CRANIOMAXILLOFACIAL DEFORMITIES/COSMETIC SURGERY
Virtual Surgical Planning for Orthognathic Surgery Using Digital Data Transfer and an Intraoral Fiducial Marker: The Charlotte Method Sam Bobek, DMD, MD,* Brian Farrell, DDS, MD,y Chris Choi, DDS, MD,z Bart Farrell, DDS, MD,x Katie Weimer, MS,k and Myron Tucker, DDS{ Purpose:
We describe an alternative workup protocol for virtual surgical planning of orthognathic surgery using an intraoral fiducial marker, clinical photography, and the digital transfer of occlusal data. We also discuss our initial experience using this protocol in a series of patients.
Patients and Methods:
A retrospective cohort study was performed of consecutive patients who had undergone combined maxillary and mandibular osteotomies for the correction of dentofacial deformities at 1 center. These patients underwent treatment planning using the modified virtual surgical planning protocol described in the present report. The primary outcome evaluated was the accuracy of the method, which was determined through superimposition of the surgical plan to the postoperative cone-beam computed tomography (CBCT) scan. The secondary outcomes included the accuracy of the natural head position readings and the adequacy of the CBCT scanned stone models for the fabrication of occlusal splints.
Results:
The population included 25 patients. The root mean standard deviation (RMSD) from the preoperative plan to the postoperative scan at the maxillary cephalometric points was 1.2, 1.4, and 2.1 mm in the axial, sagittal, and coronal planes, respectively. The RMSD of the superimposed plan to the postoperative scan at the 3 mandibular cephalometric points was 1.2, 0.8, and 0.7 mm in the axial, sagittal, and coronal planes, respectively. The average variance from the axial, sagittal, and coronal planes for the natural head position was 0.05, 2.22, and 0.69 mm, respectively. All splints fabricated from the CBCT occlusal data fit the stone models and were used intraoperatively. In the subset of patients whose models were both digitally transferred and laser scanned, the superimposition of the laser scan data to the CBCT scanned data was found to have a maximum variation of 0.2 mm at the occlusal level.
Conclusions: The use of an intraoral fiducial marker changed the workflow for the data collection needed for virtual surgical planning of the correction of dentofacial deformities, while still obtaining accurate results. Because the device does not cause lip distortion, the possibility of virtually predicting a more expectant postoperative lip position exists without the need for additional scans. Furthermore, this work flow allows the transfer of data to be isolated to digital media. Ó 2015 American Association of Oral and Maxillofacial Surgeons J Oral Maxillofac Surg -:1-16, 2015
*Surgeon, Swedish Medical Center, Seattle, WA.
Address correspondence and reprint requests to Dr Bobek: Swed-
yFellowship Director, Carolinas Center for Oral & Facial Surgery,
ish Medical Center, 1221 Madison Street, Arnold Pavilion, Suite 1523,
Charlotte, NC. zSurgeon, Inland Empire Oral & Maxillofacial Surgeons, Rancho
Seattle, WA 98104; e-mail: [email protected] Received April 15 2014
Cucamonga, CA.
Accepted December 5 2014
xSurgeon, Carolinas Center for Oral & Facial Surgery, Charlotte, NC.
Ó 2015 American Association of Oral and Maxillofacial Surgeons
kSenior Manager, Medical Modeling, Golden, CO.
0278-2391/14/01799-6
{Surgeon, Carolinas Center for Oral & Facial Surgery, Charlotte,
http://dx.doi.org/10.1016/j.joms.2014.12.008
NC.
1
2
VIRTUAL SURGICAL PLANNING USING THE CHARLOTTE METHOD
FIGURE 1. Lip distortion with A, an external fiducial marker, B, removal of the external fiducial marker digitally, and C, an internal fiducial marker. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
With the original published protocol of computerassisted surgical simulation, Xia et al1 helped change the philosophy of the orthognathic surgical workup. Additional studies were published that touted the benefits of computer-assisted surgical simulation2-4 and verified the accuracy of the method.5 The components of the original method of computer-assisted surgical simulation (CASS) were clinical examination and photography, alginate impressions producing stone models, a bite jig for recording a centric relation (CR) bite that protrudes through the lips for attachment of extraoral fiducial markers, and 3 natural head position (NHP) readings attained using a gyroscope. Once those components of the workup have been completed, a computed tomography (CT) scan of the patient in their CR bite with the attached fiducial
markers is obtained. Finally, the stone models are used to establish the final bite. This bite is registered on the casts with reference lines and can be indexed with an occlusal wafer. The stone models, final bite, fiducial markers, CR bite, NHP readings, and copied patient CT scan are all sent to a planning company to be used for surgical planning.1 The models are then laser scanned, and that digital data are merged with the patient scan using the fiducial marker to enhance the occlusal representation to plan the orthognathic surgery and fabricate the occlusal splints for transfer of the virtual plan to the operating room. When using the traditional workup method for orthognathic surgery, the lip position will be distorted by the bite jig (Fig 1). The lip position is a crucial aspect in planning orthognathic surgery, and significant
BOBEK ET AL
3
FIGURE 2. Intraoral fiducial marker. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
distortion of the preoperative position will make accurate prediction of the postoperative position impossible and hinders additional evolution of imaging programs. To avoid this shortcoming, an intraoral fiducial marker was developed. The use of this fiducial marker necessitated changes in the existing algorithm for virtual planning and subsequently allowed a more streamlined and efficient method of data collection. The intraoral fiducial marker required an additional change in the work flow—all-digital transfer of the data to the planning company. This method has been described in detail, along with our experience using this workup technique in 25 consecutive patients.
