Вы находитесь на странице: 1из 10

ORIGINAL ARTICLE

Method to classify dental arch forms


Shin-Jae Lee,a Sungim Lee,b Johan Lim,c Heon-Jin Park,d and Timothy T. Wheelere Seoul, Gyeonggi-Do, and Incheon, Korea, and Gainesville, Fla

Introduction: The aim of this study was to propose a method to classify dental arch forms of subjects with normal occlusion into several types that can ensure both goodness of t and clinical application. Methods: We selected 306 subjects with normal occlusion from 15,836 young adults, recorded 14 reference points that dened the distance between 2 arch forms as the area between 2 arches, and then classied the dental arch forms by using the partitioning around medoids clustering and silhouette method. We measured tooth size, arch width, basal arch width, arch depth, mesiodistal angulations, and buccolingual inclinations. Results: We identied 3 types of arch forms, and cross-classication of the maxillary by mandibular arch forms showed a more frequent distribution in the diagonal elements than in the off-diagonal elements. The 3 arch forms showed differences in tooth size, arch width, basal arch width, and inclination of the posterior teeth. Conclusions: By dening area discrepancies as distance measures and applying them to the cluster method by using medoids, the dental arch form can be classied keeping control for the extremes without bias. It is hoped that this method will have possible clinical applications in determining the shape and number of preformed orthodontic arch forms. (Am J Orthod Dentofacial Orthop 2011;140:87-96)

he development of the preadjusted bracket allowed orthodontists to concentrate more on diagnosis, treatment planning, and treatment mechanics rather than on complicated archwire bending. However, with preadjusted brackets, selection of a preformed archwire has become a more important step in clinical orthodontic practice that is often overlooked. As evidenced by the many types of arch forms available, orthodontists do not agree on a single arch-form shape. Previous studies have mainly focused on describing and tting arch forms geometrically or mathematically in a more accurate manner. Many authors claimed that there might be a standard form of dental arch and some geometric or mathematical curve; eg, semicircle1; ellipse2,3; parabola, hyperbola4; catenary curve5,6; cubic spline
a Associate professor, Department of Orthodontics, School of Dentistry and Dental Research Institute, Seoul National University, Seoul, Korea. b Associate professor, Department of Information and Statistics, Dankook University, Gyeonggi-Do, Korea. c Associate professor, Department of Statistics, Seoul National University, Seoul, Korea. d Professor, Department of Statistics, Inha University, Incheon, Korea. e Eminent scholar and chair, Department of Orthodontics, College of Dentistry, University of Florida, Gainesville. The authors report no commercial, proprietary, or nancial interest in the products or companies described in this article. Supported by grant no. 03-2010-0023 from the SNUDH Research Fund. Reprint requests to: Timothy T. Wheeler, Department of Orthodontics, College of Dentistry, University of Florida, Box 100444 JHMHSC, Gainesville, FL 326100444; e-mail, twheeler@dental.u.edu. Submitted, August 2009; revised and accepted, May 2010. 0889-5406/$36.00 Copyright 2011 by the American Association of Orthodontists. doi:10.1016/j.ajodo.2011.03.016

function7,8; conic sections9,10; polynomial functions including second-order polynomial,11 fourth-order polynomial,12-14 sixth-order polynomial15,16; Euclidean distance matrices17; Fourier series18,19; beta function15,20; and Bezier cubic equation21 were once described as standard ideal forms that should t the dental arch and ensure accuracy in describing arch forms. Today, however, many normal variations are emphasized more than the specic type of arch forms.9 Recent studies nullify the existence of a single ideal arch-form template, indicating that dental arch forms are highly individual, and dening a single generalized shape and using variations of it should be avoided.16,22 Once the arch form has been described with a certain geometric or mathematical function, the problem of how to dene or classify the dental arch into several types systematically without observer bias still remains. Arch-form classication is especially important when using shape-memory wires. Although there are commercial forms, the forms are classied not by scientic basis but merely by beliefs of some orthodontists concerning arch forms. Some previous investigators used commercial arch forms for their arch-form related research: eg, OrthoForm (3M Unitek, Monrovia, Calif),23 Pentamorphic arch template (Rocky Mountain Orthodontics, Denver, Colo),21 Tru-Arch (Ormco, Sybron Dental Specialties, Orange, Calif),24 and Brader (American Orthodontics, Sheboygan, Wis)24 among others. However, there is a certain degree of subjectivity in the classication of dental arch forms. The form is not based on any data but, rather, is predicated on the clinical experiences
87

