- Email: [email protected]
A new navigation system based on cephalograms and dental casts for oral and maxillofacial surgery
Int. J. Oral Maxillofac. Surg. 2006; 35: 828–836 doi:10.1016/j.ijom.2006.02.024, available online at http://www.sciencedirect.com
Research Paper Computer Assisted Surgery
A new navigation system based on cephalograms and dental casts for oral and maxillofacial surgery
M. Tsuji, N. Noguchi, M. Shigematsu, Y. Yamashita, K. Ihara, M. Shikimori, M. Goto Department of Oral and Maxillofacial Surgery, Saga Medical School, 5-1-1 Nabeshima, Saga 849-0937, Japan
M. Tsuji, N. Noguchi, M. Shigematsu, Y. Yamashita, K. Ihara, M. Shikimori, M. Goto: A new navigation system based on cephalograms and dental casts for oral and maxillofacial surgery. Int. J. Oral Maxillofac. Surg. 2006; 35: 828–836. # 2006 International Association of Oral and Maxillofacial Surgeons. Published by Elsevier Ltd. All rights reserved. Abstract. Intraoperative navigation systems help surgeons to accurately carry out preoperative plans without injuring anatomically important structures. A system is evaluated that uses cephalograms instead of computed tomographic (CT) scans to create images. Three-dimensional (3D) dental casts provide registration between imaging data and the patient. Cephalograms are widely employed in orthognathic and oral and maxillofacial surgery and expose patients to lower doses of radiation than CT. The system uses a dental cast to register the operation field to a pair of frontal and lateral cephalograms. The cast is transformed to 3D data with a laser scanner and a programme that runs on a personal computer. 3D data describing the dental cast, cephalograms and the oral and maxillofacial region of the patient are integrated with specialized software. The optical tracking system for navigation uses charged-coupled-device (CCD) video cameras and light-emitting diodes (LEDs). Two CCD video cameras follow the 3D coordinates of LED assemblies attached to the head, lower jaw and a handpiece. Errors occurring when a dental cast was transformed to 3D data ranged from 0.08 to 0.21 mm. Mean errors were 0.71 mm (0.21–1.09 mm) for the right maxillary central incisor, 0.62 mm (0.04– 1.69 mm) for the right maxillary 2nd molar and 1.02 mm (0.23–1.47 mm) for the left maxillary 2nd molar. This surgical navigation system is sufficiently accurate for use in oral and maxillofacial surgery.
Navigation systems enable surgeons to preoperatively simulate surgical procedures and avoid injury to anatomically important structures, such as the mandibular canal, maxillary sinus and orbit. Three-dimen0901-5027/090828 + 09 $30.00/0
sional (3D) stereolithographic models and surgical simulation software are now widely used in dentistry and medicine, including oral and maxillofacial surgery6,7,13,14,16. Surgical simulation has
Key words: computer-aided surgery; navigation system; cephalograms; dental implantology. Accepted for publication 27 February 2006 Available online 9 May 2006
greatly facilitated jaw reconstruction after tumour resection, dental implant placement and orthognathic surgery. Accurate realization of preoperative plans, however, developed with the use of these tools requires a
# 2006 International Association of Oral and Maxillofacial Surgeons. Published by Elsevier Ltd. All rights reserved.
Navigation system using cephalograms
829
Fig. 1. First, integration of cephalograms and the patient’s dental cast was performed. Then, the dental cast and patient were registered. Finally, registration of cephalogram to the patient was performed via the dental cast, instead of using ‘markers’.
reliable navigation system. In oral and maxillofacial surgery, navigation systems are especially useful for dental implant placement, orthognathic surgery, endoscopic surgery and removal of foreign bodies. Computed tomography (CT) is often used for computer-aided surgery in the oral and maxillofacial region, including the placement of dental implants2,27–31. Currently, CT equipment is available only at universities and major hospitals. In contrast, cephalograms can be obtained even at smaller hospitals and clinics. Other advantages of cephalograms over CT include lower cost, a simpler procedure and much lower exposure to radiation. In addition, cephalograms are standardized, allowing easy comparison of traced images, a feature that CT lacks. Since light is employed, most optical tracking systems are marginally less precise than mechanical techniques, which directly measure objects17,24. Nevertheless, optical systems facilitate the transfer of preoperative simulation data and are unaffected by patients’ anatomical characteristics and surgical complexity2,27,29–31. Optical tracking also allows real-time feedback that can be used, for example, to guide a drill during dental implant placement. If anatomic features differ from the preoperative simulation, optical navigation systems allow plans to be immediately revised, enabling surgical procedures to be more precisely performed. In this study is evaluated an optically based surgical navigation system that uses cephalograms to create images and a dental cast to provide registration between imaging data and the patient.
Materials and methods
The system under study uses frontal and lateral cephalograms to create imaging data and 3D dental casts to provide registration between the imaging data and the patient, unlike navigation systems that use ‘markers’ for registration (Fig. 1)1,3,9,31. Charge-coupled-device (CCD) cameras (KP-M1, Hitachi Denshi, Tokyo, Japan) and light-emitting diodes (LEDs) were used for the optical tracking system. The system uses 2 CCD cameras to follow the 3D coordinates of LED assemblies attached to the head, the mental region of the patient and a handpiece. We fixed LED assemblies to the patient’s head with special headgear and to the mental region with a mouthpiece fixed on the mandibular teeth. If the patient lacked adequate teeth in the mandible, the LED assembly was fixed directly to the jawbone. Movement
of the mandible as well as of the maxilla was tracked. Three LEDs were mounted on LED assemblies at the vertices of an isosceles triangle. To assess the accuracy of the system, an LED assembly was attached to the handpiece of a drill, and the drill tip used as a probe, to enable realtime navigation without switching between the probe and a tool. We used a laser scanner (VMS-250R, UNISN, Osaka, Japan) to transform the 3D morphological data of dental casts into digital data (Fig. 2). Frontal and lateral cephalograms were also digitized with a scanner (GT-9600, Seiko-Epson, Nagano, Japan). A scale was included in the cephalograms to allow their pixel size to be adjusted on the computer. 3D morphological data describing dental casts of the upper and lower jaws were integrated with the cephalogram data using computer software (OMF, UNISN,
Fig. 2. 3D morphological data describing the teeth of the patient were obtained from a dental cast by a laser scanner.
830
Tsuji et al.
Fig. 3. 3D data of the dental cast of the maxilla were integrated with cephalograms using computer software. Three points were selected on the teeth of the dental cast. These points formed a triangular plane on the display (blue triangle).
Osaka, Japan). Data integration was done with a projection-matching technique22 (Figs 3 and 4). To integrate the data, 3 points were selected on the dental cast of the upper jaw that were readily identifiable on cephalograms, e.g. median edge of the right maxillary central incisor, mesiobuccal cusp tip of the right maxillary 2nd
molar, and buccal cusp tip of the left maxillary 1st premolar. Each point was then input into a computer. These points formed a triangular plane on the screen (Fig. 3). If the intended morphological point was not identifiable, markers were cemented on the teeth and impressions taken of the teeth with the markers. Then, 3 other points in locations corresponding
Fig. 4. After selecting 3 points on the dental cast, the corresponding points on the cephalograms were marked in the software to form a triangular plane. 3D integration of the dental cast and the frontal and lateral cephalograms was performed by matching these 2 planes.
