(2001) 30, 330 ± 335 ã 2001 Nature Publishing Group. All rights reserved 0250 ± 832X/01 $15.00 www.nature.com/dmfr Comparison of three-dimensional computed tomography with rapid prototype models in the management of coronoid hyperplasia J Asaumi*,1, N Kawai 1, Y Honda 1, H Shigehara 1, T Wakasa 1 and K Kishi 1 1 Department of Oral Radiology, Okayama University Dental School, Okayama, Japan Objectives: To compare (1) the accuracy of 3DCT wilt rapid prototyping (RP) models and (2) their application in the management of coronoid hyperplasia. Methods: An adult dry skull was used to compare 3DCT and a RP model for accuracy of reproduction; Polymerisation contraction and 17 distances between 27 anatomical points were measured. Two patients with coronoid process hyperplasia were assessed by means of 3DCT and plastic models. Results: Di erences in measurements with the RP model and the dried skull were smaller than those with 3DCT (P=0.04). Polymerisation contraction was greater than the errors in reproduction with the RP model and approximately equal to those with the 3DCT. The coronoid process was thick and prominent in the patients with coronoid process hyperplasia and the small space and early contact between zygomatic arch and the coronoid process demonstrated. The plastic model duplicated the exostosis of the inner surface of the malar bone facing the concavity of the elongated process. The coronoid process had an anterior tilt in one case. Conclusion: The plastic model duplicated the relationship between the maxillofacial deformity and the coronoid process with tolerable accuracy. Trismus may be due to the direction as well as the length of the enlarged coronoid process. Keywords: tomography, X-ray computed; models, structural; mandible; mandibular disease Introduction The advances in CT and computer graphics have made 3D reconstruction of anatomical structures possible. 1,2 We have applied 3DCT to the diagnosis and evaluation of various diseases of the maxillofacial region and for pre-operative surgical planning. 3±5 The development of stereolithography (referred to here as rapid prototyping (RP)) to make plastic models of the skull has made it possible to reproduce more complex structures. 6 Plastic models may make the 3D analysis of the facial bones easier than it is from CT reconstructions. Elongation of the coronoid process can be demonstrated by various radiographic methods. Panoramic radiography 7±9 and tomography 7,10 ± 13 clearly show the length and shape of the coronoid process. Waters' projection 8,9,14 plays an important role in revealing not only coronoid hyperplasia but also any malar exostosis. *Correspondence to: J Asaumi, Department of Oral Radiology, Okayama University Dental School, 2-5-1 Shikata-cho, Okayama 700-8525, Japan; E-mail: asaumi@md.okayama-u.ac.jp Received 31 October 2000; revised 24 May 2001; accepted 10 August 2001 However, it is not possible to con rm if there is any contact between the two. Such a relationship is more clearly visualized by sagittal tomography 12 or sagittal CT reconstruction. 7,15 In the present study, we compared the accuracy of a rapid prototyping (RP) model with 3DCT and then in the diagnosis of patients with elongation of the coronoid process. Materials and methods Laboratory study An in vitro study was undertaken to determine the accuracy of the RP model in reproducing a dry skull, in which there was no abnormality such as malocclusion and skeletal deformity, compared with 3DCT. CT scanning was performed with a TCT-900S scanner (Toshiba, Tokyo, Japan) at 120 kv and 100 ma. Slices were obtained at 1 mm thickness and 1 mm spacing.