Technique Description During the preoperative visit, 1 to 2 weeks before surgery, clinical photographs and 4 alginate dental impressions are obtained. The clinical photographs should include a frontal and lateral photograph of the patient in the natural head position. These clinical photographs are the basis for natural head position in this technique. A CR wax bite is then created using medium base plate wax (Shur Wax, Modern Materials) folded to encase the intraoral fiducial marker (Medical Modeling, Golden, CO; Fig 2) within the warmed wax. To minimize lip distortion, the excess wax should be trimmed from the buccal and labial surfaces (Fig 3). The bite is cooled and checked for accuracy. For the purposes of our report, this bite has been referred to as the fiducial bite. To obtain an additional NHP reading, a technique similar to that of Damstra et al6 is used with the lines established using a laser light level (GLL2-10, Bosch, Gerlingen, Germany) projected onto the standing patient with their eyes closed (Fig 4). Once the laser is not over the patient’s eyes, they are instructed to look in the distance at an object on eye level. Four
FIGURE 3. Wax centric relation bite registration with intraoral fiducial marker (VSP) included within. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
radiopaque stickers (IZI Medical Products, Owings Mills, MD) are then placed on the patient’s face along the true horizontal and true vertical reference lines. Two radiopaque markers should be placed on the vertical reference line (approximately glabella and menton) and two on the horizontal reference line (malar eminences). The fiducial bite is then returned to the patient’s mouth, and a CBCT (i-CAT Next Generation, Imaging Sciences International, Hatfield PA) scan is obtained to record the existing dentofacial deformity in CR with the intraoral fiducial and radiopaque skin markers for NHP capture within the scan’s field of view (Fig 5). The patient is discharged from the clinic once the data collection has been completed. The alginate impressions are poured with die stone (gypsum ISO type 4, Snap-Stone, Whip-Mix, Louisville, KY). The highest quality models will be ‘‘hand articulated’’ haptically to establish the desired final bite. The cuspal reduction is performed as needed, and the reduced cusps are marked. The models are then mounted on a Galetti articulator (Kerr USA, Orange, CA) in a final bite directed by the surgeon. A final bite is indexed using warmed base plate wax (Shur Wax, Modern Materials). If segmental surgery is indicated, the cast should be sectioned and secured in the final position with wax and stone. CBCT scans of the stone models are then performed (Fig 6). Depending on the planned surgery, 3 or 4 scans should be obtained of the stone models: 1) the dental models alone to record the occlusal anatomy; 2) the dental models positioned in the fiducial bite; 3) the dental models in a final wax bite; and 4) the segmented model (if segmental surgery is planned). The CBCT settings used are listed in Table 1. These digital data for the model scans are sent, in conjunction with the patient
4
VIRTUAL SURGICAL PLANNING USING THE CHARLOTTE METHOD
FIGURE 4. Obtaining the natural head position (NHP). Clinical photography is the basis for the NHP in this technique. In addition to this technique, radiopaque markers, placed on the patient’s face along the true horizontal and true vertical lines, were used. Note that 6 markers were used in this photograph and that 2 had been inaccurately placed. Only 4 markers are needed to obtain the NHP. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
CBCT scan, to the planning company (in this case, Medical Modeling) using a secure Internet site. No stone models or wax bites are sent in the mail to the planning company. The digital data received by the service center is processed to remove scatter, and the models are re-created virtually (Fig 7). The fiducial marker is used to merge the skeletal data from the patient scan with an accurate occlusal record from the model scans (Fig 8). This virtual patient with detailed occlusal anatomy is oriented in the NHP by aligning the radiopaque stickers to the virtual horizontal, vertical, and sagittal planes. This will generate an accurate clinical, skeletal, and dental representation of the
existing clinical situation. The surgical order form (outlining the planned surgical movements and types of osteotomies) is sent digitally to allow the biomedical engineers familiar with the planning software the opportunity to prepare for the pending virtual planning session (Fig 9). In a web meeting, the surgeon verifies the head position compared with the clinical photographs, making adjustments as needed. If fiducial markers have not been used, the head position should be determined by the surgeon from the clinical photographs. The surgeon should then guide the engineers experienced with the software to correct the dentofacial
5
BOBEK ET AL
FIGURE 5. Patient scan capturing the dentofacial deformity, intraoral fiducial marker within the centric relation bite, and radiopaque skin markers for the natural head position with the field of view. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
deformity and malocclusion in multiple dimensions.1,5 Once the virtual planning has been completed, virtual occlusal splints are fabricated using additive manufacturing and used to transfer the virtual plan to the patient. The occlusal splints sent to the surgeon by mail for surgical intervention are the only physical information transferred in the modified
workflow. The summarized workflow is presented in Figure 10.
Patients and Methods We retrospectively reviewed the data from 25 consecutive patients who had undergone combined
Table 1. I-CAT (IMAGING SCIENCES INTERNATIONAL) SETTINGS USED FOR EACH CBCT SCAN
CBCT Scan
Variable Patients Models Models with VSP bite Models with final bite Segmentalized cast*
Field of Voxel Duration View (cm) Size (mm) (s) 23 17 88 88 88 88
0.3 0.2 0.2 0.2 0.2
17.8 26.9 26.9 26.9 26.9
FIGURE 6. Models oriented in cone-beam computed tomography scanner with wax bite in place.
Abbreviation: CBCT, cone-beam computed tomography. * If needed.
Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
6
VIRTUAL SURGICAL PLANNING USING THE CHARLOTTE METHOD
FIGURE 7. Reconstruction of virtual occlusion using digital transfer of model scans. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
maxillary and mandibular osteotomies in 2012 (June 1, 2012 to August 15, 2012) at the Carolinas Center for Oral & Facial Surgery. The institutional review board granted an exception for the present study through the Western Institutional Review Board (Puyallup, WA). The exclusion criteria were total joint reconstruction and single jaw osteotomies. Details of the study population are presented in Table 2. All patients underwent virtual surgical planning using Dolphin software (Chatsworth, CA) through Medical Modeling (Golden, CO), using data obtained as detailed in the preceding section. The positioning of the NHP was checked visually by the surgeon before planning the case by comparing the virtual patient
FIGURE 8. Merger using superimposition of the fiducial skeletal positioning with accurate occlusal records. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
head position with the clinical photographs. If any discrepancy existed, the head position noted on the clinical photography was used. If no fiducial markers had been used for the NHP, the head position was set using the clinical photographs by the surgeon. The virtual occlusion and the condylar position were also evaluated by the surgeon. If they were acceptable, the virtual surgery was then performed during a webinar session between the surgeon operating the case and an engineer. The virtual surgical plan was performed using the intermediate and final occlusal splint combined with an external reference point (glabella pin) used for verification of the vertical dimension. In all cases, mandibular surgery was performed first. Postoperative CBCT scans were completed for all patients in the first postoperative week. All information gathered was performed by the lead authors (S.B., C.C.), with the aid of Excel (Microsoft, Redmond, WA), preserving the patients’ anonymity. The postoperative scans were uploaded to Medical Modeling, and the postoperative scan was compared against the virtually planned surgery. The surgical plan was aligned to the postoperative anatomy using best fit surface alignment to the specific anatomy, depending on the maxilla or mandibular alignment. For the maxilla, the midface was used for the best fit surface. The mandible was aligned by the best fit of the condylar head position and the vertical position of the mandibular incisors. The superimpositions were performed by 1 engineer (C.B.). Three cephalometric points (midline upper incisor, mesiobuccal cusp of both first molars) were plotted on the anatomy of the surgical plan and the postoperative scan. These cephalometric points were used to calculate the centroids for each object (maxilla and mandible) to allow alignment of the planned and actual outcomes. The
7
BOBEK ET AL
FIGURE 9. Creation of virtual osteotomies, mimicking outcomes anticipated clinically. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
centroids were then tracked as the scans were allowed to return to their unpaired positions, and the discrepancies between the planned and actual results were evaluated (Figs 11 to 13). The 2 statistical methods used to evaluate the difference between the planned and actual result were intentionally similar to that described by Hsu et al5 to present data similar to those from the published technique. The first method was to calculate the RMSDs of the maxilla and mandible in the mediolateral, anteroposterior, and superoinferior directions.
This absolute value of the difference in the linear centroid position was used to evaluate the accuracy of the surgical plan in all directions. The second method was used to assess the measurement agreement using the Bland-Altman technique.7 A lack of agreement between the planned result and the actual result was estimated using the mean differences, standard deviations, and 95% confidence intervals. The 95% confidence intervals were computed for both the upper and lower limits of agreement. For both the RMSD and the Bland-Altman agreement
FIGURE 10. Work flow of the patient workup used in the present study. CBCT, cone-beam computed tomography; Lab, laboratory; NHP, natural head position; Preop, preoperatively. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
8
VIRTUAL SURGICAL PLANNING USING THE CHARLOTTE METHOD
Table 2. STUDY POPULATION
Variable Gender Male Female Diagnosis Asymmetry Maxillary hypoplasia Mandibular hypoplasia Vertical maxillary excess Mandibular hyperplasia Maxillary osteotomy 3-Piece Le Fort 2-Piece Le Fort 1-Piece Le Fort Mandibular osteotomy BSSO IVRO
Patients (n)
8 17 15 13 7 3 6 8 2 15 24 1
Diagnoses summed to greater than the study population because the patients had multiple diagnoses. Abbreviations: BSSO, bilateral sagittal split osteotomy; IVRO, intraoral vertical ramus osteotomy. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
this step. A superimposition of the virtual occlusal surfaces (digital and laser scanned physical models) was performed by 1 engineer (C.B.). The 10 superimpositions were compared quantitatively (Fig 14). In addition, the splints fabricated from the CBCT digitally transferred occlusion were verified for fit on the actual stone models.
Results All 25 surgeries were completed using the intermediate and final splints created using the virtual surgical workup. The absolute mean RMSD at the 3 maxillary and 3 mandibular points are presented in Table 3. All values, with the exception of the vertical position of the maxilla (2.1 mm), were not clinically different (ie, they were 3 MM)
Pt. No. Variable
18
10
9
Planned mean vertical position of maxilla (mm) Postoperative mean vertical position of maxilla (mm) Difference (mm) Significant clinical diagnoses
2.2
2.6
3.0
2.8
6.1
6.1
5.0 Asymmetric Class III open bite; maxillary hypoplasia; previously repaired cleft palate/cleft lip
3.5 Skeletal Class II deep bite; vertical maxillary insufficiency
3.1 Skeletal Class II deep bite; vertical maxillary insufficiency
Negative results correspond to inferior positioning of the maxilla. Abbreviation: Pt. No., patient number. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
14
VIRTUAL SURGICAL PLANNING USING THE CHARLOTTE METHOD
Table 6. DIFFERENCE IN PLANNED POSITION OF MAXILLA VERSUS ACTUAL POSITION IN 3 PLANES OF SPACE (CENTROIDS) FOR PATIENTS 9, 10, AND 18
Patient 18 Plane Mediolateral Anteroposterior Superoinferior
Patient 10
Patient 9
Uncorrected
Corrected
Uncorrected
Corrected
Uncorrected
Corrected
1.4 3.0 5.0
1.0 1.2 0.1
0.2 1.6 3.5
0.3 0.4 1.5
0.1 1.7 3.0
0.2 0.1 0.4
Positive values in mediolateral indicate more leftward shift. These patients had the largest variation in the vertical dimension. A correction of the vertical position of the maxilla (central incisors aligned vertically) was performed using the axis of rotation of the condyles (which would be the case in mandibular first surgery). The influence of this correction is shown in all 3 planes of space. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
FIGURE 15. Color map of superimposed preoperative plan to postoperative cone-beam computed tomography scan indicating variance at all anatomic points in the A, maxilla and B, distal segment of the mandible. Legend of map indicates differences in millimeters. (Fig 15 continued on next page.) Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