88

Lee et al

or preferred forms of certain clinicians or companies. Furthermore, the studies of Felton et al13 and Camporesi et al24 showed neither a predominance of any particular arch form among commercially available arch forms, nor a superior closeness of t. We postulated that the classications and clinical applications are inversely related to each other. Thus, if a classication is not accurate, there will be no accuracy in t. Inversely, if classication becomes too prescriptive, it interferes with the practical useclinicians' selection, inventory management, and so forth. Therefore, cluster analysis helps to determine the number of classications so that the best t can be achieved with maximum between-group distances. Originally, cluster analysis was used to identify certain groups of observations that are cohesive and separated from other groups. Multivariate cluster analysis is a relatively new method in biomedical science and can interpret an entire set of data while preserving information about individuals. In dentistry, a possible application of this methodology could be to classify dental arch forms without practitioner bias. The aim of this study was to develop a method to classify dental arch forms that are difcult to identify from subjective visual inspection into several forms that can ensure both goodness of t and pragmatic clinical application.
MATERIAL AND METHODS

A total of 306 subjects with normal occlusion were selected from 15,836 young Korean adults who responded to a community dental health survey from 1997 to 2005 in Seoul, Korea. They included 187 men and 119 women with a mean age of 20.0 years (range, 17-24 years). The selection criteria were (1) Class I molar and canine relationships with normal occlusal interdigitation, (2) complete permanent dentition erupted except the third molars, (3) normal overjet and overbite (about 2-4 mm), (4) minimal crowding (\2 mm) and spacing (\1 mm), and (5) no previous orthodontic or prosthodontic treatment. In addition, subjects with proximal caries or llings that affected a tooths size and shape, gross restorations, signicant attrition, congenital defects, or deformed teeth were excluded. Absence of tooth anomalies of structure and development was also considered. The subjects were part of the Korean Standard Occlusion Study, which has been ongoing since 1997.25 Dental stone casts were made from alginate impressions, and the sizes of the teeth were measured by using digital Vernier calipers (Mitutoyo; Kawasaki, Kanakawa, Japan) with sharpened points under an illuminated

magnier (Otsuka Optics, Tokyo, Japan) from the central incisors to the second molars (accurate to 0.01 mm). The size of each tooth was dened as the mesiodistal width measured from its mesial contact point to its distal contact point of its greatest distance, as suggested by Jensen et al.26 Arch width, basal arch width, and arch depth were measured from the canines up to the second molars. A centroid was constructed for each cusp tip, which was relatively independent of cusp wear or abrasion. The definition of basal arch width referred to the distance between the apical third of the alveolus as suggested by Howes.27 In addition, several proportional variables were measured, including arch width-arch depth, intercanineintermolar width, anterior curvature (sum of the anterior tooth sizes and intercanine width), and arch width ratio (maxillary arch width to mandibular arch width as a percentage). An angulation-and-inclination measuring gauge (Invisitech, Seoul, Korea) was used to record angulation and inclination of a tooth on a dental cast. A level and a dental surveyor kit with a laboratory jack were used to ensure the atness of the measuring eld, and the dental cast was positioned parallel to the functional occlusal plane as described by Lee et al,28 Kim et al,29 and Lee et al.30 The 3 pointers on the measuring gauge were directed toward the facial axis of the clinical crown with the middle pointer aimed toward the facial axis point as described by Andrews.31 Dental casts were photocopied, the 14 reference points from the central incisors to the second molars were digitized (Intuos 2 Graphic Tablet; Wacom, Toyonodai, Kazo-shi, Saitama, Japan) and converted into Cartesian coordinates with custom-made software in Delphi (CodeGear, Scotts Valley, Calif) programming language (Fig 1). The set of data points in Figure 1 was chosen to represent the dental arch forms of the maxilla and the mandible. The problem was that the reference points are neither on a smooth curve nor symmetric. To overcome the problem, we interpolated the data points of each subject using the piece-wise linear function for further cluster analysis (Fig 2). The observed data points were not registered well because superimpositions of the observed arches were inconsistent in the origin and therefore randomly rotated left or right. To resolve this difculty, we dened a new distance between 2 arch forms that is invariant to location shift and rotation transformation. Suppose x ; f x and x ; gx are 2 continuous functions representing 2 arch forms. The new distance dfx ; f x ; x ; gx g, simply denoted by df ; g, is dened as:

July 2011  Vol 140  Issue 1

American Journal of Orthodontics and Dentofacial Orthopedics

Lee et al

89

Fig 1. Observed arch forms in the maxilla (left) and the mandible (right).