Clinical study For the clinical cases, slices were obtained as 2 mm thickness and 2 mm spacing to reduce the absorbed dose. A threshold value of 200 HU using a bone algorithm was applied to generate the surface contour from a sequence of CT images. The data was stored on an optical disk and converted to digital format by the RP (laser lithography) method developed by D-Mec (Tokyo, Japan). 16,17 Measurement of polymerization shrinkage O-Methacryloyl-N-acyl tyrosines, 5 mol%, hydroxyethyl methacrylate, 95 mol%, and 0.2 wt/wt% to monomer total weight of benzoyl peroxide as the inhibitor were placed in a polymerization ampoule that was sealed with an oxygen torch under reduced pressure. Polymerization was performed in a constant temperature water bath, electrically heated and controlled at 60+0.028C for 24 h. The solidi ed polymer was removed from the ampoule and its density determined by directly comparing the weight of equal volumes of solid and water at 48C. Polymerization shrinkage was determined by comparing the density of polymerized mass with that of monomer mixture. 16 The plastic model (Desolite SCR 301., Japan Gum Co., Tokyo, Japan) was examined for volume and linear shrinkage by polymerization as follows: volume shrinkage=speci c gravity of the monomer ± speci c gravity of polymer/speci c gravity of the polymer; (linear shrinkage) 3 =volume shrinkage. 18 Comparison of the accuracy of the plastic model with that of 3DCT The accuracy of the plastic model compared with 3DCT was examined by marking 27 anatomical points, listed in the Key to Table 1, and measuring 14 distances as described by Kamijou. 19 Three additional distances are shown in Figure 1. The anatomical landmarks were marked on the surface of the skull 331 Figure 1 Measurement points on the RP models and the dried skull. (a) Length of coronoid process. (b) Length of condylar process. (c) Kondylion-gnathion (kdl-gn) Table 1 Comparison of the measurements on the dried skull with 3DCT and the RP model pr-ba na-ba na-pr upper facial width Dry skull 87.9 92.9 57.4 97.6 3DCT 86.7 (71.4%) 93.2 (0.3%) 60.2 (4.5%) 102.2 (4.7%) Plastic model 88.9 (1.1%) 93.0 (70.1%) 57.2 (70.3%) 97.8 (0.2%) Midfacial width Width of alveolar bone g-op ba-b width between the ears 87.1 60.0 177.2 134.0 98.0 85.2 (72.1%) 59.0 (71.7%) 177.9 (0.4%) 96.1 (71.9%) 85.8 (71.5%) 59.8 (70.3%) 175.1 (71.2%) 134.3 (0.2) 97.7 (70.3%) Maxillary alveolar Length of the palate Width of the palate Defect in the anterior wall of the maxillary sinus 48.9 44.0 38.9 13.3614.7 40.5 (717.2%) 45.3 (3.0%) 38.0 (72.3%) 13.5613.9 48.8 (70.2%) 43.0 (72.3%) 38.6 (70.8%) 12.7613.9 kdl-gn Length of coronoid process Length of condylar Distance between the Average error process condylar processes 113.1 54.9 60.8 110.0 113.2 (0.1%) 53.4 (72.7%) 59.8 (71.6%) 106.0 (73.6%) 2.16% 113.2 (0.1%) 54.6 (70.5%) 60.3 (70.8%) 109.8 (70.2%) 0.63% Values in parentheses are di erences between measurements from the 3DCT and RP model and the skull respectively. Measurement unit: mm. Key: pr, prosthion; ba, basion; na, nasion; g, glabella; op, opisthokranion; b, bregma; kdl, lateral kondylion; gn, gnathion; upper facial width (between the frontozygomatic sutures); midfacial width (between the zygomaticomaxillary sutures); maxillary aveolar width, between the middle of the central incisors and the edge of maxillary alveolar processes
332 with gutta-percha prior to 3DCT. Measurements were made from 3DCT scans as described by Ono et al. 20 Brie y, 3D images from six directions (horizontal from below, horizontal from above, right sagittal, left sagittal, front and back) were displayed simultaneously and the measurement points located on the monitor with the cursor on the monitor. The measurement points on the plastic model were marked directly on the model. The measurements were made using calipers 19 by two investigators (JA and NK) and repeated after 1 week. The measurements on the skull were compared with those from 3DCT and the plastic model and the mean error for each measurement point with each technique calculated. Clinical study In the two cases of coronoid process hyperplasia, the length of the condylar process and the length and thickness of the coronoid process were measured using calipers 19 twice by two investigators (JA