to those selected on the dental cast were selected on frontal and lateral cephalograms, forming a similar triangular plane. The images from the dental cast of the upper jaw and the frontal/lateral cephalograms were integrated by aligning these 2 triangular planes by means of a least squares technique (Fig. 4). We scanned upper and lower jaw models separately after fixing them in centric occlusion by using an occlusal record and a special tool. Centric occlusion of the jaw models could be reproduced by the computer software. The dental cast of the lower jaw was also automatically integrated with the cephalograms. Planar registration decreased errors during integration as compared to point-to-point registration. Navigational surgery requires registration of the cephalogram to the patient, for which dental casts were used (Fig. 1). The technique described above was used to integrate the dental casts and cephalograms. A separate ‘integration’ step was unnecessary because dental casts created by taking impressions closely match patients’ teeth. For surgical navigation, designated points were displayed on cephalograms by aligning the coordinates of the patient with those of the cephalogram. This was accomplished by registering the coordinates of the dental casts and the patient’s teeth in separate spaces (Fig. 1). Finally, a drill tip was used to indicate 3 points on the patient’s teeth in locations corresponding to those used for integration of the dental cast and cephalograms. A pair of CCD cameras detected an LED assembly on the handpiece, and the coordinates of the 3 points were calculated as 3D data based on the relative position between the LED assembly and the drill tip. The points were displayed as a triangle on a computer screen. Although the triangle derived from these points is congruent to that constructed by integrating the dental cast and cephalograms, these 2 triangles have different coordinate systems, skewed to each other. The procedure described above, however, can determine the positional relation between the skewed congruent triangles. Coordinates on the patient and cephalograms can be registered by aligning these triangles with the use of a coordinate transformation matrix created by Affine transformation (Fig. 5). First, the points located on the patient were transformed to a dental cast using a coordinate transformation matrix, and therefore projected onto frontal and lateral cephalograms. Navigation was thus performed in a ‘reference position’, where the patient was registered to the cephalograms
Navigation system using cephalograms
Fig. 5. Three points marked on the patient were displayed as a triangle on a computer screen. Although the triangle was congruent to that constructed by integrating a dental cast with the cephalograms, these 2 triangles had different coordinate systems, skewed to each other. The relative positional relation between the skewed congruent triangles was determined. Coordinates on the patient and cephalograms can be registered by aligning these triangles.
in real time (0.1 s delay). Points at the drill tip and calculated drill axis were displayed on images (Figs 6 and 7). The maxilla is usually used to determine 3 points on the patient’s teeth in a reference position. The mandible should be in centric occlusion to the maxilla. Cephalograms should, therefore, also be obtained in centric occlusion. Although navigation is performed in a reference position, the head can move, and the jaw position changes when the mouth
is opened and closed during surgery. Points, therefore, need to be tracked with CCD cameras via the LED assemblies attached to the head and mental region, transformed to coordinates in the reference position and displayed on cephalograms. When the head is stationary and the mandible is moved from the reference position with the drill tip positioned at a point on the mandible, the position of the
Fig. 6. LED assemblies were attached to the head, mental region and handpiece. Two CCD cameras tracked the LED, and real-time surgical navigation was performed.
831
drill tip is stored in a computer via the LED assembly attached to the drill tip and monitored by the CCD camera. The coordinates of the LED assemblies attached to the head and the mental region are simultaneously stored in the computer. The head remains stationary; therefore, its movement is 0. The CCD camera tracks the position of the LED assembly fixed to the mental region in real time. An LED is mounted on each of the 3 LED assemblies to form an isosceles triangle that is tracked by the camera. When a point on the mandibular teeth is marked with the drill tip, the relation between the triangle on the LED assembly of the mandible and the triangle in the reference position is stored in the computer. A coordinate transformation matrix, based on Affine transformation, superposes the 2 triangles of the mandible and transforms the points indicated after mandibular movement to coordinates in the reference position. The position and axis direction of the drill tip can, thus, be displayed on cephalograms created in centric occlusion (Fig. 8). When both the head and mandible move from the reference position, a coordinate transformation matrix is created in the computer from the post-movement coordinates of the LED assemblies attached to the head and mental region relative to the coordinates in the reference position. The indicated coordinate of the drill tip touching the mandible can be transformed onto an image in reference position by transforming the head LED to the reference position, followed by transformation of the mental region LED to the reference position. This allows real-time surgical navigation coordinated with movement of the head and mandible. Drills and burs of various lengths are used for surgery. When the tip position of these tools changes, the relative position of the tip to the LED assembly attached to the handpiece changes. The drill-tip position must, therefore, be calibrated against the LED assembly for each drill, using a calibrator (Fig. 9). The device used to calibrate tip position has 3 LEDs that allow the CCD camera to detect the coordinates of the drill tip. A drill tip, attached to a handpiece, is connected to the calibrator. The CCD camera detects the relative position of the drill tip between the 3 LEDs on the calibrator and the LED assembly attached to the handpiece. The computer calculates relative positions of the drill tip between these points and calibrates the position of the drill tip. The direction of the drill axis is also calculated.
832
Tsuji et al.
Fig. 7. The point of the drill tip and the drill axis were visualized on a pair of cephalograms.
Verification of accuracy
The accuracy of the LED 3D position was first evaluated by a stereo photogrammetry technique using 2 CCD cameras. A single LED was mounted on a 3-axis calibrator. Measurements were made at 5 locations 30 mm apart along each axis (125 points in total) within a cube measuring 120 mm on each side (Fig. 10). Next, the accuracy of transforming a dental cast to 3D data was verified using a preoperative dental cast of a patient who was scheduled to undergo orthognathic
surgery. Three landmarks were designated on the dental cast, which was then transformed to 3D data by computer-aided design (CAD) software (Surfacer 9.0, Imageware, USA). These points were located in the lower mesial angle on a buccal side bracket of the right maxillary 2nd molar, the lower distal angle on a labial side bracket of the right maxillary central incisor, and the lower mesial angle on a buccal side bracket of the left maxillary 2nd molar. A coordinate was measured for each point 10 times and the mean calculated. The distance measured from
the coordinate was compared with the actual distance on the dental cast. Finally, the accuracy of our surgical navigation system was verified using a phantom head as a model of a patient. A dental cast was attached to the phantom head, and 4 landmarks, differing from those on the dental cast, were designated as reference points on the face of the phantom: the right lateral canthus (point A), the right inner canthus (point B), the glabella (point C) and the nasal apex (point D). These reference points were selected so that the CCD camera could
Fig. 8. The patient’s head is fixed, and the mandible is moved from the reference position. The coordinates of the LED assemblies attached to the head and the mental region in reference position are simultaneously stored in the computer (upper left). The CCD camera tracks the position of the LED assembly fixed to the mental region in real time (upper right). When a point on the mandibular teeth is marked with the drill tip, the relation between a triangle on the LED assembly of the mandible (blue) and a triangle in the reference position (green) is stored in the computer. A coordinate transformation matrix, based on Affine transformation, superposes the 2 triangles of the mandible and transforms the points indicated after mandibular movement to coordinates in the reference position.
Navigation system using cephalograms
833
Mean errors of the CCD camera with respect to the LED position, measured with a 3D calibrator, were 0.034 mm for the X-axis, 0.031 mm for the Y-axis and 0.041 mm for the Z-axis. The maximum errors were 0.134 mm for the X-axis, 0.093 mm for the Y-axis and 0.143 mm for the Z-axis. Differences between distances among the 3 landmarks on the teeth as measured with the CAD software and the actual distances measured with calipers were used to estimate errors when the dental cast was transformed to 3D data with a laser scanner. Errors were 0.08 mm between points on the right maxillary 1st molar and central incisor, 0.21 mm between points on the right maxillary central incisor and the left maxillary 2nd molar; and 0.14 mm between points on the right maxillary 2nd molar and the left maxillary 2nd molar. To determine errors associated with the surgical navigation system, distances measured with digital calipers from points A to D on the head phantom to the dental cast attached to the head phantom were compared with the corresponding values calculated based on the coordinates stored in the system. Mean errors were 0.71 mm (0.21–1.09 mm) for the right maxillary central incisor, 0.62 mm (0.04–1.69 mm) for the right maxillary 2nd molar and 1.02 mm (0.23–1.47 mm) for the left maxillary 2nd molar (Table 1).
Fig. 9. A drill tip attached to a handpiece was connected to the calibrator.