Fig 2. Interpolated arch forms by using piece-wise linear function: left, the maxilla; right, the mandible.

Z df ; g5 inf
T d;qTU

jf x T d; qgx jdx ;

where T d; q is the transformation that shifts the original function by d and rotates it by q, and TU is the collections of all T d; qs. This new distance is illustrated in Figure 3. The left panel of Figure 3 plots 2 interpolated arch forms from original data points. To evaluate the distance, we xed 1 curve and then moved the other curve with location shift and rotation transformation until the area between 2 transformed curves was minimized. The right panel of Figure 3 shows the nal tted curves and their distances. With this distance, dental arch forms were clustered by using the partitioning around medoids (PAM)

method.32,33 A medoid, similar to a median, is more robust to uncontrolled noise or outliers than the mean. Thus, PAM is also more robust than the conventional K-means clustering method.33 In addition, the PAM with silhouettes provides information about the appropriate number of clusters to use for the analysis.32 The number of arch-form clusters was selected via the average silhouette width and practical viewpoints.
Statistical analysis

After cluster analysis for dental arch forms, all subjects were assigned to a cross-classication table composed of maxillary and mandibular clusters. The McNemar-Bowker test for the contingency table was used to determine whether the subjects were equally

American Journal of Orthodontics and Dentofacial Orthopedics

July 2011  Vol 140  Issue 1

90

Lee et al

Fig 3. New denition of distance measure. The left panel plots 2 interpolated arch forms from original data points. To evaluate the distance, 1 curve was xed, and then the other curve was location shifted and rotation transformed until the area between the 2 transformed curves was minimized. The right panel shows the nal transformed curves and their distances.

likely to fall into each classication category. After checking the assumption of homoscedasticity and normality, 1-way analysis of variance was performed to nd several important dental variables that are supposed to inuence the characteristic arch forms. All reported P values were based on 2-sided levels of signicance.
RESULTS

Table I. Clustering results for 306 dental arch forms by using the PAM
Mandible Narrow Maxilla Narrow Middle Wide Total 60 9 0 69 Middle 33 90 14 137 Wide 3 22 75 100 Total 96 121 89 306

For the linear measurements, the intraexaminer and interexaminer reliability coefcients were 0.9979 and 0.9963, respectively. For the angular measurements, the interexaminer errors of the estimation were as follows: tooth angulation, from 0.50 to 1.97 with a mean of 1.36 ; and tooth inclination, from 0.16 to 1.79 with a mean of 0.90 . In terms of root mean squares, the random errors of intraexaminer and interexaminer evaluations were lower than 0.083 and 0.111 mm, respectively. After cluster analysis by using the PAM method, 3 types of arch forms were identied in both the maxilla and the mandible. The PAM algorithm provided the silhouettes with which to display each cluster graphically. Most subjects were classied as having the intermediate type of arch form. Table I shows that there could be various combinations between the maxillary and mandibular arch forms to achieve normal occlusion. Most subjects (224 of 306), however, had a congruent intermaxillary arch form. The McNemar-Bowker test for the contingency table showed that the subjects with an

incongruent intermaxillary arch form were not evenly split (P 5 0.0003). It was apparent that the upper right off-diagonal cells (58 of 306) were more prevalent than the lower left off-diagonals (23 of 306), indicating a specic pattern of combinations between the maxillary and the mandibular arch types. The intermaxillary arch width ratio among the subjects who were assigned to the contingency table (Table I) showed no signicant difference except for the rst molar interarch width ratio (Table II). In orthodontic practice, only the subjects in diagonal elements could be regarded as having a normal occlusion. All subjects had normal occlusion. We found it interesting that, as shown in Table I, there were various combinations between the maxillary and mandibular arch forms to achieve normal occlusion. Nonetheless, in clinical orthodontics, to treat a patients malocclusion, applying the same arch form on both the mandibular

July 2011  Vol 140  Issue 1

American Journal of Orthodontics and Dentofacial Orthopedics

Lee et al

91

Table II. Intermaxillary arch width ratio difference between subjects according to assignment to the contingency table in Table I
Arch width ratio Intercanine Interrst premolar Intersecond premolar Interrst molar Intersecond molar NS, Statistically not signicant. *P \0.001. Upper off-diagonal (n 5 58) 76.70 6 4.26 81.21 6 3.59 84.19 6 3.74 86.16 6 3.26 87.11 6 2.75 Diagonal (n 5 224) 76.03 6 3.37 80.38 6 3.01 82.95 6 5.14 84.85 6 2.42 86.35 6 2.73 Lower off-diagonal (n 5 23) 77.01 6 2.35 80.40 6 2.75 81.78 6 3.27 83.31 6 2.01 86.28 6 1.91 Signicance NS NS NS * NS

