Adv 6 mm, lengthening R 2 mm Nil Nil Nil Nil Vertical lengthening 3 mm, adv 5 mm Nil
ANS adv 5.7 mm, down 1.7 mm U1 to L 2.2 mm, adv 2 mm, down 2 mm, U6R to R 0. 9 mm, adv 0.6 mm, up 2.3 mm U6L to R 0.4 mm, adv 4.3 mm, down 0.6 mm. Adv 5 mm, shortening 2 mm. Adv 4 mm, shift to R 1 mm, impaction R 1 mm Adv 6 mm, shift to L 1 mm Adv 5 mm L adv 5 mm, R adv 8 mm, anterior extrusion 2 mm Impaction R 5 mm L 2 mm, adv R 2.7 mm L 3.8 mm, shift to R, yaw rotation L adv
R setback 11.5 mm upper border, 15.5 mm lower border, L setback 6.0 mm upper border, 9.2 mm lower border. Setback R 7 mm L 4 mm Setback R 4 mm L 2 mm Setback R 5 mm L 1 mm Setback R 10 mm L 0 mm Setback 4 mm Setback 5 mm
Nil
Nil Adv 6 mm, lengthening 3 mm Nil Nil Adv 6 mm Nil
H.-W. Chang et al. / Journal of Cranio-Maxillo-Facial Surgery 43 (2015) 1119e1126
Patient no.
F, female; M, male; CLP, cleft lip and palate; LFI, LeFort I osteotomy; post, posterior; U1, central incisor; U6, first molar; ANS, anterior nasal spine; BSSO, bilateral sagittal split osteotomy; R, right; L, left; adv, advancement. All patients had maxillary hypoplasia and class III malocclusion.
1123
1124
H.-W. Chang et al. / Journal of Cranio-Maxillo-Facial Surgery 43 (2015) 1119e1126
Fig. 7. Intraoperative check of the new position of the maxillomandibular complex.
position and gently pushing upward to the joint space. During the ramus fixation, the position between the proximal and distal segments was maintained the same as in the simulation images. Repeated checks by the navigation system were performed in the fixation process. Then the intermaxillary fixation was released, and the dental occlusion was checked. Genioplasty was performed as required. There was no intermaxillary fixation after the surgery. The final phase was postoperative assessment. CBCT was taken 1 week after surgery, and the data were imported to Simplant Pro software. A postoperative 3D skull image was superimposed with the surgical simulation image. A color map ranging from 0 to 3 mm was used to show difference between virtual surgical simulation (T0) and postoperative image (T1) (Fig. 8). To evaluate the accuracy between simulation and actual surgical outcome, the distances from designed landmarks to Frankfort horizontal, midesagittal, and coronal planes were measured (Table 3). 2.3. Statistical analysis A t-test and Pearson correlation coefficient (PCC) were used to analyze the differences and correlations between virtual surgery (T0) and actual postoperative (T1) results. To determine the intraobserver reliability, 10 patients were randomly selected, and 11 bony landmarks were identified and measured two times at 1-week intervals by one investigator. 3. Results 3.1. Validation test on the skull model For the registration method, the hard tissue landmark registration had the lowest mean TRE (0.6 ± 0.27 mm), followed by the surface matching (0.84 ± 0.30 mm) and the fiducial marker (1.0 ± 0.52 mm). The differences between the simulation and
Fig. 8. Superimposition of simulation and postoperative images showing the difference in color map.
postsurgical images (T1eT0) from the validation points were 0.87 ± 0.96 mm to MSP, followed by 0.89 ± 0.53 mm to COP, and 0.92 ± 0.84 mm to FHP. 3.2. Clinical experience The intraobserver reliability from the repeated test revealed the mean difference of 0.37 ± 0.18 mm. There was no statistical
H.-W. Chang et al. / Journal of Cranio-Maxillo-Facial Surgery 43 (2015) 1119e1126
1125
Table 3 Distance from landmark points to reference planes and the difference between the virtual surgical simulation (T0) and postoperative images (T1) (mm). Surgical simulation (T0) Mean ± SD Distance to Frankfort horizontal pane A 31.37 ± 4.33 U1 49.88 ± 3.76 U3L 50.13 ± 3.36 U6L 46.73 ± 3.62 U3R 49.86 ± 3.61 U6R 46.43 ± 3.98 Average Distance to midesagittal plane A 2.18 ± 2.36 U1 1.45 ± 1.16 U3L 15.02 ± 3.15 U6L 25.08 ± 2.46 U3R 15.49 ± 3.04 U6R 24.29 ± 1.92 Average Distance to coronal plane A 83.47 ± 6.06 U1 86.15 ± 6.87 U3L 80.65 ± 7.96 U6L 65.91 ± 7.25 U3R 80.66 ± 8.07 U6R 66.31 ± 6.52 Average
Postoperative result (T1)
Min
Max
Mean ± SD
21.40 43.67 44.68 41.69 44.70 40.06
38.07 58.21 56.42 53.45 58.09 53.84
32.59 51.03 51.35 47.95 51.22 47.87
0.07 0.02 9.04 21.74 8.95 20.89
72.58 75.62 67.21 54.92 68.86 57.58
8.34 4.51 19.40 30.22 20.15 29.06
95.24 99.34 96.49 82.37 97.89 84.37
2.1 1.63 14.49 24.7 15.53 24.45
82.21 85.5 80.14 65.37 79.82 64.99
± ± ± ± ± ±
± ± ± ± ± ±
± ± ± ± ± ±
5.07 4.12 3.77 3.97 3.77 3.86
2.57 1.21 3.36 2.00 2.9 1.8
6.07 7.14 8.03 7.45 8.35 6.36
Min 21.12 44.40 45.30 41.57 45.60 42.71
0.07 0.02 7.46 21.41 9.64 21.22
71.07 72.87 64.79 52.31 64.92 55.16
Difference (T1eT0) Max 40.68 58.67 57.85 55.04 58.42 55.40
7.90 4.00 19.26 28.44 19.81 28.48
93.93 101.69 96.49 82.91 97.00 83.58
Mean ± SD
PCC p Value
1.45 1.32 1.34 1.34 1.42 1.48 1.39
± ± ± ± ± ± ±
1.33 1.20 0.95 0.76 1.28 1.18 1.11
0.42 0.36 0.29 0.32 0.25 0.25 0.20
0.99
0.83 0.84 0.98 1.03 1.30 1.02 0.96
± ± ± ± ± ± ±
0.47 0.48 0.64 0.56 1.44 0.59 0.60
0.92 0.63 0.61 0.60 0.77 0.79 0.96
0.99
1.78 2.15 2.11 2.12 2.34 2.24 2.12
± ± ± ± ± ± ±
1.63 1.59 1.66 1.67 2.29 2.15 1.82
0.52 0.77 0.84 0.82 0.75 0.52 0.54
0.97
Max, maximum; Min, minimum; PCC, Pearson correlation coefficient.