Discussion Fig. 10. Two CCD cameras used in stereo photogrammetry technique to evaluate accuracy of LED 3D position.
easily detect the handpiece LED assembly when the drill tip pointed to these positions. Three points on the teeth were selected similarly to the procedure used to verify the accuracy of 3D transformation of the dental cast. A drill tip was pointed at the 7 coordinates described above 20 times each and the information stored in a computer. The mean of each coordinate was then calculated. The distances between points A–D and each of the landmarks on the dental cast attached to the phantom head were measured and compared to the actual distances measured with calipers. Errors were then estimated.
frontal and lateral cephalograms. The CCD camera tracked the position of the LED assembly attached to the handpiece to maintain the point of the drill tip and the drill axes in view. The point of the drill tip was re-calculated by means of a coordinate transformation matrix when the maxilla and mandible were moved from the reference position. It took 0.1 s to transform the point and display it on cephalograms. Navigation was performed on a real-time basis, concurrently with movement of the head and mandible.
Table 1. Accuracy of measuring distances from 4 landmarks on the head to 3 landmarks on teeth of phantom head
Results
Points indicated by the drill tip on the mandible and maxilla of the patient as well as the drill axes were displayed on
Dental casts represent the shape of teeth much more precisely than CT images and, therefore, provide much more precise markers for navigation systems. If readily identifiable landmarks are needed to register a patient to a radiograph via a dental cast, markers as small as 1 mm in diameter can be used as long as they are reproducible. A key benefit of this system is its relative simplicity compared to CT. Surgical navigation systems originally developed for stereotaxic cranial nerve surgery21,23 have been used for surgery of the skull base, sinuses, spine, and oral and maxillofacial region4,8,11,12,15,25,32. For maxillofacial surgery, the head is fixed in place by conventional methods; however, patients sometimes move their head
Right molar Right incisor Left molar
Point A
Point B
Point C
Point D
Mean difference (mm)
0.65 0.91 0.23
0.08 0.61 1.47
0.04 1.09 0.97
1.69 0.21 1.42
0.62 0.71 1.02
834
Tsuji et al.
during operation or open or close their mouth, moving the jaw, especially during oral surgery. Surgical navigation must, therefore, be modified to accommodate movement. This system enables real-time surgical navigation by tracking mandibular motion, as well as gross head movement, by means of an LED assembly. The LED assembly, used to track both maxillary and mandibular movement, needs to be fixed firmly to avoid dislodgement due to body movement. Dislodgment of the LED assembly decreases accuracy, even with highly precise surgical navigation systems. Headgear was used to fix an LED assembly on the head, but one could not be fixed to the mandible because the skin and muscles move too far when the mouth is opened and closed. The assembly was instead attached to a frame fixed to the teeth outside of the surgical site. This method cannot be used if patients are edentulous or have too few teeth. To obtain stable fixation in these patients, the mouthpiece for the LED assembly is fixed directly to the bone of the lower jaw, using screws designed for fixation of bone fractures. A surgical template can be used for dental implant placement to replicate the preoperative simulation during the operation5,18,26,28. This method is effective when enough teeth remain to mount the template on the bone surface. A template cannot provide good results when oral stabilization is difficult, such as in edentulous patients or in those undergoing tumour resection and reconstruction6. The precision of preoperative simulation images greatly affects operative results because re-registration and image correction are not possible during surgery. In addition, the technique depends in large part on the surgeon’s experience. A template-based method can provide a high degree of precision if the template can be firmly fixed to the surface of the jawbone. Another method to transfer preoperative simulation to the patient is mechanical transfer using special equipment during surgery17,24. Transfer accuracy is high with this method; however, 3D models used for simulation need to match the patient’s anatomy, and precise 3D images are required. Preoperative simulation and simulation transfer are carried out with devices that use 3D models. Error may be introduced during the production of these models. At present, a large space near the patient is needed to deploy the equipment in the surgical field. The application of mechanical transfer techniques is thus limited to facial fracture or orthognathic surgery.
Registration accuracy greatly affects the precision of navigation systems. When CT is used, markers need to be mounted in advance on the head and face for registration to the patient1,25,31. These markers should be large enough to compensate for the CT slice interval and width. Continuous 3D images must be constructed from intermittent image sets. Artifacts can influence the construction of 3D images, potentially increasing error. Markers attached to the skin may be displaced by body and muscle movements or swelling, which can also contribute to error10. 3D models produced on the basis of erroneous images may lead to further errors. With this navigation system, information on the cast is processed to 3D data for use in registration. These data allow preoperative simulation using CAD software, which is especially useful for osteotomies including teeth. Our system also enables the postoperative location of bone chips to be modelled with CAD software on the basis of spatial coordinates determined after displacement. Surgery can thus be evaluated in 3D. 3D images are also useful for obtaining informed consent before operation because they facilitate explanation of intended procedures to patients (Fig. 11). Precision of the CCD cameras, accuracy of transforming dental casts to 3D data, and accuracy of registration between imaging data and the patient are the main
factors affecting the performance of surgical navigation systems. The camera precision has an especially important role in the overall system. In this study, mean error was 0.035 mm and maximum error was 0.143 mm. Greater error appeared in the Z-axis than the X- or Y-axis, demonstrating that some axes or directions in space have reduced accuracy with 2-camera systems. Such inaccuracy may occur because the 2 cameras are configured to use a stereo photogrammetry technique to search space from different directions; therefore, accuracy near the centre of the field of view of both cameras is higher than that at the edges of the LED. When a navigation system is used, the surgical field needs to be located in the space providing the highest accuracy. Accuracy in recording a dental cast as 3D data influences data integration (done with computer software) as well as registration, both of which are important navigational components. Differences between dimensions obtained with the CAD software and those measured with calipers ranged from 0.08 to 0.21 mm. This result agrees with the findings of MOTOHASHI et al.19,20 (0.12 0.08 mm) who performed verification studies using an equivalent scanner. In the present study, 3D data of the dental cast were collected separately for teeth anterior and those posterior to the premolars, and then aligned in a computer. Discrepancies could have occurred during the alignment
Fig. 11. Upper panel: dentition before operation. Lower panel: dentition after operation. The dental cast is transformed to 3D data and processed using CAD software. It is possible to predict the operation outcome with this software, and to use the information to obtain informed consent from the patient29–31,2.