and NK). The measurements were repeated after 1 week (Figure 2). The angle between the posterior margin of ramus and the coronoid process (d in Figure 2) in the sagittal plane was measured. Intra- and interobserver error and the kappa values were calculated. The kappa values were calculated with a minimal unit of mm. Statistical analysis was performed using Statview J 5.0 software (SAS Institute Inc. San Francisco, CA, USA). Comparison between values was carried out using Student's t-test. A signi cant di erence was de ned as P50.05. Results Contraction of the plastic material on polymerisation The speci c gravity of the monomer was 1.08 and the polymer 1.15. The contraction rate was therefore 1.08 ± 1.15/1.15 and thus the volume shrinkage 76.09% and the linear shrinkage 72.03%. Comparison of RP model with 3DCT Both 3DCT and the RP model successfully reproduced the complex anatomical structures of the dry skull. The RP model cost US$ 120071400 and took a month to manufacture. The average error with 3DCT compared with the dried skull was 2.16% and ranged from 717.2% for alveolar width in the maxilla to 4.7% for upper facial width. The average error with the RP model was 0.63%, ranging from 71.5% for middle facial width to 1.1% for the distance between prosthion and basion. The values with the RP model were consistently smaller than with 3DCT for all the lengths measured (Table 1) (P=0.04). In the laboratory study, the average intraobserver errors for observer A were 0.47 (0.09 ± 2.4), 0.55 (0.06 ± 2.0), and 0.48 (0 ± 0.8)%, and the kappa values 0.47, 0.53, and 0.44 respectively for the dry skull, RP model, and 3DCT and for observer B, 0.41 (0 ± 2.0), 0.95(0 ± 1.5), and 0.83 (0 ± 3.0)% and the kappa values 0.71, 0.29, and 0.3. The average interobserver errors were 0.76 (0 ± 2.1), 1.2 (0.2 ± 5.9) %, and 1.25 (0.2 ± 2.8), and the kappa values were 0.47, 0.35 and 0.13. Clinical cases Case 1: A 14-year-old boy had maximal mouth opening of 2 mm due to bilteral coronoid hyperplasia con rmed by both 3DCT and the RP model. When viewed from below, CT showed a depression on the right side of the maxilla with a small maxillary sinus, a marked curvature of the right zygomatic arch and a deformed skull. The right side of the skull was smaller than the left side and the space between the skull and the zygomatic arch on the right was smaller than that on the left. This might cause early contact between the zygomatic arch and the coronoid process. The concavity in the left coronoid process and the hyperplasia on the inner surface of the malar bone facing it were easily recognized on the plastic model (Figure 3). The similar relationship between the right coronoid process and the malar bone in occlusion was also clearly seen. Case 2: A 7-year-old boy had maximum mouth opening of 17 mm due to bilateral coronoid process hyperplasia which was con rmed by both 3DCT and the RP model. Prominent hyperplasia of the malar bones on both sides facing the elongated coronoid processes was more clearly recognized on the RP model Figure 2 Measurement points on the RP models of two clinical cases with coronoid hyperplasia. (a) Length of condylar process. (b) Length of coronoid process. (c) Thickness of coronoid process. (d) Angle between the posterior margins of the ascending ramus and coronoid process
a b 333 Figure 3 Case 1. An RP model showing (a) the concavity in the left coronoid process (arrow) and (b) the exostosis-like hyperplasia of the left malar bone (arrow) than on 3DCT (Figure 4). Severe deformity of the skull, the maxillary bone, and the zygomatic arch were also visible. The right side of the skull was remarkably smaller than the left and the space between the skull and the zygomatic arch on the left was smaller than on the right. This could cause early contact between the zygomatic arch and the coronoid process. Measurements using plastic models in clinical cases Table 2 shows the results of the measurements on plastic models. The elongation and thickening of the coronoid process in these cases is clearly seen. In Case 2, the angle between the posterior margins of ramus and the coronoid process was much wider. The results show that the coronoid process of Case 2 tilted anteriorly as it expanded. This caused the early contact of both