Fig 4. Estimated standard arch forms by using symmetric cubic spline method: left, the maxilla; right, the mandible.

and maxillary dentition seemed more reasonable than accepting the variability of the arch form for each dentition. For this reason, we estimated the standard arch forms using subjects in the diagonals of Table I. The standard arch forms determined by tting the symmetric cubic spline are plotted in Figure 4.34 In general, arch-form type was inuenced by tooth sizes, arch width, basal arch width (Table III), and inclination of the posterior teeth (Table IV). Mesiodistal angulation of the teeth did not make a signicant difference (Table IV). Higher values of the arch width to arch depth ratio were supposed to indicate relatively broader arches rather than narrower arches; higher values of the intercanine to intermolar width ratio indicated square in contrast to tapered arches; and higher anterior curvature values indicated convex vs at anterior curvature arches. Proportional variables showed that the narrow arch type in both the maxilla and the mandible has tapered and convex arch forms with a lower value of arch width to arch depth ratio, a higher value of intercanine to

intermolar width ratio, and a higher anterior curvature compared with the wide type of arch form. Regarding the intercanine to intermolar width ratio, a narrow arch had the smallest intercanine and intermolar widths. The wide arch form showed not only the largest intercanine width but also had much greater intermolar width than did the narrow arch form; this resulted in the higher value of intercanine to intermolar width ratio for the tapered and convex arch form (Tables III and IV).
DISCUSSION

There are 2 broad types of grouping procedures: supervised and unsupervised classications. The supervised classication was the conventional and traditional tool used in dentistry. For example, Noroozi et al,15 using 23 sets of dental casts, classied arch forms into square, ovoid, and tapered. In this study, the rst step was tting the arch form via mathematic modeling with the sixth-order polynomial function. Then, they selected linear variables to describe the arch form, the (Wc/

American Journal of Orthodontics and Dentofacial Orthopedics

July 2011  Vol 140  Issue 1

92

Lee et al

Table III. Linear measurements (mm) according to arch-form clusters


Maxilla Narrow (n 5 96) Tooth size U1 8.28 6 0.49 U2 6.77 6 0.46 U3 7.79 6 0.44 U4 7.17 6 0.37 U5 6.68 6 0.36 U6 10.29 6 0.57 U7 9.45 6 0.56 Sum of anterior tooth sizes U123 22.84 6 1.19 Total sum of tooth sizes U1234567 56.43 6 2.38 Arch width U3-U3 35.41 6 1.60 U4-U4 43.11 6 1.69 U5-U5 48.61 6 1.85 U6-U6 53.38 6 1.98 U7-U7 59.16 6 2.28 Basal arch width U3-U3 34.54 6 2.81 U4-U4 44.55 6 2.55 U5-U5 53.47 6 2.78 U6-U6 60.13 6 2.47 U7-U7 64.74 6 2.42 Arch depth 37.43 6 2.01 Middle (n 5 121) Wide (n 5 89) Signicance y * y z y y z y z z z z z z z z z z z NS L1 L2 L3 L4 L5 L6 L7 L123 Narrow (n 5 69) Mandible Middle (n 5 137) Wide (n 5 100) Signicance NS NS y NS z * z * z z z z z z z z z z z *

8.41 6 0.44 8.50 6 0.50 6.93 6 0.47 6.95 6 0.56 7.89 6 0.41 7.99 6 0.46 7.33 6 0.38 7.41 6 0.38 6.79 6 0.41 6.89 6 0.38 10.48 6 0.54 10.55 6 0.53 9.62 6 0.73 9.84 6 0.63 23.23 6 1.12 23.44 6 1.22 57.44 6 2.46 58.14 6 2.55 36.65 6 1.75 45.05 6 1.56 50.89 6 1.58 55.93 6 1.57 61.72 6 2.16 35.89 6 2.93 46.73 6 2.49 55.82 6 2.68 62.44 6 2.57 67.24 6 2.44 37.38 6 1.93 46.67 6 2.08 53.19 6 1.94 58.90 6 1.98 64.92 6 2.18 36.71 6 2.88 48.06 6 2.67 57.30 6 2.61 64.71 6 2.56 69.85 6 2.51