Navigation system using cephalograms process. In addition, distances between coordinates of corresponding points were measured to verify accuracy. Measured differences might have included errors that occurred during coordinate calculation. Another factor potentially affecting the accuracy of infrared navigation is lighting, especially the surgical light used in an operating room. If light is too bright or produces wavelengths similar to the LED, greater sampling errors occur when the CCD camera attempts to detect the LED. Preoperative registration of patients with a computerized dental cast is necessary in most operating rooms, and patients need to be fixed in a reference position during this procedure to avoid reducing the accuracy of navigation. To verify the accuracy of this navigation system, the system and digital calipers were used to measure distances from 4 points on the phantom head to 3 landmarks on the dental cast attached to the phantom head. The results were then compared. Points A–D were selected so that the CCD camera could readily detect LED locations indicated by the drill tip. When the left maxillary 2nd molar on the dental cast was indicated by the drill tip, the LED assembly attached to the handpiece showed a shorter angle toward the CCD camera than the other 2 points. In other words, when the CCD camera detected an isosceles triangle created by the LED assembly, the area of the triangle was very small as compared to other landmarks, which might have caused a detection error by the CCD camera. In addition, individual LEDs appeared to the CCD camera to be compressed to an oval shape, rather than round. Errors seem to have occurred when LED light was converted to pixels and the centroid was calculated. It appears therefore that LED-related detection errors resulted in errors in registering the drill-tip coordinates. The distance between the LED and CCD camera can lead to error. This effect was considered to be negligible in this study because the CCD cameras were positioned at optimal distances; however, some distance is necessary between the patient and camera in actual surgery because an operator stands between them. With this surgical navigation system, mean errors were 0.62 mm (SD 0.76 mm) at the right maxillary 2nd molar, 0.71 mm (SD 0.38 mm) at the right maxillary central incisor and 1.02 mm (SD 0.57 mm) at the left maxillary 2nd molar. The method used coordinates of pairs of points; the results may, therefore, have included measurement errors from both points. To reduce error, it appears that
LED assemblies attached to the patient and a handpiece must be at appropriate angles and distances with respect to the CCD camera. In conclusion, it has been shown here that this surgical navigation system provides sufficient accuracy for highly precise surgery. The system should be useful in oral and maxillofacial surgery, especially dental implant placement, endoscopic surgery and orthognathic surgery. References 1. Alp MS, Dujovny M, Misra M, Charbel FT, Ausman JI. Head registration techniques for image-guided surgery. Neurol Res 1998: 20: 31–37. 2. Birkfeller W, Huber K, Larson A, Hanson D, Diemling M, Homolka P, Bergmann H. A modular software system for computer-aided surgery and its first application in oral implantology. IEEE Trans Med Imaging 2000: 19: 616–620. 3. Birkfellner W, Watzinger F, Wanschitz F, Ewers R, Bergmann H. Calibration of tracking systems in a surgical environment. IEEE Trans Med Imaging 1998: 17: 737–742. 4. Claudius GN, Alexander S, Beat H, Sergio R, Daniel C, Wolf L, Rainer S. Computer-assisted secondary reconstruction of unilateral posttraumatic orbital deformity. Plast Reconstr Surg 2002: 110: 1417–1429. 5. Fortin T, Bosson JL, Coudert JL, Isidori M. Reliability of preoperative planning of an image-guided system for oral implant placement based on 3-dimensional images: an in vivo study. Int J Oral Maxillofac Implants 2003: 18: 886–893. 6. Fortin T, Champleboux G, Bianchi S, Buatois H, Coudert JL. Precision of transfer of preoperative planning for oral implants based on cone-beam CT-scan images through a robotic drilling machine. Clinical Oral Implants Res 2002: 13: 651–656. 7. Girod S, Keebe E, Girod B. Advance in interactive craniofacial surgery planning by 3D simulation and visualization. Int J Oral Maxillofac Surg 1995: 24: 120–125. 8. Grevers G, Leunig A, Klemens A, Hagedorn H. CAS of the paranasal sinuses—technology and clinical experience with the Vector-Vision-CompactSystem in 102 patients. Laryngorhinootologie 2002: 81: 476–483. 9. Grimson WEL, Ettinger GJ, Pe´rez TL, Wells WM, Kikinis R. An automatic registration method for frameless stereotaxy, image guided surgery, and enhanced reality visualization. IEEE Trans Med Imaging 1996: 15: 129–140. 10. Hassfeld S, Mu¨hling J. Computer assisted oral and maxillofacial surgery—a review and an assessment of technology. Int J Oral Maxillofac Surg 2001: 30: 2–13.
835
11. Ikeda K, Shoin K, Taguchi H, Yamano J, Yamashita J. Cranionavigator combining a high-speed drill and a navigation system for skull base surgery—technical note. Neurol Med Chir 1999: 39: 701– 708. 12. Itoh H, Ebara S, Kamimura M, Tateiwa Y, Kinoshita T, Yuzawa Y, Takaoka K, Ohtsuka K. Computerassisted pedicle screw insertion for surgery of spinal deformity. J Jpn Scoliosis Soc 1998: 13: 55–60. 13. Jacbs R, Adriansens A, Verstreken K, Suetens P. Predictability of a threedimensional planning system for oral implant surgery. Dentomaxillofac Radiol 1999: 28: 105–111. 14. Keeve E, Girod S, Kikinis R, Girod B. Deformable modeling of facial tissue for craniofacial surgery simulation. Comput Aided Surg 1999: 3: 228–238. 15. Khan M, Ecke U, Mann WJ. The application of an optical navigation system in endonasal sinus surgery. HNO 2003: 51: 209–215. 16. Kopp KC, Koslow AH, Abdo OS. Predictable implant placement with a diagnostic/surgical template and advanced radiographic imaging [in German]. J Prosthet Dent 2003: 89: 611–615. 17. Marmulla R, Niederdellmann H. Computer-assisted bone segment navigation. J Craniomaxillofac Surg 1998: 26: 347–359. 18. Mizrahi B, Thunthy KH, Finger I. Radiographic/surgical template incorporating metal telescopic tubes for accurate implant placement. Pract Periodont Aesthet Dent 1998: 10: 757–765. 19. Motohashi N, Kuroda A. A 3D computer-aided design system applied to diagnosis and treatment planning in orthodontics and orthognathic surgery. Eur J Orthod 1999: 21: 263–274. 20. Motohashi N, Nunota A, Kuroda T. Study on CAM system for dental model [in Japanese]. J Stomal Soc Jpn 1997: 64: 427–435. 21. Nakao N, Nakai K, Itakura T. Updating of neuronavigation based on images intraoperatively acquired with a mobile computerized tomographic scanner: technical note. Minim Invasive Neurosurg 2003: 46: 117–120. 22. Noguchi N, Goto M. Computer simulation system for orthognathic surgery. Orthod Craniofacial Res 2003: 6: 176– 178. 23. Ohue S, Kumon Y, Kohno K, Shiraishi M, Sakaki S, Kikuchi K, Miki H. Preoperative neurosurgical simulation and intraoperative navigation using threedimensional images. Jpn J Neurosurg 1999: 8: 657–663. 24. Santler G. 3-D COSMOS a new 3-D model based computerized operation simulation and navigation system. J Craniomaxillofac Surg 2000: 28: 287–293. 25. Schultes G, Ka¨rcher H, Gaggl A, Santler G. Computer assisted tumour
836
Tsuji et al.