significant difference (p > 0.05), and the Pearson correlation coefficient showed a high correlation between the two measurement tasks (PCC > 0.7). Three patients did not have registration verification records. The mean TRE was 1.07 ± 0.18 mm for the landmark registration and 2.80 ± 0.30 mm for the surface matching registration method. The surface matching registration was not accurate, and therefore the landmark registration method was chosen for patient registration to the navigation system. For the accuracy between simulation and postoperative images, the lowest mean differences (T1eT0) was observed in the validation points to MSP (0.96 ± 0.60 mm), followed by FHP (1.39 ± 1.11 mm) and COP (2.12 ± 1.82 mm) (Table 3). The differences were not statistically significant (p > .05). The Pearson correlation coefficient showed a high correlation between virtual surgical simulation (T0) and actual postoperative (T1) result (PCC > 0.7). The surgeons and patients were satisfied with the surgical outcome during follow-up. 4. Discussion The model experiment revealed high accuracy with low registration errors between the navigation system and the model. The hard tissue landmarks were more accurate than the surface scanning and the fiducial marker method. Using the landmark method, the mean differences from the simulation to surgical outcome were less than 1 mm in this study. The results from the experiment indicate that the system is accurate for navigation and transferring virtual planning to actual surgery, and could meet the requirements for clinical application (Lübbers et al., 2012). There was no clear minimum requirement of accuracy for orthognathic surgery navigation in the literature. A 1.5-mm difference was used as a clinically acceptable standard (Sun et al., 2013a), which was described as unnoticeable discrepancy by the naked eye (Lübbers et al., 2012). Compared with previous registration methods, Sun et al. presented an in vitro study using six anatomical landmarks as registration points with 85 target points in different maxillofacial areas and had a mean TRE of 0.93 mm. The accuracy decreased in the regions
farther away from the registration points. Therefore, the authors suggested that surgeons take the mean TRE and the marginal accuracy of 2 mm into consideration (Sun et al., 2013b). Luebbers et al. reported a synthetic skull model registered into the VectorVision® navigation system, and obtained an average precision of 1 mm regardless of the registration methods (Lübbers et al., 2008). Hernandez-Alfaro et al. presented a study using a CAD/CAM splint to reposition the mandible, and reported its average difference between the virtual plan and the actual operation with the greatest ndezdiscrepancy of 1.26 mm in the vertical dimension (Herna Alfaro and Guijarro-Martínez, 2013). The accuracy of our experimental model surgery was in agreement with that in previous studies. Therefore this navigation system can be reliably applied for single-splint, two-jaw orthognathic surgery. In the clinical situation, the TRE of landmark registration method slightly increased as compared with the model surgery (1.07 ± 0.18 mm vs. 0.6 ± 0.27 mm). This error was acceptable without sacrificing surgical precision. It was found that the hard tissue landmarks of the anterior nasal spine and dental cusps were convenient for the registration task. This study showed that the surface matching registration method was not accurate for patients undergoing orthognathic surgery, a finding supporting those in previous studies. The differences between the simulation and actual postoperative images (T1eT0) did not show significant change, indicating acceptable transferring of the virtual surgery (Table 3). However, it was observed that the differences in the real surgery were larger than those in the model surgery (i.e., 0.96 ± 0.60 mm vs. 0.87 ± 0.96 mm to midesagittal plane, 1.39 ± 1.11 mm vs. 0.92 ± 0.84 mm to the Frankfort horizontal plane, and 2.12 ± 1.82 mm vs. 0.89 ± 0.53 mm to the coronal plane). The postoperative CBCT was taken in 1 week and suggested possible displacement of the maxillary segment during this period. The displacement was mild posterior and inferior movement, and may reflect the soft tissue constraint causing the relapse. It is to be noted that intermaxillary fixation was not applied in these patients due to discomfort and airway concerns. Outliers were actually found in
1126
H.-W. Chang et al. / Journal of Cranio-Maxillo-Facial Surgery 43 (2015) 1119e1126
our postoperative evaluation, causing the aberration of the differences. An outlier (patient 19) had 3 standard deviations away from the mean difference in the anteroposterior dimension (to COP), and a greater than 3-mm difference from the UI and A point in the vertical dimension (to FHP). Another two outliers (patients 5 and 7) had 2 standard deviations from the mean value in the horizontal and vertical dimensions. Postoperative maxillary midline was well maintained in this study group, as shown from the negligible difference to midesagittal plane. As such, facial midline and symmetry were maintained. Compared with conventional surgical planning, computerassisted surgical simulation helps to place the maxillary and mandibular segments in the ideal position. The “ideal position: refers not only to maintaining adequate dental occlusion, level, and midline, but also to achieving esthetic outcomes in terms of symmetry and proportion. To transfer the surgical plan, some authors reported experiences in navigating orthognathic surgery. A navigated orthognathic surgery for a patient with hemi-mandibular hyperplasia by a preplanned mirror image for guiding was performed (Bell, 2011). An additional patient with class II malocclusion underwent navigation-assisted surgery combined with intraoperative CT scanning before and after maxillary reposition (Bell, 2011). A “waferless” orthognathic surgery with an interactive image-guided visualization display and the headset registration method to navigate the position of the maxilla was implemented (Zinser et al., 2013a). The in vivo management achieved the precision less than 1 mm in both the anteroposterior and mediolateral dimensions, but decreased in the vertical dimension (Zinser et al., 2013a, 2013b; Mischkowski et al., 2006). The Stryker eNlite navigation system was applied and obtained good reproducibility as compared with conventional positioning using surgical splints (Mazzoni et al., 2010). Our method supports the concept of intraoperative navigation, and we found it convenient and useful for single-splint, two-jaw surgery. There are concerns using this technique when compared with conventional orthognathic surgery. Additional time is required for preoperative preparation, including image processing, virtual operation, and preregistration of landmarks. During the surgery, the setup of the navigation system involves reference tracker application, registering the patient's position, as well as validation. If positioning guides are used, computer-aided design and manufacturing has to be performed. The cost of the hardware, software, and materials need to be taken into consideration for the simulation and navigation of orthognathic surgery. However, a better treatment outcome is obtained, especially for patients with complicated deformity and asymmetry. In our experience, postoperative facial symmetry and harmony have much been improved (Lin et al., 2015). The cost and time are expected to be reduced in the future because of improvements in technology and experience. 5. Conclusion This study showed that the navigation system was accurate and helpful for two-jaw orthognathic surgery using the single-splint method. Conflict of interest There is no conflict of interest in regard to this work. Acknowledgment We would like to thank Drs. Chiung-Shing Huang, Ellen WenChing Ko and Dr. Betty Chien-Jung Pai for the orthodontic