15
BOBEK ET AL
FIGURE 15 (cont’d). Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
positioning of the maxillary segments intraoperatively. Accurate positioning of segmental maxillary surgery is dependent on the ability of the surgeon to seat the segments within the occlusal splint. If minor errors occur, occlusal discrepancies can result. The final consideration, and arguably the most important, was our relatively small patient population. Any surgical result that was an outlier will skew the results. We identified 3 patients who had the most deviation in the vertical position of the maxilla. One patient (patient 18) was a patient with a cleft and maxillary hypoplasia, apertognathia, and mandibular asymmetry. The other 2 patients (patients 9 and 10) were twins with a Class II deep bite and vertical maxillary insufficiency. The deviation from vertical in these 3 patients is presented in Table 5. For the patient with
the cleft, the more downward position of the final maxilla can be explained by the surgical difficulty in achieving the planned maxillary advancement of 10 mm. The outcome of this patient also resulted in the outlier for the anteroposterior positional difference in the maxilla, with 7 mm of advancement achieved (3-mm difference from surgical plan). The scar tissue prevented full rotation of the mandible, and a vertical position more inferior than planned was accepted. For the twins, the amount of inferior positioning of the maxilla (3 mm) was not thought to be enough, and the surgical plan was changed intraoperatively. Although it could be argued that these patients should have been excluded from the present study, we believed that their inclusion illustrates the
16
VIRTUAL SURGICAL PLANNING USING THE CHARLOTTE METHOD
challenges encountered in clinical orthognathic surgical practice and demonstrates the surgical shortcomings in achieving a surgical plan. We performed a second superimposition of the maxilla in these patients, assuming that the final vertical position of the maxilla was placed as planned. The results of this superimposition are presented in Table 6. Our analysis of the accuracy of the surgical results to the surgical plans was limited to 3 occlusal cephalometric points. Our purpose was to compare our data to that found in the report by Hsu et al.5 Certainly, more points than these are clinically relevant and should be considered in evaluating the accuracy of this method. For Figure 15, a color map was created for a representative mandible and maxilla that illustrates the variation seen at, not only the occlusal level, but also at the bony level. This type of analysis was not completed for all patients owing to time constraints and the statistical difficulty in quantifying the differences. Future studies evaluating the accuracy of bony landmarks would better characterize the accuracy of this method. Finally, an important difference in our report compared with the report by Hsu et al5 was the investigator who had performed the superimpositions. In our study, an engineer from Medical Modeling performed this key evaluation of the data. This is a source of bias that should not be overlooked. Because the statistical analysis was performed by the senior author and the results were less accurate than those published, we believed this source of bias to be minor. In conclusion, the incorporation of an intraoral fiducial marker into the CASS protocol necessitated multiple changes in the workflow of data acquisition. These changes included an alternative NHP reading and scanning the dental models in a CBCT scanner. The protocol accounted for NHP using clinical photographs and radiographic fiducial markers aligned to the true vertical and horizontal lines. The inability to laser scan the fiducial marker was overcome by performing a CBCT scan of the patients’ dental models and of the dental models in the fiducial and final bite. This ‘‘Charlotte protocol’’ was successfully used for 25 patients. This protocol eliminates the significant lip distortion in the planning CBCT scan and initiates the all-digital transfer of data to the planning company. Other than the surgeon-controlled variable
of the vertical position of the maxilla, no clinically significant variation from the surgical plan was seen. The upper and lower limits of the positional difference were found to have clinically significant differences from the virtual surgical plan. This can be explained largely by surgical shortcomings in a small patient population. Acknowledgments The authors thank Chris Beaudreau (Medical Modeling) for performing the superimpositions.
References 1. Xia JJ, Gateno J, Teichgraeber JF: New clinical protocol to evaluate craniomaxillofacial deformity and plan surgical correction. J Oral Maxillofac Surg 67:2094, 2009 2. Ritter L: Registration accuracy of three-dimensional surface and cone beam computed tomography data for virtual implant planning. Clin Oral Implants Res 23:447, 2012 3. Plooij JM: Digital three-dimensional image fusion processes for planning and evaluating orthodontics and orthognathic surgery: A systematic review. Int J Oral Maxillofac Surg 40: 341, 2011 4. Zisner MJ: Computer-assisted orthognathic surgery: Feasibility study using multiple CAD/CAM surgical splints. Oral Surg Oral Med Oral Pathol Oral Radiol 113:673, 2012 5. Hsu SS, Gateno J, Bell RB, et al: Accuracy of a computer aided surgical simulation protocol for orthognathic surgery: A prospective multicenter study. J Oral Maxillofac Surg 71: 128, 2013 6. Damstra J, Fourie Z, Ren Y: Simple technique to achieve a natural position of the head for cone beam computed tomography. Br J Oral Maxillofac Surg 48:236, 2010 7. Bland JM, Altman DG: Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1:307, 1986 8. Tarazona B, Llamas JM, Cibrian R, et al: A comparison between dental measurements taken from CBCT models and those taken from a digital method. Eur J Orthod 35:1, 2013 9. Zizelmann C, Hammer B, Gellrich NC, et al: An evaluation of face bow transfer for the planning of orthognathic surgery. J Oral Maxillofac Surg 70:1944, 2012 10. Agbaje J, Sun Y, Lambrichts I, Politis C: Problems during orthognathic surgery resulting from errors in diagnostic wax bite. J Craniofac Surg 24:999, 2013 11. Bae MJ, Kim JY, Park JT, et al: Accuracy of miniscrew surgical guides assessed from cone-beam computed tomography and digital models. Am J Orthod Dentofacial Orthop 143:893, 2013 12. Dreiseidler T, Tandon D, Kreppel M, et al: CBCT device dependency on the transfer accuracy from computer aided implantology procedures. Clin Oral Implants Res 23:10897, 2012 13. Lightbeart KG, English JD, Kau CH, et al: Surface analysis of study models generated from OrthoCAD and cone-beam computed tomography imaging. Am J Orthod Dentofacial Orthop 141:686, 2012
Virtual Surgical Planning for Orthognathic Surgery Using Digital Data Transfer and an Intraoral Fiducial Marker: The Charlotte Method Sam Bobek, DMD, MD,* Brian Farrell, DDS, MD,y Chris Choi, DDS, MD,z Bart Farrell, DDS, MD,x Katie Weimer, MS,k and Myron Tucker, DDS{ Purpose:
We describe an alternative workup protocol for virtual surgical planning of orthognathic surgery using an intraoral fiducial marker, clinical photography, and the digital transfer of occlusal data. We also discuss our initial experience using this protocol in a series of patients.