coronoid processes, especially on the left. The mean intra-observer error of observer A, was 2.6 (0 ± 7.0)% and the kappa value was 0.2 and for observer B, intra-observer error 1.9 (0 ± 5.0)% and the kappa value 0.3. Interobserver error was 1.4 (0 ± 5.9) and the kappa value was 0.5. Discussion Three-dimensional images do not su er from superimposition and therefore allow views that are Table 2 The results of the measurements on the plastic models in two cases of bilateral coronoid hyperplasia compared with a dry skull a (mm) b (mm) c (mm) d (degree) Rt. Lt. Rt. Lt. Rt. Lt. Rt. Lt. Case 1 10 13 26 32 11 9 23 17 Case 2 11 9 20 21 8 10 35 42 Dry skull 17 19 12 12 5 5 22 26 Key: for a, b, c and d see Figure 2 impossible with conventional radiographs. These have proved very bene cial in developing 3DCT reconstruction for diagnosis and treatment planning. 13 ± 15,21 We have shown the accuracy of reproduction of the plastic model was notably superior to that of 3DCT. The intra- and interobserver variation on the dry skull was smaller than that associated with 3DCT and approximately similar to that found with the RP model. The contraction associated with polymerisation was larger than the measurement variation with the plastic model and approximately equal to that with 3DCT. The contraction strongly a ected the accuracy of reproduction of the dried skull. However, the combined contraction and measurement errors may not in uence surgery since they are within approximately 2%. The kappa values were low for intraobserver examination of the RP models, the 3DCT measurements by observer B and for the interobserver examination of the 3DCT, while they were relatively
334 Figure 4 (arrows) Case 2. RP model viewed from above showing prominent hyperplasia of both the malar bones adjacent to the coronoid processes high for the dried skull. It is not surprising that the measurements from a dried skull are easier than those from a RP model with a rough surface and from 3DCT on the monitor. Abnormalities of the coronoid process are extremely rare compared with those of the condylar head. Hyperplasia of the coronoid process can occur either unilaterally or bilaterally. Coronoid process hyperplasia which results in restricted mandibular opening has been reported frequently in the dental literature: 7 ± 9,14,22 it is believed that the occurrence of the trismus is due to its impingement on adjacent bones. 14,22 Nakao et al. reported that it was due to contact between the coronoid processes and the malar bones. 22 In this study, we compared 3DCT with RP models in two patients in order to observe the relationship between the coronoid process and the malar bone from various directions. In both cases, the direct, functional cause of the patient's trismus was more readily seen on the plastic models. Severe deformity of the maxilla, both sides of the zygomatic arches, the malar bone, and the skull were visualized in both cases. The small space and early contact between the zygomatic arch and the coronoid process, led to trismus. The concavity in the coronoid process and the hyperplasia on the inner surface of the malar bone was easily overlooked on 3DCT. Further, we could con rm in more detail the early contact between the zygomatic arch and the coronoid process on opening of the mouth with the RP models. If the cost and the radiation dose is reduced, we would be able to use the RP model for simulation of surgery more easily. We found in Case 2 that the enlarged coronoid process had tilted anteriorly. Thus, the direction as well as the length of the coronoid process may have led to the early contact with the zygomatic arch, leading to trismus. However, more cases are needed to con rm this notion. In conclusion, the RP model duplicated the relationship between the maxillofacial deformity and the coronoid process with acceptable accuracy. Analysis of the RP model clari ed the cause of the trismus in coronoid hyperplasia. References 1. Hemmly DC, Zonneveld FW, Lobregt S, Fukuta K. A decade of clinical three-dimensional imaging: A review. Part 1: Historical development. Invest Radiol 1994; 29: 489 ± 496. 2. Zonneveld FW, Fukuta K. A decade of clinical three-dimensional imaging: A review. Part 2: Clinical applications. Invest Radiol 1994; 29: 574 ± 589.