resection of the skull base: case report. J Craniomaxillofac Surg 2001: 29: 326– 331. 26. Sicilia A, Noguerol B, Cobo J, Zabalegui I. Profile surgical template: a systematic approach to precise implant placement. A technical note. Int J Oral Maxillofac Implants 1998: 13: 109– 114. 27. Sießegger M, Schneider BT, Mischkowski RA, Lazar F, Krug B, Klesper B, Zo¨ller JE. Use of an imageguided navigation system in dental implant surgery in anatomically complex operation sites. J Craniomaxillofac Surg 2001: 29: 276–281. 28. Steenberghe D, Naert I, Andersson M, Brajnovic I, Cleynenbreugel J, Suetens P. A Custom template and definitive prosthesis allowing immediate implant loading in the maxilla: a clinical
report. Int J Oral Maxillofac Implants 2002: 17: 663–670. 29. Wagner A, Undt G, Watzinger F, Wanschitz F, Schicho K, Yerit K, Kermer C, Birkfellner W, Ewers R. Principles of computer-assisted arthroscopy of the temporomandibular joint with optoelectronic tracking technology. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2001: 92: 30–37. 30. Wagner A, Wanschits F, Birkfellner W, Zauza K, Klug C, Schicho K, Kainberger F, Czerny C, Bergmann H, Evers R. Computer-aided placement of endosseous oral implants in patients after ablative tumour surgery: assessment of accuracy. Clin Oral Implants Res 2003: 14: 340–348. 31. Watzinger F, Birkfellner W, Wanschitz F. Positioning of dental implants using computer-aided naviga-
tion and an optical tracking system: case report and presentation of a new method. J Craniomaxillofac Surg 1999: 27: 77–81. 32. Watzinger F, Wanschitz F, Wanger A, Enislidis G, Millesi W, Baumann A, Ewers R. Computer-aided navigation in secondary reconstruction of post-traumatic deformities of the zygoma. J Craniomaxillofac Surg 1997: 25: 198–202. Address: Mitsuhiro Tsuji Department of Oral and Maxillofacial Surgery Saga Medical School 5-1-1 Nabeshima Saga 849-0937 Japan Tel: +81 952342397 Fax: +81 952342044 E-mail: [email protected]
Research Paper Computer Assisted Surgery
A new navigation system based on cephalograms and dental casts for oral and maxillofacial surgery
M. Tsuji, N. Noguchi, M. Shigematsu, Y. Yamashita, K. Ihara, M. Shikimori, M. Goto Department of Oral and Maxillofacial Surgery, Saga Medical School, 5-1-1 Nabeshima, Saga 849-0937, Japan
M. Tsuji, N. Noguchi, M. Shigematsu, Y. Yamashita, K. Ihara, M. Shikimori, M. Goto: A new navigation system based on cephalograms and dental casts for oral and maxillofacial surgery. Int. J. Oral Maxillofac. Surg. 2006; 35: 828–836. # 2006 International Association of Oral and Maxillofacial Surgeons. Published by Elsevier Ltd. All rights reserved. Abstract. Intraoperative navigation systems help surgeons to accurately carry out preoperative plans without injuring anatomically important structures. A system is evaluated that uses cephalograms instead of computed tomographic (CT) scans to create images. Three-dimensional (3D) dental casts provide registration between imaging data and the patient. Cephalograms are widely employed in orthognathic and oral and maxillofacial surgery and expose patients to lower doses of radiation than CT. The system uses a dental cast to register the operation field to a pair of frontal and lateral cephalograms. The cast is transformed to 3D data with a laser scanner and a programme that runs on a personal computer. 3D data describing the dental cast, cephalograms and the oral and maxillofacial region of the patient are integrated with specialized software. The optical tracking system for navigation uses charged-coupled-device (CCD) video cameras and light-emitting diodes (LEDs). Two CCD video cameras follow the 3D coordinates of LED assemblies attached to the head, lower jaw and a handpiece. Errors occurring when a dental cast was transformed to 3D data ranged from 0.08 to 0.21 mm. Mean errors were 0.71 mm (0.21–1.09 mm) for the right maxillary central incisor, 0.62 mm (0.04– 1.69 mm) for the right maxillary 2nd molar and 1.02 mm (0.23–1.47 mm) for the left maxillary 2nd molar. This surgical navigation system is sufficiently accurate for use in oral and maxillofacial surgery.
Navigation systems enable surgeons to preoperatively simulate surgical procedures and avoid injury to anatomically important structures, such as the mandibular canal, maxillary sinus and orbit. Three-dimen0901-5027/090828 + 09 $30.00/0
sional (3D) stereolithographic models and surgical simulation software are now widely used in dentistry and medicine, including oral and maxillofacial surgery6,7,13,14,16. Surgical simulation has
Key words: computer-aided surgery; navigation system; cephalograms; dental implantology. Accepted for publication 27 February 2006 Available online 9 May 2006
greatly facilitated jaw reconstruction after tumour resection, dental implant placement and orthognathic surgery. Accurate realization of preoperative plans, however, developed with the use of these tools requires a
# 2006 International Association of Oral and Maxillofacial Surgeons. Published by Elsevier Ltd. All rights reserved.
Navigation system using cephalograms
829
Fig. 1. First, integration of cephalograms and the patient’s dental cast was performed. Then, the dental cast and patient were registered. Finally, registration of cephalogram to the patient was performed via the dental cast, instead of using ‘markers’.
reliable navigation system. In oral and maxillofacial surgery, navigation systems are especially useful for dental implant placement, orthognathic surgery, endoscopic surgery and removal of foreign bodies. Computed tomography (CT) is often used for computer-aided surgery in the oral and maxillofacial region, including the placement of dental implants2,27–31. Currently, CT equipment is available only at universities and major hospitals. In contrast, cephalograms can be obtained even at smaller hospitals and clinics. Other advantages of cephalograms over CT include lower cost, a simpler procedure and much lower exposure to radiation. In addition, cephalograms are standardized, allowing easy comparison of traced images, a feature that CT lacks. Since light is employed, most optical tracking systems are marginally less precise than mechanical techniques, which directly measure objects17,24. Nevertheless, optical systems facilitate the transfer of preoperative simulation data and are unaffected by patients’ anatomical characteristics and surgical complexity2,27,29–31. Optical tracking also allows real-time feedback that can be used, for example, to guide a drill during dental implant placement. If anatomic features differ from the preoperative simulation, optical navigation systems allow plans to be immediately revised, enabling surgical procedures to be more precisely performed. In this study is evaluated an optically based surgical navigation system that uses cephalograms to create images and a dental cast to provide registration between imaging data and the patient.
Materials and methods
The system under study uses frontal and lateral cephalograms to create imaging data and 3D dental casts to provide registration between the imaging data and the patient, unlike navigation systems that use ‘markers’ for registration (Fig. 1)1,3,9,31. Charge-coupled-device (CCD) cameras (KP-M1, Hitachi Denshi, Tokyo, Japan) and light-emitting diodes (LEDs) were used for the optical tracking system. The system uses 2 CCD cameras to follow the 3D coordinates of LED assemblies attached to the head, the mental region of the patient and a handpiece. We fixed LED assemblies to the patient’s head with special headgear and to the mental region with a mouthpiece fixed on the mandibular teeth. If the patient lacked adequate teeth in the mandible, the LED assembly was fixed directly to the jawbone. Movement
of the mandible as well as of the maxilla was tracked. Three LEDs were mounted on LED assemblies at the vertices of an isosceles triangle. To assess the accuracy of the system, an LED assembly was attached to the handpiece of a drill, and the drill tip used as a probe, to enable realtime navigation without switching between the probe and a tool. We used a laser scanner (VMS-250R, UNISN, Osaka, Japan) to transform the 3D morphological data of dental casts into digital data (Fig. 2). Frontal and lateral cephalograms were also digitized with a scanner (GT-9600, Seiko-Epson, Nagano, Japan). A scale was included in the cephalograms to allow their pixel size to be adjusted on the computer. 3D morphological data describing dental casts of the upper and lower jaws were integrated with the cephalogram data using computer software (OMF, UNISN,
Fig. 2. 3D morphological data describing the teeth of the patient were obtained from a dental cast by a laser scanner.
830
Tsuji et al.
Fig. 3. 3D data of the dental cast of the maxilla were integrated with cephalograms using computer software. Three points were selected on the teeth of the dental cast. These points formed a triangular plane on the display (blue triangle).
Osaka, Japan). Data integration was done with a projection-matching technique22 (Figs 3 and 4). To integrate the data, 3 points were selected on the dental cast of the upper jaw that were readily identifiable on cephalograms, e.g. median edge of the right maxillary central incisor, mesiobuccal cusp tip of the right maxillary 2nd
molar, and buccal cusp tip of the left maxillary 1st premolar. Each point was then input into a computer. These points formed a triangular plane on the screen (Fig. 3). If the intended morphological point was not identifiable, markers were cemented on the teeth and impressions taken of the teeth with the markers. Then, 3 other points in locations corresponding
Fig. 4. After selecting 3 points on the dental cast, the corresponding points on the cephalograms were marked in the software to form a triangular plane. 3D integration of the dental cast and the frontal and lateral cephalograms was performed by matching these 2 planes.