Patients and Methods:
A retrospective cohort study was performed of consecutive patients who had undergone combined maxillary and mandibular osteotomies for the correction of dentofacial deformities at 1 center. These patients underwent treatment planning using the modified virtual surgical planning protocol described in the present report. The primary outcome evaluated was the accuracy of the method, which was determined through superimposition of the surgical plan to the postoperative cone-beam computed tomography (CBCT) scan. The secondary outcomes included the accuracy of the natural head position readings and the adequacy of the CBCT scanned stone models for the fabrication of occlusal splints.
Results:
The population included 25 patients. The root mean standard deviation (RMSD) from the preoperative plan to the postoperative scan at the maxillary cephalometric points was 1.2, 1.4, and 2.1 mm in the axial, sagittal, and coronal planes, respectively. The RMSD of the superimposed plan to the postoperative scan at the 3 mandibular cephalometric points was 1.2, 0.8, and 0.7 mm in the axial, sagittal, and coronal planes, respectively. The average variance from the axial, sagittal, and coronal planes for the natural head position was 0.05, 2.22, and 0.69 mm, respectively. All splints fabricated from the CBCT occlusal data fit the stone models and were used intraoperatively. In the subset of patients whose models were both digitally transferred and laser scanned, the superimposition of the laser scan data to the CBCT scanned data was found to have a maximum variation of 0.2 mm at the occlusal level.
Conclusions: The use of an intraoral fiducial marker changed the workflow for the data collection needed for virtual surgical planning of the correction of dentofacial deformities, while still obtaining accurate results. Because the device does not cause lip distortion, the possibility of virtually predicting a more expectant postoperative lip position exists without the need for additional scans. Furthermore, this work flow allows the transfer of data to be isolated to digital media. Ó 2015 American Association of Oral and Maxillofacial Surgeons J Oral Maxillofac Surg -:1-16, 2015
*Surgeon, Swedish Medical Center, Seattle, WA.
Address correspondence and reprint requests to Dr Bobek: Swed-
yFellowship Director, Carolinas Center for Oral & Facial Surgery,
ish Medical Center, 1221 Madison Street, Arnold Pavilion, Suite 1523,
Charlotte, NC. zSurgeon, Inland Empire Oral & Maxillofacial Surgeons, Rancho
Seattle, WA 98104; e-mail: [email protected] Received April 15 2014
Cucamonga, CA.
Accepted December 5 2014
xSurgeon, Carolinas Center for Oral & Facial Surgery, Charlotte, NC.
Ó 2015 American Association of Oral and Maxillofacial Surgeons
kSenior Manager, Medical Modeling, Golden, CO.
0278-2391/14/01799-6
{Surgeon, Carolinas Center for Oral & Facial Surgery, Charlotte,
http://dx.doi.org/10.1016/j.joms.2014.12.008
NC.
1
2
VIRTUAL SURGICAL PLANNING USING THE CHARLOTTE METHOD
FIGURE 1. Lip distortion with A, an external fiducial marker, B, removal of the external fiducial marker digitally, and C, an internal fiducial marker. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
With the original published protocol of computerassisted surgical simulation, Xia et al1 helped change the philosophy of the orthognathic surgical workup. Additional studies were published that touted the benefits of computer-assisted surgical simulation2-4 and verified the accuracy of the method.5 The components of the original method of computer-assisted surgical simulation (CASS) were clinical examination and photography, alginate impressions producing stone models, a bite jig for recording a centric relation (CR) bite that protrudes through the lips for attachment of extraoral fiducial markers, and 3 natural head position (NHP) readings attained using a gyroscope. Once those components of the workup have been completed, a computed tomography (CT) scan of the patient in their CR bite with the attached fiducial
markers is obtained. Finally, the stone models are used to establish the final bite. This bite is registered on the casts with reference lines and can be indexed with an occlusal wafer. The stone models, final bite, fiducial markers, CR bite, NHP readings, and copied patient CT scan are all sent to a planning company to be used for surgical planning.1 The models are then laser scanned, and that digital data are merged with the patient scan using the fiducial marker to enhance the occlusal representation to plan the orthognathic surgery and fabricate the occlusal splints for transfer of the virtual plan to the operating room. When using the traditional workup method for orthognathic surgery, the lip position will be distorted by the bite jig (Fig 1). The lip position is a crucial aspect in planning orthognathic surgery, and significant
BOBEK ET AL
3
FIGURE 2. Intraoral fiducial marker. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
distortion of the preoperative position will make accurate prediction of the postoperative position impossible and hinders additional evolution of imaging programs. To avoid this shortcoming, an intraoral fiducial marker was developed. The use of this fiducial marker necessitated changes in the existing algorithm for virtual planning and subsequently allowed a more streamlined and efficient method of data collection. The intraoral fiducial marker required an additional change in the work flow—all-digital transfer of the data to the planning company. This method has been described in detail, along with our experience using this workup technique in 25 consecutive patients.