treatment and surgical planning, Dr. Wei-Min Yang for making the palatal plate with fiducial markers, Mr. Keven Song for help of intraoperative registration using the BrainLab, and Miss Pei-Ju Lin for the help of statistical analysis. The study was supported by a grant from Chang Gung Memorial Hospital, CMRPG381601-3, and a grant from Ministry of Science and Technology, MOST 103–2314-B182-028-MY3. References Bell RB: Computer planning and intraoperative navigation in orthognathic surgery. J Oral Maxillofac Surg 69: 592e605, 2011 Bettschart C, Kruse A, Matthews F, Zemann W, Obwegeser JA, Gr€ atz KW, et al: Pointto-point registration with mandibulo-maxillary splint in open and closed jaw position. Evaluation of registration accuracy for computer-aided surgery of the mandible. J Craniomaxillofac Surg 40: 592e598, 2012 Hoffmann J, Westendorff C, Leitner C, Bartz D, Reinert S: Validation of 3D-laser surface registration for image-guided cranio-maxillofacial surgery. J Craniomaxillofac Surg 33: 13e18, 2005 ndez-Alfaro F, Guijarro-Martínez R: New protocol for three-dimensional Herna surgical planning and CAD/CAM splint generation in orthognathic surgery: an in vitro and in vivo study. Int J Oral Maxillofac Surg 42: 1547e1556, 2013 Lo LJ, Marsh JL, Vannier MW, Patel VV: Craniofacial computer-assisted surgical planning and simulation. Clin Plast Surg 21: 501e516, 1994 €tz KW, Kruse A: A simple and Lübbers HT, Obwegeser JA, Matthews F, Eyrich G, Gra flexible concept for computer-navigated surgery of the mandible. J Oral Maxillofac Surg 69: 924e930, 2011a Lübbers HT, Matthews F, Zemann W, Gr€ atz KW, Obwegeser JA, Bredell M: Registration for computer-navigated surgery in edentulous patients: a problem based decision concept. J Craniomaxillofac Surg 39: 453e458, 2011b Lübbers HT, Medinger L, Kruse AL, Gr€ atz KW, Obwegeser JA, Matthews F: The influence of involuntary facial movements on craniofacial anthropometry: a survey using a three-dimensional photographic system. Br J Oral Maxillofac Surg 50: 171e175, 2012 Luebbers HT1, Messmer P, Obwegeser JA, Zwahlen RA, Kikinis R, Graetz KW, et al: Comparison of different registration methods for surgical navigation in craniomaxillofacial surgery. J Craniomaxillofac Surg 36: 109e116, 2008 Lin HH, Chang HW, Wang CH, Kim SG, Lo LJ: Three-dimensional computer-assissted orthognathic surgery: experience of 37 patients. Ann Plast Surg 74 Suppl. 2: S118e126, 2015 €ller JE: Application of an Mischkowski RA, Zinser MJ, Kübler AC, Krug B, Seifert U, Zo augmented reality tool for maxillary positioning in orthognathic surgery-a feasibility study. J Craniomaxillofac Surg 34: 478e483, 2006 €ller JE: Intraoperative Mischkowski RA, Zinser MJ, Ritter L, Neugebauer J, Keeve E, Zo navigation in the maxillofacial area based on 3D imaging obtained by a conebeam device. Int J Oral Maxillofac Surg 36: 687e694, 2007 Marmulla R, Mühling J, Lüth T, Hassfeld S: Physiological shift of facial skin and its influence on the change in precision of computer-assisted surgery. Br J Oral Maxillofac Surg 44: 273e278, 2006 Mazzoni S, Badiali G, Lancellotti L, Babbi L, Bianchi A, Marchetti C: Simulationguided navigation: a new approach to improve intraoperative threedimensional reproducibility during orthognathic surgery. J Craniofac Surg 21: 1698e1705, 2010 Polley JW, Figueroa AA: Orthognathic positioning system: intraoperative system to transfer virtual surgical plan to operating field during orthognathic surgery. J Oral Maxillofac Surg 71: 911e920, 2013 Shim BK, Shin HS, Nam SM, Kim YB: Real-time navigation-assisted orthognathic surgery. J Craniofac Surg 24: 221e225, 2013 Sun Y, Luebbers HT, Agbaje JO, Schepers S, Vrielinck L, Lambrichts I, et al: Evaluation of 3 different registration techniques in image-guided bimaxillary surgery. J Craniofac Surg 24: 1095e1099, 2013a Sun Y, Luebbers HT, Agbaje JO, Schepers S, Vrielinck L, Lambrichts I, et al: Validation of anatomical landmarks-based registration for image-guided surgery: an invitro study. J Craniomaxillofac Surg 41: 522e526, 2013b Yu CC, Bergeron L, Lin CH, Chu YM, Chen YR: Single-splint technique in orthognathic surgery: intraoperative checkpoints to control facial symmetry. Plast Reconstr Surg 124: 879e886, 2009 € ller JE: Computer-assisted orthognathic Zinser MJ, Mischkowski RA, Sailer HF, Zo surgery: feasibility study using multiple CAD/CAM surgical splints. Oral Surg Oral Med Oral Pathol Oral Radiol 113: 673e687, 2012 € ller JE: Zinser MJ, Mischkowski RA, Dreiseidler T, Thamm OC, Rothamel D, Zo Computer-assisted orthognathic surgery: waferless maxillary positioning, versatility and accuracy of an image-guided visualization display. Br J Oral Maxillofac Surg 51: 827e833, 2013a €ller JE: A paradigm shift in Zinser MJ, Sailer HF, Ritter L, Braumann B, Maegele M, Zo orthognathic surgery? A comparison of navigation, computer-aided designed/ computer-aided manufactured splints, and “classic” intermaxillary splints to surgical transfer of virtual orthognathic planning. J Oral Maxillofac Surg 71: 2151e1ee21, 2013b
Contents lists available at ScienceDirect
Journal of Cranio-Maxillo-Facial Surgery journal homepage: www.jcmfs.com
Intraoperative navigation for single-splint two-jaw orthognathic surgery: From model to actual surgery Hsin-Wen Chang a, Hsiu-Hsia Lin a, Peerasak Chortrakarnkij b, Sun Goo Kim c, Lun-Jou Lo d, * a
Craniofacial Research Center, Chang Gung Memorial Hospital, Taoyuan, Taiwan Division of Plastic Surgery, Siriraj Hospital, Mahidol University, Bangkok, Thailand Seran Plastic Surgery Clinic, Incheon, South Korea d Plastic & Reconstructive Surgery, and Craniofacial Research Center, Chang Gung Memorial Hospital, Chang Gung University, Taoyuan, Taiwan b c
a r t i c l e i n f o
a b s t r a c t
Article history: Paper received 29 January 2015 Accepted 8 June 2015 Available online 18 June 2015
Objective: This study reported an intraoperative navigation system for single-splint two-jaw orthognathic surgery, and assessed the accuracy of transferring the computer assisted surgical simulation. Methods: A skull model was used for validation, and twenty patients receiving such procedure were enrolled. The procedure contained five phases, including virtual surgery on three-dimensional images, fabrication of surgical positioning guides, preparation of registration and validation landmarks, confirmation of bony position during surgery, and postoperative assessment. Target registration error (TRE) and differences between simulation (T0) and postoperative images (T1) were measured from landmarks to Frankfort horizontal plane (FHP), midesagittal plane (MSP), and coronal plane (COP). Results: For the model experiment, mean TRE was lowest using the hard tissue landmarks (0.60 ± 0.27 mm), and the mean difference (T1-T0) was less than 1 mm to all three planes. For the patients, mean TRE was 1.07 ± 0.18 mm from the hard tissue landmarks. The mean difference was 0.96. ± 0.60 mm from MSP, 1.39 ± 1.11 mm from FHP, and 2.12 ± 1.82 mm from COP. The differences were not significant. Both surgeons and patients were satisfied with the surgical outcome. Conclusion: This study showed that the navigation system had acceptable accuracy and was useful for the two-jaw orthognathic surgery using single-splint method. © 2015 European Association for Cranio-Maxillo-Facial Surgery. Published by Elsevier Ltd. All rights reserved.