to those selected on the dental cast were selected on frontal and lateral cephalograms, forming a similar triangular plane. The images from the dental cast of the upper jaw and the frontal/lateral cephalograms were integrated by aligning these 2 triangular planes by means of a least squares technique (Fig. 4). We scanned upper and lower jaw models separately after fixing them in centric occlusion by using an occlusal record and a special tool. Centric occlusion of the jaw models could be reproduced by the computer software. The dental cast of the lower jaw was also automatically integrated with the cephalograms. Planar registration decreased errors during integration as compared to point-to-point registration. Navigational surgery requires registration of the cephalogram to the patient, for which dental casts were used (Fig. 1). The technique described above was used to integrate the dental casts and cephalograms. A separate ‘integration’ step was unnecessary because dental casts created by taking impressions closely match patients’ teeth. For surgical navigation, designated points were displayed on cephalograms by aligning the coordinates of the patient with those of the cephalogram. This was accomplished by registering the coordinates of the dental casts and the patient’s teeth in separate spaces (Fig. 1). Finally, a drill tip was used to indicate 3 points on the patient’s teeth in locations corresponding to those used for integration of the dental cast and cephalograms. A pair of CCD cameras detected an LED assembly on the handpiece, and the coordinates of the 3 points were calculated as 3D data based on the relative position between the LED assembly and the drill tip. The points were displayed as a triangle on a computer screen. Although the triangle derived from these points is congruent to that constructed by integrating the dental cast and cephalograms, these 2 triangles have different coordinate systems, skewed to each other. The procedure described above, however, can determine the positional relation between the skewed congruent triangles. Coordinates on the patient and cephalograms can be registered by aligning these triangles with the use of a coordinate transformation matrix created by Affine transformation (Fig. 5). First, the points located on the patient were transformed to a dental cast using a coordinate transformation matrix, and therefore projected onto frontal and lateral cephalograms. Navigation was thus performed in a ‘reference position’, where the patient was registered to the cephalograms
Navigation system using cephalograms
Fig. 5. Three points marked on the patient were displayed as a triangle on a computer screen. Although the triangle was congruent to that constructed by integrating a dental cast with the cephalograms, these 2 triangles had different coordinate systems, skewed to each other. The relative positional relation between the skewed congruent triangles was determined. Coordinates on the patient and cephalograms can be registered by aligning these triangles.
in real time (0.1 s delay). Points at the drill tip and calculated drill axis were displayed on images (Figs 6 and 7). The maxilla is usually used to determine 3 points on the patient’s teeth in a reference position. The mandible should be in centric occlusion to the maxilla. Cephalograms should, therefore, also be obtained in centric occlusion. Although navigation is performed in a reference position, the head can move, and the jaw position changes when the mouth
is opened and closed during surgery. Points, therefore, need to be tracked with CCD cameras via the LED assemblies attached to the head and mental region, transformed to coordinates in the reference position and displayed on cephalograms. When the head is stationary and the mandible is moved from the reference position with the drill tip positioned at a point on the mandible, the position of the
Fig. 6. LED assemblies were attached to the head, mental region and handpiece. Two CCD cameras tracked the LED, and real-time surgical navigation was performed.
831
drill tip is stored in a computer via the LED assembly attached to the drill tip and monitored by the CCD camera. The coordinates of the LED assemblies attached to the head and the mental region are simultaneously stored in the computer. The head remains stationary; therefore, its movement is 0. The CCD camera tracks the position of the LED assembly fixed to the mental region in real time. An LED is mounted on each of the 3 LED assemblies to form an isosceles triangle that is tracked by the camera. When a point on the mandibular teeth is marked with the drill tip, the relation between the triangle on the LED assembly of the mandible and the triangle in the reference position is stored in the computer. A coordinate transformation matrix, based on Affine transformation, superposes the 2 triangles of the mandible and transforms the points indicated after mandibular movement to coordinates in the reference position. The position and axis direction of the drill tip can, thus, be displayed on cephalograms created in centric occlusion (Fig. 8). When both the head and mandible move from the reference position, a coordinate transformation matrix is created in the computer from the post-movement coordinates of the LED assemblies attached to the head and mental region relative to the coordinates in the reference position. The indicated coordinate of the drill tip touching the mandible can be transformed onto an image in reference position by transforming the head LED to the reference position, followed by transformation of the mental region LED to the reference position. This allows real-time surgical navigation coordinated with movement of the head and mandible. Drills and burs of various lengths are used for surgery. When the tip position of these tools changes, the relative position of the tip to the LED assembly attached to the handpiece changes. The drill-tip position must, therefore, be calibrated against the LED assembly for each drill, using a calibrator (Fig. 9). The device used to calibrate tip position has 3 LEDs that allow the CCD camera to detect the coordinates of the drill tip. A drill tip, attached to a handpiece, is connected to the calibrator. The CCD camera detects the relative position of the drill tip between the 3 LEDs on the calibrator and the LED assembly attached to the handpiece. The computer calculates relative positions of the drill tip between these points and calibrates the position of the drill tip. The direction of the drill axis is also calculated.
832
Tsuji et al.
Fig. 7. The point of the drill tip and the drill axis were visualized on a pair of cephalograms.
Verification of accuracy
The accuracy of the LED 3D position was first evaluated by a stereo photogrammetry technique using 2 CCD cameras. A single LED was mounted on a 3-axis calibrator. Measurements were made at 5 locations 30 mm apart along each axis (125 points in total) within a cube measuring 120 mm on each side (Fig. 10). Next, the accuracy of transforming a dental cast to 3D data was verified using a preoperative dental cast of a patient who was scheduled to undergo orthognathic
surgery. Three landmarks were designated on the dental cast, which was then transformed to 3D data by computer-aided design (CAD) software (Surfacer 9.0, Imageware, USA). These points were located in the lower mesial angle on a buccal side bracket of the right maxillary 2nd molar, the lower distal angle on a labial side bracket of the right maxillary central incisor, and the lower mesial angle on a buccal side bracket of the left maxillary 2nd molar. A coordinate was measured for each point 10 times and the mean calculated. The distance measured from
the coordinate was compared with the actual distance on the dental cast. Finally, the accuracy of our surgical navigation system was verified using a phantom head as a model of a patient. A dental cast was attached to the phantom head, and 4 landmarks, differing from those on the dental cast, were designated as reference points on the face of the phantom: the right lateral canthus (point A), the right inner canthus (point B), the glabella (point C) and the nasal apex (point D). These reference points were selected so that the CCD camera could
Fig. 8. The patient’s head is fixed, and the mandible is moved from the reference position. The coordinates of the LED assemblies attached to the head and the mental region in reference position are simultaneously stored in the computer (upper left). The CCD camera tracks the position of the LED assembly fixed to the mental region in real time (upper right). When a point on the mandibular teeth is marked with the drill tip, the relation between a triangle on the LED assembly of the mandible (blue) and a triangle in the reference position (green) is stored in the computer. A coordinate transformation matrix, based on Affine transformation, superposes the 2 triangles of the mandible and transforms the points indicated after mandibular movement to coordinates in the reference position.
Navigation system using cephalograms
833
Mean errors of the CCD camera with respect to the LED position, measured with a 3D calibrator, were 0.034 mm for the X-axis, 0.031 mm for the Y-axis and 0.041 mm for the Z-axis. The maximum errors were 0.134 mm for the X-axis, 0.093 mm for the Y-axis and 0.143 mm for the Z-axis. Differences between distances among the 3 landmarks on the teeth as measured with the CAD software and the actual distances measured with calipers were used to estimate errors when the dental cast was transformed to 3D data with a laser scanner. Errors were 0.08 mm between points on the right maxillary 1st molar and central incisor, 0.21 mm between points on the right maxillary central incisor and the left maxillary 2nd molar; and 0.14 mm between points on the right maxillary 2nd molar and the left maxillary 2nd molar. To determine errors associated with the surgical navigation system, distances measured with digital calipers from points A to D on the head phantom to the dental cast attached to the head phantom were compared with the corresponding values calculated based on the coordinates stored in the system. Mean errors were 0.71 mm (0.21–1.09 mm) for the right maxillary central incisor, 0.62 mm (0.04–1.69 mm) for the right maxillary 2nd molar and 1.02 mm (0.23–1.47 mm) for the left maxillary 2nd molar (Table 1).
Fig. 9. A drill tip attached to a handpiece was connected to the calibrator.