3. KishiK,AkitaK,NakatsuT,AdachiK,UemuraM,OchiS, Sato EA. Evaluation of in ammatory diseases of jaw bones with three-dimensional CT imaging: Comparison with conventional radiography and high resolution CT. Oral Radiol 1992; 8: 21 ± 31. 4. KishiK,FujikiT,HirakawaK,HasegawaI,UemuraM,Adachi K,ShigeharaH,YamamotoM,HirakiY.ThreedimensionalCT imaging in the evaluation of fracture of the jaws. Oral Radiol 1993; 9: 17 ± 25. 5. Komatsu H, Wakasa T, Kameyama T, Hirakawa K, Nakatsu T, Kishi K. A study of three-dimensional CT reconstruction conditions of the malignant tumors in maxillo-facial region. J Okayama Dent Soc 1990; 9: 331 ± 338. 6. Zonneveld F. A decade of clinical three dimensional imaging: A review. Part 3: Images analysis and interaction, display options, and physical models. Invest Radiol 1994; 29: 716 ± 725. 7. Tucker MR, Guilford WB, Howard CW. Coronoid process hyperplasia causing restricted opening and facial asymmetry. Oral Surg Oral Med Oral Pathol 1984; 58: 130 ± 132. 8. Isberg A, Isacsson G, Nah K-S. Mandibular coronoid process locking: A prospective study of frequency and association with internal derangement of the temporomandibular joint. Oral Surg Oral Med Oral Pathol 1987; 63: 275 ± 279. 9. Kreutz RW, Sanders B. Bilateral coronoid hyperplasia resulting in severe limitation of mandibular movement. Report of a case. Oral Surg Oral Med Oral Pathol 1985; 60: 482 ± 484. 10. Allison ML, Wallace WR, von Wyl H. Coronoid abnormalities causing limitation of mandibular movement. JOralSurg1969; 27: 229 ± 233. 11. Fukumori T, Tagawa T, Inui M. Bilateral coronoid process hyperplasia and short stature: A case report. Int J Oral Maxillofac Surg 1993; 22: 139 ± 141. 12. Honig JF, Merten HA, Korth OE, Halling F. Case report: Coronoid process enlargement. Dentomaxillofac Radiol 1994; 23: 108 ± 110. 13. Hecker R, Corwin JO. Bilateral coronoid hyperplasia: review of the literature and report of case. JOralSurg1980; 38: 606 ± 608. 14. Takahashi A, Hao-Zong W, Murakami S, Kondoh H, Fujishita M, Fuchihata H. Diagnosis of coronoid process hyperplasia by three-dimensional computed tomographic imaging. Dentomaxillofac Radiol 1993; 22: 149 ± 154. 15. Hall RE, Orbach S, Landesberg R. Bilateral hyperplasia of the mandibular coronoid processes: A report of two cases. Oral Surg Oral Med Oral Pathol 1989; 67: 141 ± 145. 16. Tanaka J, Suzuki K, Nakai H. Syntheses of O-methacryloyl-Nacyl tyrosines and their adhesiveness to unetched human dentin. Dental Materials Journal 1994; 13: 55 ± 65. 17.KozakiH,NishiiY,SakamotoT,SueishiK,IsshikiY.The Shikwa Gakuho 1998; 98: 547 ± 553 (in Japanese). 18. Komori T, Takato T, Miyamoto M, Nakatsuru M, Mori Y, Sakamoto Y, Akagawa T. Application of a laser-hardened type of three-dimensional model in jaw deformities. Jpn J Maxillofac Surg 1993; 39: 871 ± 878 (in Japanese). 19. Kamijou Y. Oral anatomy: 1. Osteology (Craniology) 2nd ed. Tokyo, Anatohmu 1982 (in Japanese). 20. Ono I, Ohura T, Narumi E, Kawashima K, Matsuo I, Nakamura S, et al. Three-dimensional analysis of craniofacial bones using three-dimensional computer tomography. J Cranio-maxillo-fac Surg 1992; 20: 49 ± 60. 21. Gillespie JE, Quayle AA, Barker G, Isherwood I. Three dimensional CT reformations in the assessment of congenital and traumatic cranio-facial deformities. Brit J Oral Maxillofac Surg 1987; 25: 171 ± 177. 22. Nakao K, Osaki T, Ohno A and Kato I. A case of jaw trismus caused by the hyperplastic coronoid process interfered with the zygomatic arch, and studies on the positional interrelation between the process and the arch. Jpn J Oral Maxillofac Surg 1985; 31: 331 ± 336 (in Japanese). 335