Keywords: Surgical navigation Orthognathic surgery Single-splint Computer assisted simulation
1. Introduction Computer-assisted surgical planning and simulation has long been applied in craniofacial surgery but less so in maxillofacial surgery (Lo et al., 1994). Because of current improvement in imaging and software, computer-assisted surgery has been increasingly used in orthognathic surgery to improve surgical outcomes. This is particularly helpful for patients with significant facial deformity and asymmetry. However, the accuracy and convenience of transferring the virtual surgery from the image laboratory to the operation theater has yet to be established. Studies have reported
* Corresponding author. Plastic & Reconstructive Surgery, Chang Gung Memorial Hospital, 5, Fu-Shin Street, Kwei Shan, Taoyuan, 333, Taiwan. Tel.: þ886 3 3281200x2105; fax: þ886 3 3271029. E-mail address: [email protected] (L.-J. Lo).
the adoption of intraoperative aids to improve the connection between simulation and actual surgical execution (Zinser et al., 2013b; Mischkowski et al., 2007; Lübbers et al., 2011a). An approach of using a series of surgical splints in conjunction with two-dimensional planning was suggested (Zinser et al., 2012). Computer-aided design/computer-aided manufacture (CAD/CAM) positioning guides were introduced (Polley and Figueroa, 2013). The navigation system was found to be useful for complex two-jaw orthognathic surgery, especially for patients with facial asymmetry (Bell, 2011). A navigation system with preplanned safety margin and registration instrument for accurate surgical implementation was reported (Shim et al., 2013). An augmented reality technique was applied into a waferless navigated orthognathic surgery with an interactive image-guided visualization display (Mischkowski et al., 2006; Zinser et al., 2013a). Navigation system can be used for intraoperative guiding as well as for validation of tissue position. However, inaccuracy of the
http://dx.doi.org/10.1016/j.jcms.2015.06.009 1010-5182/© 2015 European Association for Cranio-Maxillo-Facial Surgery. Published by Elsevier Ltd. All rights reserved.
1120
H.-W. Chang et al. / Journal of Cranio-Maxillo-Facial Surgery 43 (2015) 1119e1126
navigation system could occur, and the registration error has been reported as an important factor. Luebbers et al. reported low accuracy from laser facial skin-based registration method, which was around 2 mm (Sun et al., 2013a, 2013b; Lübbers et al., 2011b). This error was caused largely by distortion of facial tissues from nasotracheal intubation (Sun et al., 2013a, 2013b), shifting of facial skin surface between supine and upright position, which could increase mean target registration error (TRE) from 1.1 to 1.7 mm (Sun et al., 2013b; Marmulla et al., 2006), and the higher noise of cone beam computed tomography (CBCT) (Sun et al., 2013a). Therefore, the facial skin-based registration method was not adequate for imageguided bimaxillary surgery (Sun et al., 2013a, 2013b). A landmarkbased registration method was reported as being more reliable, and resulted in better accuracy in the clinical practice of orthognathic surgery; however, it depends on selection or creation of landmarks. Increase in TRE was found because of less definable and reliable landmarks on the facial skeleton (Sun et al., 2013a, 2013b). Previous studies also indicated that longer distance between the registration points induced higher precision in the navigation system. Additionally, TRE was found to be positively related to the closeness from the registered landmarks when checking the accuracy during the operation (Sun et al., 2013b; Bettschart et al., 2012; Hoffmann et al., 2005). To address this issue, the noticeable dental landmarks were used as points for registration (Sun et al., 2013a, 2013b). Other noninvasive methods such as template-based registration techniques were shown to be useful, which was reported to be 1e2 mm of TRE but required more preoperative effort (Sun et al., 2013a, 2013b; Hoffmann et al., 2005). We have found that an intraoperative navigation system is helpful for single-splint, two-jaw orthognathic surgery, and our experience is reported herein. 2. Material and methods The experimental work and clinical application were approved by the institutional review board of Chang Gung Memorial Hospital, Taoyuan, Taiwan (numbers 101e4244C and 103e1929B). Informed consent was obtained from all patients. Surgical planning and simulation were performed using single-splint method for two-jaw orthognathic surgery (Yu et al., 2009). The maxillary and mandibular segments were fixed into the final occlusal splint and moved as a unit of maxillomandibular complex (MMC) to the planned position. No intermediate occlusal splint was applied. 2.1. Validation test on plastic skull model A synthetic skull model (type: A20, 3B Scientific GmbH, Hamburg, Germany) was used for testing the registration accuracy and navigation method. For fiducial marker registration, four 7-mm plaster markers were attached to the palatal plate fabricated with hygienic orthodontic resin. This resin plate was attached to the palatal surface of the skull model (Fig. 1A). The skull model with fiducial markers underwent CBCT scanning using an i-CATTM scanner (Imaging Sciences International, Hatfield, PA, USA) (voxel resolution: 0.4 mm). Data were stored in the Digital Image Communications in Medicine (DICOM) format. The image data were subsequently registered to the navigation system. DICOM files were imported into a virtual surgery planning software, Simplant Pro (Materialize Dental, Leuven, Belgium). Dental casts were fabricated, a two-jaw surgical plan was made, and the final occlusion splint was created for the skull model. The dental casts with the splint for the planned final occlusion were scanned by a three-dimensional (3D) laser scanner (3shape R700, Denmark). Data were stored in stereolithographic (STL) format for maxillary, mandibular, and the final occlusion of dentition.
Fig. 1. Validation study on a plastic skull model. (A) Palatal plate with four fiducial markers; (B) simulation surgery on the skull model.
The study processes consisted of five phases. The first phase was performing virtual surgical planning on 3D CBCT. The virtual surgery included LeFort I and bilateral ramus sagittal split osteotomy. Repositioning of the bony segments was carried out according to the treatment plan (Fig. 1B). During translation and rotation of the bony segments, parameters such as the facial symmetry, facial profile, occlusal cant and inclination, bony collision, or gap from the segments in the maxillary and mandibular regions were taken into consideration. The second phase was designing positioning guides for the final position of MMC in virtual planning (Fig. 2). Two virtual “blank objects” were created and placed over the anterior surface of the maxilla crossing the osteotomy line and covering the piriform rim. Boolean operation was performed to subtract the overlapping part with the skull image and obtained the 3D image for the positioning guide. The positioning guides were manufactured from the CAD images by a 3D printer using biocompatible materials. The third phase was preoperative preparation for the navigation system using the software iPlan CMF 3.0 (BrainLab, Germany) by identifying registration and validation points and performing virtual osteotomies. The surgical simulation was imported into this iPlan software. Works performed in the imaging laboratory were identification of eight registration points (Fig. 3A), the LeFort I osteotomy lines (Fig. 3B), the same as in the surgical simulation,
Fig. 2. Two positioning guides were designed at LeFort I level for the final position of maxillomandibular complex.