Discussion Fig. 10. Two CCD cameras used in stereo photogrammetry technique to evaluate accuracy of LED 3D position.
easily detect the handpiece LED assembly when the drill tip pointed to these positions. Three points on the teeth were selected similarly to the procedure used to verify the accuracy of 3D transformation of the dental cast. A drill tip was pointed at the 7 coordinates described above 20 times each and the information stored in a computer. The mean of each coordinate was then calculated. The distances between points A–D and each of the landmarks on the dental cast attached to the phantom head were measured and compared to the actual distances measured with calipers. Errors were then estimated.
frontal and lateral cephalograms. The CCD camera tracked the position of the LED assembly attached to the handpiece to maintain the point of the drill tip and the drill axes in view. The point of the drill tip was re-calculated by means of a coordinate transformation matrix when the maxilla and mandible were moved from the reference position. It took 0.1 s to transform the point and display it on cephalograms. Navigation was performed on a real-time basis, concurrently with movement of the head and mandible.
Table 1. Accuracy of measuring distances from 4 landmarks on the head to 3 landmarks on teeth of phantom head
Results
Points indicated by the drill tip on the mandible and maxilla of the patient as well as the drill axes were displayed on
Dental casts represent the shape of teeth much more precisely than CT images and, therefore, provide much more precise markers for navigation systems. If readily identifiable landmarks are needed to register a patient to a radiograph via a dental cast, markers as small as 1 mm in diameter can be used as long as they are reproducible. A key benefit of this system is its relative simplicity compared to CT. Surgical navigation systems originally developed for stereotaxic cranial nerve surgery21,23 have been used for surgery of the skull base, sinuses, spine, and oral and maxillofacial region4,8,11,12,15,25,32. For maxillofacial surgery, the head is fixed in place by conventional methods; however, patients sometimes move their head
Right molar Right incisor Left molar
Point A
Point B
Point C
Point D
Mean difference (mm)
0.65 0.91 0.23
0.08 0.61 1.47
0.04 1.09 0.97
1.69 0.21 1.42
0.62 0.71 1.02
834
Tsuji et al.
during operation or open or close their mouth, moving the jaw, especially during oral surgery. Surgical navigation must, therefore, be modified to accommodate movement. This system enables real-time surgical navigation by tracking mandibular motion, as well as gross head movement, by means of an LED assembly. The LED assembly, used to track both maxillary and mandibular movement, needs to be fixed firmly to avoid dislodgement due to body movement. Dislodgment of the LED assembly decreases accuracy, even with highly precise surgical navigation systems. Headgear was used to fix an LED assembly on the head, but one could not be fixed to the mandible because the skin and muscles move too far when the mouth is opened and closed. The assembly was instead attached to a frame fixed to the teeth outside of the surgical site. This method cannot be used if patients are edentulous or have too few teeth. To obtain stable fixation in these patients, the mouthpiece for the LED assembly is fixed directly to the bone of the lower jaw, using screws designed for fixation of bone fractures. A surgical template can be used for dental implant placement to replicate the preoperative simulation during the operation5,18,26,28. This method is effective when enough teeth remain to mount the template on the bone surface. A template cannot provide good results when oral stabilization is difficult, such as in edentulous patients or in those undergoing tumour resection and reconstruction6. The precision of preoperative simulation images greatly affects operative results because re-registration and image correction are not possible during surgery. In addition, the technique depends in large part on the surgeon’s experience. A template-based method can provide a high degree of precision if the template can be firmly fixed to the surface of the jawbone. Another method to transfer preoperative simulation to the patient is mechanical transfer using special equipment during surgery17,24. Transfer accuracy is high with this method; however, 3D models used for simulation need to match the patient’s anatomy, and precise 3D images are required. Preoperative simulation and simulation transfer are carried out with devices that use 3D models. Error may be introduced during the production of these models. At present, a large space near the patient is needed to deploy the equipment in the surgical field. The application of mechanical transfer techniques is thus limited to facial fracture or orthognathic surgery.
Registration accuracy greatly affects the precision of navigation systems. When CT is used, markers need to be mounted in advance on the head and face for registration to the patient1,25,31. These markers should be large enough to compensate for the CT slice interval and width. Continuous 3D images must be constructed from intermittent image sets. Artifacts can influence the construction of 3D images, potentially increasing error. Markers attached to the skin may be displaced by body and muscle movements or swelling, which can also contribute to error10. 3D models produced on the basis of erroneous images may lead to further errors. With this navigation system, information on the cast is processed to 3D data for use in registration. These data allow preoperative simulation using CAD software, which is especially useful for osteotomies including teeth. Our system also enables the postoperative location of bone chips to be modelled with CAD software on the basis of spatial coordinates determined after displacement. Surgery can thus be evaluated in 3D. 3D images are also useful for obtaining informed consent before operation because they facilitate explanation of intended procedures to patients (Fig. 11). Precision of the CCD cameras, accuracy of transforming dental casts to 3D data, and accuracy of registration between imaging data and the patient are the main
factors affecting the performance of surgical navigation systems. The camera precision has an especially important role in the overall system. In this study, mean error was 0.035 mm and maximum error was 0.143 mm. Greater error appeared in the Z-axis than the X- or Y-axis, demonstrating that some axes or directions in space have reduced accuracy with 2-camera systems. Such inaccuracy may occur because the 2 cameras are configured to use a stereo photogrammetry technique to search space from different directions; therefore, accuracy near the centre of the field of view of both cameras is higher than that at the edges of the LED. When a navigation system is used, the surgical field needs to be located in the space providing the highest accuracy. Accuracy in recording a dental cast as 3D data influences data integration (done with computer software) as well as registration, both of which are important navigational components. Differences between dimensions obtained with the CAD software and those measured with calipers ranged from 0.08 to 0.21 mm. This result agrees with the findings of MOTOHASHI et al.19,20 (0.12 0.08 mm) who performed verification studies using an equivalent scanner. In the present study, 3D data of the dental cast were collected separately for teeth anterior and those posterior to the premolars, and then aligned in a computer. Discrepancies could have occurred during the alignment
Fig. 11. Upper panel: dentition before operation. Lower panel: dentition after operation. The dental cast is transformed to 3D data and processed using CAD software. It is possible to predict the operation outcome with this software, and to use the information to obtain informed consent from the patient29–31,2.