Technique Description During the preoperative visit, 1 to 2 weeks before surgery, clinical photographs and 4 alginate dental impressions are obtained. The clinical photographs should include a frontal and lateral photograph of the patient in the natural head position. These clinical photographs are the basis for natural head position in this technique. A CR wax bite is then created using medium base plate wax (Shur Wax, Modern Materials) folded to encase the intraoral fiducial marker (Medical Modeling, Golden, CO; Fig 2) within the warmed wax. To minimize lip distortion, the excess wax should be trimmed from the buccal and labial surfaces (Fig 3). The bite is cooled and checked for accuracy. For the purposes of our report, this bite has been referred to as the fiducial bite. To obtain an additional NHP reading, a technique similar to that of Damstra et al6 is used with the lines established using a laser light level (GLL2-10, Bosch, Gerlingen, Germany) projected onto the standing patient with their eyes closed (Fig 4). Once the laser is not over the patient’s eyes, they are instructed to look in the distance at an object on eye level. Four
FIGURE 3. Wax centric relation bite registration with intraoral fiducial marker (VSP) included within. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
radiopaque stickers (IZI Medical Products, Owings Mills, MD) are then placed on the patient’s face along the true horizontal and true vertical reference lines. Two radiopaque markers should be placed on the vertical reference line (approximately glabella and menton) and two on the horizontal reference line (malar eminences). The fiducial bite is then returned to the patient’s mouth, and a CBCT (i-CAT Next Generation, Imaging Sciences International, Hatfield PA) scan is obtained to record the existing dentofacial deformity in CR with the intraoral fiducial and radiopaque skin markers for NHP capture within the scan’s field of view (Fig 5). The patient is discharged from the clinic once the data collection has been completed. The alginate impressions are poured with die stone (gypsum ISO type 4, Snap-Stone, Whip-Mix, Louisville, KY). The highest quality models will be ‘‘hand articulated’’ haptically to establish the desired final bite. The cuspal reduction is performed as needed, and the reduced cusps are marked. The models are then mounted on a Galetti articulator (Kerr USA, Orange, CA) in a final bite directed by the surgeon. A final bite is indexed using warmed base plate wax (Shur Wax, Modern Materials). If segmental surgery is indicated, the cast should be sectioned and secured in the final position with wax and stone. CBCT scans of the stone models are then performed (Fig 6). Depending on the planned surgery, 3 or 4 scans should be obtained of the stone models: 1) the dental models alone to record the occlusal anatomy; 2) the dental models positioned in the fiducial bite; 3) the dental models in a final wax bite; and 4) the segmented model (if segmental surgery is planned). The CBCT settings used are listed in Table 1. These digital data for the model scans are sent, in conjunction with the patient
4
VIRTUAL SURGICAL PLANNING USING THE CHARLOTTE METHOD
FIGURE 4. Obtaining the natural head position (NHP). Clinical photography is the basis for the NHP in this technique. In addition to this technique, radiopaque markers, placed on the patient’s face along the true horizontal and true vertical lines, were used. Note that 6 markers were used in this photograph and that 2 had been inaccurately placed. Only 4 markers are needed to obtain the NHP. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
CBCT scan, to the planning company (in this case, Medical Modeling) using a secure Internet site. No stone models or wax bites are sent in the mail to the planning company. The digital data received by the service center is processed to remove scatter, and the models are re-created virtually (Fig 7). The fiducial marker is used to merge the skeletal data from the patient scan with an accurate occlusal record from the model scans (Fig 8). This virtual patient with detailed occlusal anatomy is oriented in the NHP by aligning the radiopaque stickers to the virtual horizontal, vertical, and sagittal planes. This will generate an accurate clinical, skeletal, and dental representation of the
existing clinical situation. The surgical order form (outlining the planned surgical movements and types of osteotomies) is sent digitally to allow the biomedical engineers familiar with the planning software the opportunity to prepare for the pending virtual planning session (Fig 9). In a web meeting, the surgeon verifies the head position compared with the clinical photographs, making adjustments as needed. If fiducial markers have not been used, the head position should be determined by the surgeon from the clinical photographs. The surgeon should then guide the engineers experienced with the software to correct the dentofacial
5
BOBEK ET AL
FIGURE 5. Patient scan capturing the dentofacial deformity, intraoral fiducial marker within the centric relation bite, and radiopaque skin markers for the natural head position with the field of view. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
deformity and malocclusion in multiple dimensions.1,5 Once the virtual planning has been completed, virtual occlusal splints are fabricated using additive manufacturing and used to transfer the virtual plan to the patient. The occlusal splints sent to the surgeon by mail for surgical intervention are the only physical information transferred in the modified
workflow. The summarized workflow is presented in Figure 10.
Patients and Methods We retrospectively reviewed the data from 25 consecutive patients who had undergone combined
Table 1. I-CAT (IMAGING SCIENCES INTERNATIONAL) SETTINGS USED FOR EACH CBCT SCAN
CBCT Scan
Variable Patients Models Models with VSP bite Models with final bite Segmentalized cast*
Field of Voxel Duration View (cm) Size (mm) (s) 23 17 88 88 88 88
0.3 0.2 0.2 0.2 0.2
17.8 26.9 26.9 26.9 26.9
FIGURE 6. Models oriented in cone-beam computed tomography scanner with wax bite in place.
Abbreviation: CBCT, cone-beam computed tomography. * If needed.
Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
6
VIRTUAL SURGICAL PLANNING USING THE CHARLOTTE METHOD
FIGURE 7. Reconstruction of virtual occlusion using digital transfer of model scans. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
maxillary and mandibular osteotomies in 2012 (June 1, 2012 to August 15, 2012) at the Carolinas Center for Oral & Facial Surgery. The institutional review board granted an exception for the present study through the Western Institutional Review Board (Puyallup, WA). The exclusion criteria were total joint reconstruction and single jaw osteotomies. Details of the study population are presented in Table 2. All patients underwent virtual surgical planning using Dolphin software (Chatsworth, CA) through Medical Modeling (Golden, CO), using data obtained as detailed in the preceding section. The positioning of the NHP was checked visually by the surgeon before planning the case by comparing the virtual patient
FIGURE 8. Merger using superimposition of the fiducial skeletal positioning with accurate occlusal records. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
head position with the clinical photographs. If any discrepancy existed, the head position noted on the clinical photography was used. If no fiducial markers had been used for the NHP, the head position was set using the clinical photographs by the surgeon. The virtual occlusion and the condylar position were also evaluated by the surgeon. If they were acceptable, the virtual surgery was then performed during a webinar session between the surgeon operating the case and an engineer. The virtual surgical plan was performed using the intermediate and final occlusal splint combined with an external reference point (glabella pin) used for verification of the vertical dimension. In all cases, mandibular surgery was performed first. Postoperative CBCT scans were completed for all patients in the first postoperative week. All information gathered was performed by the lead authors (S.B., C.C.), with the aid of Excel (Microsoft, Redmond, WA), preserving the patients’ anonymity. The postoperative scans were uploaded to Medical Modeling, and the postoperative scan was compared against the virtually planned surgery. The surgical plan was aligned to the postoperative anatomy using best fit surface alignment to the specific anatomy, depending on the maxilla or mandibular alignment. For the maxilla, the midface was used for the best fit surface. The mandible was aligned by the best fit of the condylar head position and the vertical position of the mandibular incisors. The superimpositions were performed by 1 engineer (C.B.). Three cephalometric points (midline upper incisor, mesiobuccal cusp of both first molars) were plotted on the anatomy of the surgical plan and the postoperative scan. These cephalometric points were used to calculate the centroids for each object (maxilla and mandible) to allow alignment of the planned and actual outcomes. The
7
BOBEK ET AL
FIGURE 9. Creation of virtual osteotomies, mimicking outcomes anticipated clinically. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
centroids were then tracked as the scans were allowed to return to their unpaired positions, and the discrepancies between the planned and actual results were evaluated (Figs 11 to 13). The 2 statistical methods used to evaluate the difference between the planned and actual result were intentionally similar to that described by Hsu et al5 to present data similar to those from the published technique. The first method was to calculate the RMSDs of the maxilla and mandible in the mediolateral, anteroposterior, and superoinferior directions.
This absolute value of the difference in the linear centroid position was used to evaluate the accuracy of the surgical plan in all directions. The second method was used to assess the measurement agreement using the Bland-Altman technique.7 A lack of agreement between the planned result and the actual result was estimated using the mean differences, standard deviations, and 95% confidence intervals. The 95% confidence intervals were computed for both the upper and lower limits of agreement. For both the RMSD and the Bland-Altman agreement
FIGURE 10. Work flow of the patient workup used in the present study. CBCT, cone-beam computed tomography; Lab, laboratory; NHP, natural head position; Preop, preoperatively. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
8
VIRTUAL SURGICAL PLANNING USING THE CHARLOTTE METHOD
Table 2. STUDY POPULATION
Variable Gender Male Female Diagnosis Asymmetry Maxillary hypoplasia Mandibular hypoplasia Vertical maxillary excess Mandibular hyperplasia Maxillary osteotomy 3-Piece Le Fort 2-Piece Le Fort 1-Piece Le Fort Mandibular osteotomy BSSO IVRO
Patients (n)
8 17 15 13 7 3 6 8 2 15 24 1
Diagnoses summed to greater than the study population because the patients had multiple diagnoses. Abbreviations: BSSO, bilateral sagittal split osteotomy; IVRO, intraoral vertical ramus osteotomy. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
this step. A superimposition of the virtual occlusal surfaces (digital and laser scanned physical models) was performed by 1 engineer (C.B.). The 10 superimpositions were compared quantitatively (Fig 14). In addition, the splints fabricated from the CBCT digitally transferred occlusion were verified for fit on the actual stone models.
Results All 25 surgeries were completed using the intermediate and final splints created using the virtual surgical workup. The absolute mean RMSD at the 3 maxillary and 3 mandibular points are presented in Table 3. All values, with the exception of the vertical position of the maxilla (2.1 mm), were not clinically different (ie, they were 3 MM)
Pt. No. Variable
18
10
9
Planned mean vertical position of maxilla (mm) Postoperative mean vertical position of maxilla (mm) Difference (mm) Significant clinical diagnoses
2.2
2.6
3.0
2.8
6.1
6.1
5.0 Asymmetric Class III open bite; maxillary hypoplasia; previously repaired cleft palate/cleft lip
3.5 Skeletal Class II deep bite; vertical maxillary insufficiency
3.1 Skeletal Class II deep bite; vertical maxillary insufficiency
Negative results correspond to inferior positioning of the maxilla. Abbreviation: Pt. No., patient number. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
14
VIRTUAL SURGICAL PLANNING USING THE CHARLOTTE METHOD
Table 6. DIFFERENCE IN PLANNED POSITION OF MAXILLA VERSUS ACTUAL POSITION IN 3 PLANES OF SPACE (CENTROIDS) FOR PATIENTS 9, 10, AND 18
Patient 18 Plane Mediolateral Anteroposterior Superoinferior
Patient 10
Patient 9
Uncorrected
Corrected
Uncorrected
Corrected
Uncorrected
Corrected
1.4 3.0 5.0
1.0 1.2 0.1
0.2 1.6 3.5
0.3 0.4 1.5
0.1 1.7 3.0
0.2 0.1 0.4
Positive values in mediolateral indicate more leftward shift. These patients had the largest variation in the vertical dimension. A correction of the vertical position of the maxilla (central incisors aligned vertically) was performed using the axis of rotation of the condyles (which would be the case in mandibular first surgery). The influence of this correction is shown in all 3 planes of space. Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
FIGURE 15. Color map of superimposed preoperative plan to postoperative cone-beam computed tomography scan indicating variance at all anatomic points in the A, maxilla and B, distal segment of the mandible. Legend of map indicates differences in millimeters. (Fig 15 continued on next page.) Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
15
BOBEK ET AL
FIGURE 15 (cont’d). Bobek et al. Virtual Surgical Planning Using the Charlotte Method. J Oral Maxillofac Surg 2015.