H.-W. Chang et al. / Journal of Cranio-Maxillo-Facial Surgery 43 (2015) 1119e1126
1121
Fig. 3. Preoperative preparation for the navigation. The original cone beam computed tomography (CBCT) image (gray) is superimposed with the surgical simulation image (blue) in B and C. (A) Localization of registration points; (B) marking the LeFort I osteotomy line; (C) localization of validation points.
Fig. 4. Intraoperative registration using hard tissue landmark registration method.
and several validation points for intraoperative checking of MMC position (Fig. 3C). The fourth phase was to register the skull model to the navigation system in the operation theater. There were two steps to prepare for this navigation system (Kolibri 2.0®, BrainLab, Germany). The first step was to apply the required instrument for the navigation-guided surgery. The reference tracker was fixed to upper lateral frontal bone, where it was easily detected by the infrared camera of the navigation system. After the camera detected the tracker, the second step began with the registration process. Three
registration methods were performed, including surface matching, bone landmark registration (Fig. 4), and fiducial marker registration. For surface matching, Z-touch® (BrainLab, Germany) laser scanner was used for this marker-free registration. The system camera acquired signals on the facial surface by detecting the reflection from the laser beam projection. Approximately 200 points were required to complete the registration, mainly from the forehead, nose, and cheek areas. For landmark registration, eight bony or dental landmarks were selected before the operation, including the anterior nasal spine, both infraorbital foramens, the lateral corner of one central incisor, the left and right canine tips, and the mesiobuccal cusp of first maxillary molars (Table 1). The spikes on the palatal fiducial markers were used for registration. While the registration was performed, the TRE was calculated by the distance between the actual position and the target point in the virtual procedure (Fig. 5). Operation and navigation were started after all registration procedures were completed. Osteotomies were performed according to plan in the virtual surgery. The fabricated final occlusal splint was used to wire the maxillary and mandibular segments together, forming the MMC. This MMC was moved to the planned position, assisted by positioning guides. Subsequently, the MMC position was checked by the navigation system and fixed with plates and screws. The final phase was postoperative 3D assessment. CBCT was performed postoperatively. The DICOM files were imported into simulation software and superimposed with the preoperative
Table 1 Definition of hard tissue landmarks and reference planes. Hard tissue landmarks Landmark
Abbreviation
Definition
A point Upper incisor Maxillary canine Maxillary first molar Orbitale Porion Basion Nasion
A U1 U3R/U3L U6R/U6L Or Po Ba N
Maximum concavity in midline of alveolar process of maxilla Most medial and inferior corner of left central incisor Cusp of maxillary canine (right and left) Mesiobuccal cusp of first maxillary molar (right and left) Most inferior point of each infraorbital rim Most superior point of each external acoustic meatus Most anterior point of the foramen magnum Midpoint of the frontonasal suture
Plane
Abbreviation
Definition
Frankfort horizontal plane Midesagittal plane Coronal plane
FHP MSP COP
A plane through or left, or right and the middle point of two Po A plane perpendicular to FHP and through N and Ba A plane perpendicular to FHP and MSP and through Ba
Reference planes
1122
H.-W. Chang et al. / Journal of Cranio-Maxillo-Facial Surgery 43 (2015) 1119e1126
Fig. 5. Performing the intraoperative registration on the skull model and checking the target registration error.
virtual surgical planning. A color map showed the differences between virtual surgery (T0) and postoperative image (T1). The color scale ranged from 0 to 3 mm, with the minimum difference coded as green and the maximum difference coded as purple. The differences between T0 and T1 were measured by the distance from hard tissue landmarks to Frankfort horizontal plane (FHP), midesagittal plane (MSP), and coronal plane (COP) (Fig. 6 and Table 1).
Fig. 6. Measurement of landmark distance to Frankfort horizontal plane, midesagittal plane, and coronal plane.
2.2. Clinical experience Twenty patients receiving single-splint, two-jaw orthognathic surgery between March 2013 and May 2014 were enrolled (Table 2). Of the patients, 12 were female and 8 were male, with a mean age of 19.7 ± 3.6 years. Preoperative CBCT was taken by i-CAT scanner within 1 month before surgery. The DICOM data were imported for surgical simulation using Simplant Pro software. Collaborative orthodontists designed the surgical plan and fabricated the final occlusal splint. Dental casts with the final occlusal splint were scanned by the 3D laser scanner and stored as STL files individually for the maxillary, mandibular, and both dentitions in the final occlusal splint. The five phases were carried out same as in the skull model experiment. The third phase was carried out in the imaging laboratory; it consisted of preoperative preparation including identification of registration points, osteotomy lines, and validation points. At least four hard tissue landmarks were used as registration points, i.e., anterior nasal spine and dental cusps or tips. The fourth phase was in the operating theater to register the patient to the navigation system before performing surgery and navigation. The reference tracker was fixed to the frontal bone behind the hairline. After it was detected, registration was started. Both facial surface matching and hard tissue landmark registration methods were used. Orthognathic surgery and navigation were performed after the registration. Osteotomies were carried out. The maxillary and mandibular segments were wired together in the final occlusal splint. The MMC was moved to the planned position with the aid of positioning guides, which were used in 16 of the 20 patients. The MMC was temporarily fixed and the position checked by navigation system before final fixation was achieved (Fig. 7). LeFort I fixation was first carried out using miniplates and screws, followed by fixation of the mandibular ramus segments using transcutaneous bicortical titanium screws. Positioning of the ascending ramus and condyle was achieved by placing this segment in its original relaxed
Table 2 Patient information and procedures of orthognathic surgery. Gender/age (year)
Diagnosis
1 2 3 4 5 6 7 8 9 10 11 12 13
F/17 F/18 F/17 F/18 F/20 M/18 F/16 F/18 M/19 M/20 F/17 F/25 M/19