Navigation system using cephalograms process. In addition, distances between coordinates of corresponding points were measured to verify accuracy. Measured differences might have included errors that occurred during coordinate calculation. Another factor potentially affecting the accuracy of infrared navigation is lighting, especially the surgical light used in an operating room. If light is too bright or produces wavelengths similar to the LED, greater sampling errors occur when the CCD camera attempts to detect the LED. Preoperative registration of patients with a computerized dental cast is necessary in most operating rooms, and patients need to be fixed in a reference position during this procedure to avoid reducing the accuracy of navigation. To verify the accuracy of this navigation system, the system and digital calipers were used to measure distances from 4 points on the phantom head to 3 landmarks on the dental cast attached to the phantom head. The results were then compared. Points A–D were selected so that the CCD camera could readily detect LED locations indicated by the drill tip. When the left maxillary 2nd molar on the dental cast was indicated by the drill tip, the LED assembly attached to the handpiece showed a shorter angle toward the CCD camera than the other 2 points. In other words, when the CCD camera detected an isosceles triangle created by the LED assembly, the area of the triangle was very small as compared to other landmarks, which might have caused a detection error by the CCD camera. In addition, individual LEDs appeared to the CCD camera to be compressed to an oval shape, rather than round. Errors seem to have occurred when LED light was converted to pixels and the centroid was calculated. It appears therefore that LED-related detection errors resulted in errors in registering the drill-tip coordinates. The distance between the LED and CCD camera can lead to error. This effect was considered to be negligible in this study because the CCD cameras were positioned at optimal distances; however, some distance is necessary between the patient and camera in actual surgery because an operator stands between them. With this surgical navigation system, mean errors were 0.62 mm (SD 0.76 mm) at the right maxillary 2nd molar, 0.71 mm (SD 0.38 mm) at the right maxillary central incisor and 1.02 mm (SD 0.57 mm) at the left maxillary 2nd molar. The method used coordinates of pairs of points; the results may, therefore, have included measurement errors from both points. To reduce error, it appears that
LED assemblies attached to the patient and a handpiece must be at appropriate angles and distances with respect to the CCD camera. In conclusion, it has been shown here that this surgical navigation system provides sufficient accuracy for highly precise surgery. The system should be useful in oral and maxillofacial surgery, especially dental implant placement, endoscopic surgery and orthognathic surgery. References 1. Alp MS, Dujovny M, Misra M, Charbel FT, Ausman JI. Head registration techniques for image-guided surgery. Neurol Res 1998: 20: 31–37. 2. Birkfeller W, Huber K, Larson A, Hanson D, Diemling M, Homolka P, Bergmann H. A modular software system for computer-aided surgery and its first application in oral implantology. IEEE Trans Med Imaging 2000: 19: 616–620. 3. Birkfellner W, Watzinger F, Wanschitz F, Ewers R, Bergmann H. Calibration of tracking systems in a surgical environment. IEEE Trans Med Imaging 1998: 17: 737–742. 4. Claudius GN, Alexander S, Beat H, Sergio R, Daniel C, Wolf L, Rainer S. Computer-assisted secondary reconstruction of unilateral posttraumatic orbital deformity. Plast Reconstr Surg 2002: 110: 1417–1429. 5. Fortin T, Bosson JL, Coudert JL, Isidori M. Reliability of preoperative planning of an image-guided system for oral implant placement based on 3-dimensional images: an in vivo study. Int J Oral Maxillofac Implants 2003: 18: 886–893. 6. Fortin T, Champleboux G, Bianchi S, Buatois H, Coudert JL. Precision of transfer of preoperative planning for oral implants based on cone-beam CT-scan images through a robotic drilling machine. Clinical Oral Implants Res 2002: 13: 651–656. 7. Girod S, Keebe E, Girod B. Advance in interactive craniofacial surgery planning by 3D simulation and visualization. Int J Oral Maxillofac Surg 1995: 24: 120–125. 8. Grevers G, Leunig A, Klemens A, Hagedorn H. CAS of the paranasal sinuses—technology and clinical experience with the Vector-Vision-CompactSystem in 102 patients. Laryngorhinootologie 2002: 81: 476–483. 9. Grimson WEL, Ettinger GJ, Pe´rez TL, Wells WM, Kikinis R. An automatic registration method for frameless stereotaxy, image guided surgery, and enhanced reality visualization. IEEE Trans Med Imaging 1996: 15: 129–140. 10. Hassfeld S, Mu¨hling J. Computer assisted oral and maxillofacial surgery—a review and an assessment of technology. Int J Oral Maxillofac Surg 2001: 30: 2–13.
835
11. Ikeda K, Shoin K, Taguchi H, Yamano J, Yamashita J. Cranionavigator combining a high-speed drill and a navigation system for skull base surgery—technical note. Neurol Med Chir 1999: 39: 701– 708. 12. Itoh H, Ebara S, Kamimura M, Tateiwa Y, Kinoshita T, Yuzawa Y, Takaoka K, Ohtsuka K. Computerassisted pedicle screw insertion for surgery of spinal deformity. J Jpn Scoliosis Soc 1998: 13: 55–60. 13. Jacbs R, Adriansens A, Verstreken K, Suetens P. Predictability of a threedimensional planning system for oral implant surgery. Dentomaxillofac Radiol 1999: 28: 105–111. 14. Keeve E, Girod S, Kikinis R, Girod B. Deformable modeling of facial tissue for craniofacial surgery simulation. Comput Aided Surg 1999: 3: 228–238. 15. Khan M, Ecke U, Mann WJ. The application of an optical navigation system in endonasal sinus surgery. HNO 2003: 51: 209–215. 16. Kopp KC, Koslow AH, Abdo OS. Predictable implant placement with a diagnostic/surgical template and advanced radiographic imaging [in German]. J Prosthet Dent 2003: 89: 611–615. 17. Marmulla R, Niederdellmann H. Computer-assisted bone segment navigation. J Craniomaxillofac Surg 1998: 26: 347–359. 18. Mizrahi B, Thunthy KH, Finger I. Radiographic/surgical template incorporating metal telescopic tubes for accurate implant placement. Pract Periodont Aesthet Dent 1998: 10: 757–765. 19. Motohashi N, Kuroda A. A 3D computer-aided design system applied to diagnosis and treatment planning in orthodontics and orthognathic surgery. Eur J Orthod 1999: 21: 263–274. 20. Motohashi N, Nunota A, Kuroda T. Study on CAM system for dental model [in Japanese]. J Stomal Soc Jpn 1997: 64: 427–435. 21. Nakao N, Nakai K, Itakura T. Updating of neuronavigation based on images intraoperatively acquired with a mobile computerized tomographic scanner: technical note. Minim Invasive Neurosurg 2003: 46: 117–120. 22. Noguchi N, Goto M. Computer simulation system for orthognathic surgery. Orthod Craniofacial Res 2003: 6: 176– 178. 23. Ohue S, Kumon Y, Kohno K, Shiraishi M, Sakaki S, Kikuchi K, Miki H. Preoperative neurosurgical simulation and intraoperative navigation using threedimensional images. Jpn J Neurosurg 1999: 8: 657–663. 24. Santler G. 3-D COSMOS a new 3-D model based computerized operation simulation and navigation system. J Craniomaxillofac Surg 2000: 28: 287–293. 25. Schultes G, Ka¨rcher H, Gaggl A, Santler G. Computer assisted tumour
836
Tsuji et al.
resection of the skull base: case report. J Craniomaxillofac Surg 2001: 29: 326– 331. 26. Sicilia A, Noguerol B, Cobo J, Zabalegui I. Profile surgical template: a systematic approach to precise implant placement. A technical note. Int J Oral Maxillofac Implants 1998: 13: 109– 114. 27. Sießegger M, Schneider BT, Mischkowski RA, Lazar F, Krug B, Klesper B, Zo¨ller JE. Use of an imageguided navigation system in dental implant surgery in anatomically complex operation sites. J Craniomaxillofac Surg 2001: 29: 276–281. 28. Steenberghe D, Naert I, Andersson M, Brajnovic I, Cleynenbreugel J, Suetens P. A Custom template and definitive prosthesis allowing immediate implant loading in the maxilla: a clinical
report. Int J Oral Maxillofac Implants 2002: 17: 663–670. 29. Wagner A, Undt G, Watzinger F, Wanschitz F, Schicho K, Yerit K, Kermer C, Birkfellner W, Ewers R. Principles of computer-assisted arthroscopy of the temporomandibular joint with optoelectronic tracking technology. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2001: 92: 30–37. 30. Wagner A, Wanschits F, Birkfellner W, Zauza K, Klug C, Schicho K, Kainberger F, Czerny C, Bergmann H, Evers R. Computer-aided placement of endosseous oral implants in patients after ablative tumour surgery: assessment of accuracy. Clin Oral Implants Res 2003: 14: 340–348. 31. Watzinger F, Birkfellner W, Wanschitz F. Positioning of dental implants using computer-aided naviga-
tion and an optical tracking system: case report and presentation of a new method. J Craniomaxillofac Surg 1999: 27: 77–81. 32. Watzinger F, Wanschitz F, Wanger A, Enislidis G, Millesi W, Baumann A, Ewers R. Computer-aided navigation in secondary reconstruction of post-traumatic deformities of the zygoma. J Craniomaxillofac Surg 1997: 25: 198–202. Address: Mitsuhiro Tsuji Department of Oral and Maxillofacial Surgery Saga Medical School 5-1-1 Nabeshima Saga 849-0937 Japan Tel: +81 952342397 Fax: +81 952342044 E-mail: [email protected]