positioning of the maxillary segments intraoperatively. Accurate positioning of segmental maxillary surgery is dependent on the ability of the surgeon to seat the segments within the occlusal splint. If minor errors occur, occlusal discrepancies can result. The final consideration, and arguably the most important, was our relatively small patient population. Any surgical result that was an outlier will skew the results. We identified 3 patients who had the most deviation in the vertical position of the maxilla. One patient (patient 18) was a patient with a cleft and maxillary hypoplasia, apertognathia, and mandibular asymmetry. The other 2 patients (patients 9 and 10) were twins with a Class II deep bite and vertical maxillary insufficiency. The deviation from vertical in these 3 patients is presented in Table 5. For the patient with
the cleft, the more downward position of the final maxilla can be explained by the surgical difficulty in achieving the planned maxillary advancement of 10 mm. The outcome of this patient also resulted in the outlier for the anteroposterior positional difference in the maxilla, with 7 mm of advancement achieved (3-mm difference from surgical plan). The scar tissue prevented full rotation of the mandible, and a vertical position more inferior than planned was accepted. For the twins, the amount of inferior positioning of the maxilla (3 mm) was not thought to be enough, and the surgical plan was changed intraoperatively. Although it could be argued that these patients should have been excluded from the present study, we believed that their inclusion illustrates the
16
VIRTUAL SURGICAL PLANNING USING THE CHARLOTTE METHOD
challenges encountered in clinical orthognathic surgical practice and demonstrates the surgical shortcomings in achieving a surgical plan. We performed a second superimposition of the maxilla in these patients, assuming that the final vertical position of the maxilla was placed as planned. The results of this superimposition are presented in Table 6. Our analysis of the accuracy of the surgical results to the surgical plans was limited to 3 occlusal cephalometric points. Our purpose was to compare our data to that found in the report by Hsu et al.5 Certainly, more points than these are clinically relevant and should be considered in evaluating the accuracy of this method. For Figure 15, a color map was created for a representative mandible and maxilla that illustrates the variation seen at, not only the occlusal level, but also at the bony level. This type of analysis was not completed for all patients owing to time constraints and the statistical difficulty in quantifying the differences. Future studies evaluating the accuracy of bony landmarks would better characterize the accuracy of this method. Finally, an important difference in our report compared with the report by Hsu et al5 was the investigator who had performed the superimpositions. In our study, an engineer from Medical Modeling performed this key evaluation of the data. This is a source of bias that should not be overlooked. Because the statistical analysis was performed by the senior author and the results were less accurate than those published, we believed this source of bias to be minor. In conclusion, the incorporation of an intraoral fiducial marker into the CASS protocol necessitated multiple changes in the workflow of data acquisition. These changes included an alternative NHP reading and scanning the dental models in a CBCT scanner. The protocol accounted for NHP using clinical photographs and radiographic fiducial markers aligned to the true vertical and horizontal lines. The inability to laser scan the fiducial marker was overcome by performing a CBCT scan of the patients’ dental models and of the dental models in the fiducial and final bite. This ‘‘Charlotte protocol’’ was successfully used for 25 patients. This protocol eliminates the significant lip distortion in the planning CBCT scan and initiates the all-digital transfer of data to the planning company. Other than the surgeon-controlled variable
of the vertical position of the maxilla, no clinically significant variation from the surgical plan was seen. The upper and lower limits of the positional difference were found to have clinically significant differences from the virtual surgical plan. This can be explained largely by surgical shortcomings in a small patient population. Acknowledgments The authors thank Chris Beaudreau (Medical Modeling) for performing the superimpositions.
References 1. Xia JJ, Gateno J, Teichgraeber JF: New clinical protocol to evaluate craniomaxillofacial deformity and plan surgical correction. J Oral Maxillofac Surg 67:2094, 2009 2. Ritter L: Registration accuracy of three-dimensional surface and cone beam computed tomography data for virtual implant planning. Clin Oral Implants Res 23:447, 2012 3. Plooij JM: Digital three-dimensional image fusion processes for planning and evaluating orthodontics and orthognathic surgery: A systematic review. Int J Oral Maxillofac Surg 40: 341, 2011 4. Zisner MJ: Computer-assisted orthognathic surgery: Feasibility study using multiple CAD/CAM surgical splints. Oral Surg Oral Med Oral Pathol Oral Radiol 113:673, 2012 5. Hsu SS, Gateno J, Bell RB, et al: Accuracy of a computer aided surgical simulation protocol for orthognathic surgery: A prospective multicenter study. J Oral Maxillofac Surg 71: 128, 2013 6. Damstra J, Fourie Z, Ren Y: Simple technique to achieve a natural position of the head for cone beam computed tomography. Br J Oral Maxillofac Surg 48:236, 2010 7. Bland JM, Altman DG: Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1:307, 1986 8. Tarazona B, Llamas JM, Cibrian R, et al: A comparison between dental measurements taken from CBCT models and those taken from a digital method. Eur J Orthod 35:1, 2013 9. Zizelmann C, Hammer B, Gellrich NC, et al: An evaluation of face bow transfer for the planning of orthognathic surgery. J Oral Maxillofac Surg 70:1944, 2012 10. Agbaje J, Sun Y, Lambrichts I, Politis C: Problems during orthognathic surgery resulting from errors in diagnostic wax bite. J Craniofac Surg 24:999, 2013 11. Bae MJ, Kim JY, Park JT, et al: Accuracy of miniscrew surgical guides assessed from cone-beam computed tomography and digital models. Am J Orthod Dentofacial Orthop 143:893, 2013 12. Dreiseidler T, Tandon D, Kreppel M, et al: CBCT device dependency on the transfer accuracy from computer aided implantology procedures. Clin Oral Implants Res 23:10897, 2012 13. Lightbeart KG, English JD, Kau CH, et al: Surface analysis of study models generated from OrthoCAD and cone-beam computed tomography imaging. Am J Orthod Dentofacial Orthop 141:686, 2012