14
M/19
Cleft palate Right CLP Left CLP Left CLP Bilateral CLP Right CLP Right CLP Cleft palate Bilateral CLP Left CLP Left CLP Right CLP Prognathism and facial asymmetry Left CLP
15 16 17 18 19 20
F/20 F/24 M/20 M/19 M/18 F/32
Bilateral CLP Cleft palate Left CLP Left CLP Right CLP Cleft palate
Surgical procedures LeFort I osteotomy
BSSO
Genioplasty
Impaction 7 mm, adv R 9 mm L 11 mm Adv R 7 mm L 2 mm, post impaction 3 mm, shift to R 3 mm Post impaction L 4.5 mm R 2.5 mm, yaw rotation L forward, shift to L 3 mm Adv 5 mm, extrude 3 mm at U1, extrude 5 mm at L molar, shift to L 3 mm Adv ANS 6 mm, post impaction R 3.5 mm L 0.5 mm, shift to L 1 mm Adv 8 mm LFI level, U1 adv 4 mm, clockwise rotation, post impaction 2 mm Intrude 3 mm, adv 5 mm Adv 5 mm, impaction R 1 mm L 3 mm, shift to L 1.5 mm Anterior down 2 mm, posterior impaction 2 mm, adv 1.5 mm Adv R 3 mm L 7 mm, Intrusion R post 2 mm, L post 1 mm Adv 6 mm Yaw rotation, R U6 adv 3 mm, L U6 setback 2 mm, shift to R 2 mm Adv R 8.5 mm L 6 mm, yaw rotation, post impaction 4 mm
Autorotation, setback R 10 mm L 6 mm Counterclockwise rotation, setback 5 mm Setback R 7 mm L 3 mm Setback 3 mm Setback R 7 mm L 2 mm Setback R 6 mm L 3 mm Setback R 5 mm L 3 mm Setback L 6.5 mm R 5.5 mm Setback 6 mm Setback R 17 mm, L 14 mm Setback 8 mm Setback 5 mm Setback R 11 mm L 4 mm
Adv 7 mm Adv 6 mm Lengthening R 3 mm, adv L 3 mm Narrowing L side 4 mm Shift to L 2 mm, adv 5 mm Nil Adv 6 mm, lengthening R 2 mm Nil Nil Nil Nil Vertical lengthening 3 mm, adv 5 mm Nil
ANS adv 5.7 mm, down 1.7 mm U1 to L 2.2 mm, adv 2 mm, down 2 mm, U6R to R 0. 9 mm, adv 0.6 mm, up 2.3 mm U6L to R 0.4 mm, adv 4.3 mm, down 0.6 mm. Adv 5 mm, shortening 2 mm. Adv 4 mm, shift to R 1 mm, impaction R 1 mm Adv 6 mm, shift to L 1 mm Adv 5 mm L adv 5 mm, R adv 8 mm, anterior extrusion 2 mm Impaction R 5 mm L 2 mm, adv R 2.7 mm L 3.8 mm, shift to R, yaw rotation L adv
R setback 11.5 mm upper border, 15.5 mm lower border, L setback 6.0 mm upper border, 9.2 mm lower border. Setback R 7 mm L 4 mm Setback R 4 mm L 2 mm Setback R 5 mm L 1 mm Setback R 10 mm L 0 mm Setback 4 mm Setback 5 mm
Nil
Nil Adv 6 mm, lengthening 3 mm Nil Nil Adv 6 mm Nil
H.-W. Chang et al. / Journal of Cranio-Maxillo-Facial Surgery 43 (2015) 1119e1126
Patient no.
F, female; M, male; CLP, cleft lip and palate; LFI, LeFort I osteotomy; post, posterior; U1, central incisor; U6, first molar; ANS, anterior nasal spine; BSSO, bilateral sagittal split osteotomy; R, right; L, left; adv, advancement. All patients had maxillary hypoplasia and class III malocclusion.
1123
1124
H.-W. Chang et al. / Journal of Cranio-Maxillo-Facial Surgery 43 (2015) 1119e1126
Fig. 7. Intraoperative check of the new position of the maxillomandibular complex.
position and gently pushing upward to the joint space. During the ramus fixation, the position between the proximal and distal segments was maintained the same as in the simulation images. Repeated checks by the navigation system were performed in the fixation process. Then the intermaxillary fixation was released, and the dental occlusion was checked. Genioplasty was performed as required. There was no intermaxillary fixation after the surgery. The final phase was postoperative assessment. CBCT was taken 1 week after surgery, and the data were imported to Simplant Pro software. A postoperative 3D skull image was superimposed with the surgical simulation image. A color map ranging from 0 to 3 mm was used to show difference between virtual surgical simulation (T0) and postoperative image (T1) (Fig. 8). To evaluate the accuracy between simulation and actual surgical outcome, the distances from designed landmarks to Frankfort horizontal, midesagittal, and coronal planes were measured (Table 3). 2.3. Statistical analysis A t-test and Pearson correlation coefficient (PCC) were used to analyze the differences and correlations between virtual surgery (T0) and actual postoperative (T1) results. To determine the intraobserver reliability, 10 patients were randomly selected, and 11 bony landmarks were identified and measured two times at 1-week intervals by one investigator. 3. Results 3.1. Validation test on the skull model For the registration method, the hard tissue landmark registration had the lowest mean TRE (0.6 ± 0.27 mm), followed by the surface matching (0.84 ± 0.30 mm) and the fiducial marker (1.0 ± 0.52 mm). The differences between the simulation and
Fig. 8. Superimposition of simulation and postoperative images showing the difference in color map.
postsurgical images (T1eT0) from the validation points were 0.87 ± 0.96 mm to MSP, followed by 0.89 ± 0.53 mm to COP, and 0.92 ± 0.84 mm to FHP. 3.2. Clinical experience The intraobserver reliability from the repeated test revealed the mean difference of 0.37 ± 0.18 mm. There was no statistical
H.-W. Chang et al. / Journal of Cranio-Maxillo-Facial Surgery 43 (2015) 1119e1126
1125
Table 3 Distance from landmark points to reference planes and the difference between the virtual surgical simulation (T0) and postoperative images (T1) (mm). Surgical simulation (T0) Mean ± SD Distance to Frankfort horizontal pane A 31.37 ± 4.33 U1 49.88 ± 3.76 U3L 50.13 ± 3.36 U6L 46.73 ± 3.62 U3R 49.86 ± 3.61 U6R 46.43 ± 3.98 Average Distance to midesagittal plane A 2.18 ± 2.36 U1 1.45 ± 1.16 U3L 15.02 ± 3.15 U6L 25.08 ± 2.46 U3R 15.49 ± 3.04 U6R 24.29 ± 1.92 Average Distance to coronal plane A 83.47 ± 6.06 U1 86.15 ± 6.87 U3L 80.65 ± 7.96 U6L 65.91 ± 7.25 U3R 80.66 ± 8.07 U6R 66.31 ± 6.52 Average
Postoperative result (T1)
Min
Max
Mean ± SD
21.40 43.67 44.68 41.69 44.70 40.06
38.07 58.21 56.42 53.45 58.09 53.84
32.59 51.03 51.35 47.95 51.22 47.87
0.07 0.02 9.04 21.74 8.95 20.89
72.58 75.62 67.21 54.92 68.86 57.58
8.34 4.51 19.40 30.22 20.15 29.06
95.24 99.34 96.49 82.37 97.89 84.37
2.1 1.63 14.49 24.7 15.53 24.45
82.21 85.5 80.14 65.37 79.82 64.99
± ± ± ± ± ±
± ± ± ± ± ±
± ± ± ± ± ±
5.07 4.12 3.77 3.97 3.77 3.86
2.57 1.21 3.36 2.00 2.9 1.8
6.07 7.14 8.03 7.45 8.35 6.36
Min 21.12 44.40 45.30 41.57 45.60 42.71
0.07 0.02 7.46 21.41 9.64 21.22
71.07 72.87 64.79 52.31 64.92 55.16
Difference (T1eT0) Max 40.68 58.67 57.85 55.04 58.42 55.40
7.90 4.00 19.26 28.44 19.81 28.48
93.93 101.69 96.49 82.91 97.00 83.58
Mean ± SD
PCC p Value
1.45 1.32 1.34 1.34 1.42 1.48 1.39
± ± ± ± ± ± ±
1.33 1.20 0.95 0.76 1.28 1.18 1.11
0.42 0.36 0.29 0.32 0.25 0.25 0.20
0.99
0.83 0.84 0.98 1.03 1.30 1.02 0.96
± ± ± ± ± ± ±
0.47 0.48 0.64 0.56 1.44 0.59 0.60
0.92 0.63 0.61 0.60 0.77 0.79 0.96
0.99
1.78 2.15 2.11 2.12 2.34 2.24 2.12
± ± ± ± ± ± ±
1.63 1.59 1.66 1.67 2.29 2.15 1.82
0.52 0.77 0.84 0.82 0.75 0.52 0.54
0.97
Max, maximum; Min, minimum; PCC, Pearson correlation coefficient.
significant difference (p > 0.05), and the Pearson correlation coefficient showed a high correlation between the two measurement tasks (PCC > 0.7). Three patients did not have registration verification records. The mean TRE was 1.07 ± 0.18 mm for the landmark registration and 2.80 ± 0.30 mm for the surface matching registration method. The surface matching registration was not accurate, and therefore the landmark registration method was chosen for patient registration to the navigation system. For the accuracy between simulation and postoperative images, the lowest mean differences (T1eT0) was observed in the validation points to MSP (0.96 ± 0.60 mm), followed by FHP (1.39 ± 1.11 mm) and COP (2.12 ± 1.82 mm) (Table 3). The differences were not statistically significant (p > .05). The Pearson correlation coefficient showed a high correlation between virtual surgical simulation (T0) and actual postoperative (T1) result (PCC > 0.7). The surgeons and patients were satisfied with the surgical outcome during follow-up. 4. Discussion The model experiment revealed high accuracy with low registration errors between the navigation system and the model. The hard tissue landmarks were more accurate than the surface scanning and the fiducial marker method. Using the landmark method, the mean differences from the simulation to surgical outcome were less than 1 mm in this study. The results from the experiment indicate that the system is accurate for navigation and transferring virtual planning to actual surgery, and could meet the requirements for clinical application (Lübbers et al., 2012). There was no clear minimum requirement of accuracy for orthognathic surgery navigation in the literature. A 1.5-mm difference was used as a clinically acceptable standard (Sun et al., 2013a), which was described as unnoticeable discrepancy by the naked eye (Lübbers et al., 2012). Compared with previous registration methods, Sun et al. presented an in vitro study using six anatomical landmarks as registration points with 85 target points in different maxillofacial areas and had a mean TRE of 0.93 mm. The accuracy decreased in the regions
farther away from the registration points. Therefore, the authors suggested that surgeons take the mean TRE and the marginal accuracy of 2 mm into consideration (Sun et al., 2013b). Luebbers et al. reported a synthetic skull model registered into the VectorVision® navigation system, and obtained an average precision of 1 mm regardless of the registration methods (Lübbers et al., 2008). Hernandez-Alfaro et al. presented a study using a CAD/CAM splint to reposition the mandible, and reported its average difference between the virtual plan and the actual operation with the greatest ndezdiscrepancy of 1.26 mm in the vertical dimension (Herna Alfaro and Guijarro-Martínez, 2013). The accuracy of our experimental model surgery was in agreement with that in previous studies. Therefore this navigation system can be reliably applied for single-splint, two-jaw orthognathic surgery. In the clinical situation, the TRE of landmark registration method slightly increased as compared with the model surgery (1.07 ± 0.18 mm vs. 0.6 ± 0.27 mm). This error was acceptable without sacrificing surgical precision. It was found that the hard tissue landmarks of the anterior nasal spine and dental cusps were convenient for the registration task. This study showed that the surface matching registration method was not accurate for patients undergoing orthognathic surgery, a finding supporting those in previous studies. The differences between the simulation and actual postoperative images (T1eT0) did not show significant change, indicating acceptable transferring of the virtual surgery (Table 3). However, it was observed that the differences in the real surgery were larger than those in the model surgery (i.e., 0.96 ± 0.60 mm vs. 0.87 ± 0.96 mm to midesagittal plane, 1.39 ± 1.11 mm vs. 0.92 ± 0.84 mm to the Frankfort horizontal plane, and 2.12 ± 1.82 mm vs. 0.89 ± 0.53 mm to the coronal plane). The postoperative CBCT was taken in 1 week and suggested possible displacement of the maxillary segment during this period. The displacement was mild posterior and inferior movement, and may reflect the soft tissue constraint causing the relapse. It is to be noted that intermaxillary fixation was not applied in these patients due to discomfort and airway concerns. Outliers were actually found in
1126
H.-W. Chang et al. / Journal of Cranio-Maxillo-Facial Surgery 43 (2015) 1119e1126
our postoperative evaluation, causing the aberration of the differences. An outlier (patient 19) had 3 standard deviations away from the mean difference in the anteroposterior dimension (to COP), and a greater than 3-mm difference from the UI and A point in the vertical dimension (to FHP). Another two outliers (patients 5 and 7) had 2 standard deviations from the mean value in the horizontal and vertical dimensions. Postoperative maxillary midline was well maintained in this study group, as shown from the negligible difference to midesagittal plane. As such, facial midline and symmetry were maintained. Compared with conventional surgical planning, computerassisted surgical simulation helps to place the maxillary and mandibular segments in the ideal position. The “ideal position: refers not only to maintaining adequate dental occlusion, level, and midline, but also to achieving esthetic outcomes in terms of symmetry and proportion. To transfer the surgical plan, some authors reported experiences in navigating orthognathic surgery. A navigated orthognathic surgery for a patient with hemi-mandibular hyperplasia by a preplanned mirror image for guiding was performed (Bell, 2011). An additional patient with class II malocclusion underwent navigation-assisted surgery combined with intraoperative CT scanning before and after maxillary reposition (Bell, 2011). A “waferless” orthognathic surgery with an interactive image-guided visualization display and the headset registration method to navigate the position of the maxilla was implemented (Zinser et al., 2013a). The in vivo management achieved the precision less than 1 mm in both the anteroposterior and mediolateral dimensions, but decreased in the vertical dimension (Zinser et al., 2013a, 2013b; Mischkowski et al., 2006). The Stryker eNlite navigation system was applied and obtained good reproducibility as compared with conventional positioning using surgical splints (Mazzoni et al., 2010). Our method supports the concept of intraoperative navigation, and we found it convenient and useful for single-splint, two-jaw surgery. There are concerns using this technique when compared with conventional orthognathic surgery. Additional time is required for preoperative preparation, including image processing, virtual operation, and preregistration of landmarks. During the surgery, the setup of the navigation system involves reference tracker application, registering the patient's position, as well as validation. If positioning guides are used, computer-aided design and manufacturing has to be performed. The cost of the hardware, software, and materials need to be taken into consideration for the simulation and navigation of orthognathic surgery. However, a better treatment outcome is obtained, especially for patients with complicated deformity and asymmetry. In our experience, postoperative facial symmetry and harmony have much been improved (Lin et al., 2015). The cost and time are expected to be reduced in the future because of improvements in technology and experience. 5. Conclusion This study showed that the navigation system was accurate and helpful for two-jaw orthognathic surgery using the single-splint method. Conflict of interest There is no conflict of interest in regard to this work. Acknowledgment We would like to thank Drs. Chiung-Shing Huang, Ellen WenChing Ko and Dr. Betty Chien-Jung Pai for the orthodontic
treatment and surgical planning, Dr. Wei-Min Yang for making the palatal plate with fiducial markers, Mr. Keven Song for help of intraoperative registration using the BrainLab, and Miss Pei-Ju Lin for the help of statistical analysis. The study was supported by a grant from Chang Gung Memorial Hospital, CMRPG381601-3, and a grant from Ministry of Science and Technology, MOST 103–2314-B182-028-MY3. References Bell RB: Computer planning and intraoperative navigation in orthognathic surgery. J Oral Maxillofac Surg 69: 592e605, 2011 Bettschart C, Kruse A, Matthews F, Zemann W, Obwegeser JA, Gr€ atz KW, et al: Pointto-point registration with mandibulo-maxillary splint in open and closed jaw position. Evaluation of registration accuracy for computer-aided surgery of the mandible. J Craniomaxillofac Surg 40: 592e598, 2012 Hoffmann J, Westendorff C, Leitner C, Bartz D, Reinert S: Validation of 3D-laser surface registration for image-guided cranio-maxillofacial surgery. J Craniomaxillofac Surg 33: 13e18, 2005 ndez-Alfaro F, Guijarro-Martínez R: New protocol for three-dimensional Herna surgical planning and CAD/CAM splint generation in orthognathic surgery: an in vitro and in vivo study. Int J Oral Maxillofac Surg 42: 1547e1556, 2013 Lo LJ, Marsh JL, Vannier MW, Patel VV: Craniofacial computer-assisted surgical planning and simulation. Clin Plast Surg 21: 501e516, 1994 €tz KW, Kruse A: A simple and Lübbers HT, Obwegeser JA, Matthews F, Eyrich G, Gra flexible concept for computer-navigated surgery of the mandible. J Oral Maxillofac Surg 69: 924e930, 2011a Lübbers HT, Matthews F, Zemann W, Gr€ atz KW, Obwegeser JA, Bredell M: Registration for computer-navigated surgery in edentulous patients: a problem based decision concept. J Craniomaxillofac Surg 39: 453e458, 2011b Lübbers HT, Medinger L, Kruse AL, Gr€ atz KW, Obwegeser JA, Matthews F: The influence of involuntary facial movements on craniofacial anthropometry: a survey using a three-dimensional photographic system. Br J Oral Maxillofac Surg 50: 171e175, 2012 Luebbers HT1, Messmer P, Obwegeser JA, Zwahlen RA, Kikinis R, Graetz KW, et al: Comparison of different registration methods for surgical navigation in craniomaxillofacial surgery. J Craniomaxillofac Surg 36: 109e116, 2008 Lin HH, Chang HW, Wang CH, Kim SG, Lo LJ: Three-dimensional computer-assissted orthognathic surgery: experience of 37 patients. Ann Plast Surg 74 Suppl. 2: S118e126, 2015 €ller JE: Application of an Mischkowski RA, Zinser MJ, Kübler AC, Krug B, Seifert U, Zo augmented reality tool for maxillary positioning in orthognathic surgery-a feasibility study. J Craniomaxillofac Surg 34: 478e483, 2006 €ller JE: Intraoperative Mischkowski RA, Zinser MJ, Ritter L, Neugebauer J, Keeve E, Zo navigation in the maxillofacial area based on 3D imaging obtained by a conebeam device. Int J Oral Maxillofac Surg 36: 687e694, 2007 Marmulla R, Mühling J, Lüth T, Hassfeld S: Physiological shift of facial skin and its influence on the change in precision of computer-assisted surgery. Br J Oral Maxillofac Surg 44: 273e278, 2006 Mazzoni S, Badiali G, Lancellotti L, Babbi L, Bianchi A, Marchetti C: Simulationguided navigation: a new approach to improve intraoperative threedimensional reproducibility during orthognathic surgery. J Craniofac Surg 21: 1698e1705, 2010 Polley JW, Figueroa AA: Orthognathic positioning system: intraoperative system to transfer virtual surgical plan to operating field during orthognathic surgery. J Oral Maxillofac Surg 71: 911e920, 2013 Shim BK, Shin HS, Nam SM, Kim YB: Real-time navigation-assisted orthognathic surgery. J Craniofac Surg 24: 221e225, 2013 Sun Y, Luebbers HT, Agbaje JO, Schepers S, Vrielinck L, Lambrichts I, et al: Evaluation of 3 different registration techniques in image-guided bimaxillary surgery. J Craniofac Surg 24: 1095e1099, 2013a Sun Y, Luebbers HT, Agbaje JO, Schepers S, Vrielinck L, Lambrichts I, et al: Validation of anatomical landmarks-based registration for image-guided surgery: an invitro study. J Craniomaxillofac Surg 41: 522e526, 2013b Yu CC, Bergeron L, Lin CH, Chu YM, Chen YR: Single-splint technique in orthognathic surgery: intraoperative checkpoints to control facial symmetry. Plast Reconstr Surg 124: 879e886, 2009 € ller JE: Computer-assisted orthognathic Zinser MJ, Mischkowski RA, Sailer HF, Zo surgery: feasibility study using multiple CAD/CAM surgical splints. Oral Surg Oral Med Oral Pathol Oral Radiol 113: 673e687, 2012 € ller JE: Zinser MJ, Mischkowski RA, Dreiseidler T, Thamm OC, Rothamel D, Zo Computer-assisted orthognathic surgery: waferless maxillary positioning, versatility and accuracy of an image-guided visualization display. Br J Oral Maxillofac Surg 51: 827e833, 2013a €ller JE: A paradigm shift in Zinser MJ, Sailer HF, Ritter L, Braumann B, Maegele M, Zo orthognathic surgery? A comparison of navigation, computer-aided designed/ computer-aided manufactured splints, and “classic” intermaxillary splints to surgical transfer of virtual orthognathic planning. J Oral Maxillofac Surg 71: 2151e1ee21, 2013b