5.23 6 0.33 5.23 6 0.45 5.31 6 0.34 5.85 6 0.35 5.84 6 0.38 5.91 6 0.34 6.74 6 0.39 6.81 6 0.38 6.94 6 0.39 7.16 6 0.40 7.20 6 0.41 7.25 6 0.37 6.90 6 0.41 6.97 6 0.43 7.13 6 0.40 10.82 6 0.51 10.91 6 0.59 11.05 6 0.54 10.23 6 0.70 10.38 6 0.66 10.67 6 0.65 17.81 6 0.89 17.88 6 1.00 18.16 6 0.93

L1234567 52.93 6 2.32 53.35 6 2.48 54.27 6 2.29 L3-L3 L4-L4 L5-L5 L6-L6 L7-L7 L3-L3 L4-L4 L5-L5 L6-L6 L7-L7 27.01 6 1.62 34.67 6 1.80 40.40 6 3.52 44.82 6 1.99 50.35 6 2.27 28.18 6 2.02 38.69 6 1.82 47.25 6 2.21 55.08 6 2.03 62.59 6 2.34 27.85 6 1.75 36.18 6 1.89 41.89 6 2.05 47.21 6 1.71 53.03 6 1.90 28.59 6 1.88 40.14 6 1.97 48.90 6 2.15 57.14 6 2.00 64.84 6 1.64 28.26 6 1.81 37.17 6 1.98 43.95 6 2.14 49.97 6 1.83 56.22 6 2.35 29.74 6 1.89 41.50 6 1.90 50.96 6 2.17 59.78 6 2.06 67.88 6 1.99

37.69 6 3.79 37.11 6 2.13

33.09 6 2.02 32.54 6 1.99 32.15 6 2.09

NS, Statistically not signicant; U, Maxillary; L, mandibular; 1, central incisor; 2, lateral incisor; 3, canine; 4, rst premolar; 5, second premolar; 6, rst molar; 7, second molar. *P \0.05; yP \0.01; zP \0.001.

Wm)/(Dc/Dm) ratio. When this ratio was within a mean of 61 SD, they assumed (or labeled) the arch form as ovoid; if it was more than 11 SD, it was square; and if it was less than 1 SD, it was tapered. This step is called supervision: Noroozi et al15 intentionally dened and determined the way to classify the arch forms using a ratio variable. In orthodontics, although there are some guidelines to determine the arch-form shape, such as 1 or 2 mm out of the standard arch form or an aberration over 1 or 2 SD, these are more or less arbitrary.35 In their study, a signicant discrepancy was dened as a value outside 1 SD. This criterion is simply equivalent to the 70% (more exactly, 68.3%) condence interval of the mean. Thus, we thought it was an arbitrary denition to discriminate 1 form from another. Moreover, when the form and shape are not normally distributed but skewed, the use of mean and standard deviation units to indicate the arch form is inappropriate. On the other hand, cluster analysis is a method of unsupervised classication.36 In cluster analysis, no statistics of the data jointly with their class labels are known, so the goal is to group the objects into clusters based only on their observable features, so that each cluster contains objects

that share some important properties. This method is a common technique for statistical analysis in many elds, including pattern recognition and image analysis. An important step in any clustering is to select a distance measure, which will determine how the similarity (or the dissimilarity) of 2 elements is calculated. In our study, only the similarity measure in the shape of arch form was included in the classication model. A linear or ratio variable was not included in this classication method as a classication criterion. In addition, to ensure robustness and control for outliers, the PAM clustering method was performed. After cluster analysis, all subjects were assigned to a classication table. After the assignment, the previously measured variablestooth size, arch width, arch depth, angulations, and inclinationswere investigated, not to classify the arch forms, but to properly describe the shapes. After investigating the differences in these variables among the 3 clusters, nally we labeled these clusters as narrow, middle, and wide. Although the literature discusses many geometric and mathematical models of arch forms, such functions were developed mainly to describe dental arch form more accurately,2,15,18,20,37-39 to observe growth changes in

July 2011  Vol 140  Issue 1

American Journal of Orthodontics and Dentofacial Orthopedics

Lee et al

93

Table IV. Angulation ( ), inclination ( ), and several indexes according to arch-form clusters
Maxilla Narrow Middle Wide (n 5 96) (n 5 121) (n 5 89) Signicance Angulation U1 3.15 6 2.92 2.93 6 2.72 2.74 6 2.85 NS U2 6.46 6 3.24 6.20 6 3.87 6.24 6 2.98 NS U3 7.00 6 4.83 7.08 6 4.68 7.22 6 5.46 NS U4 3.73 6 4.55 4.99 6 4.34 4.61 6 3.76 NS U5 7.80 6 4.19 7.87 6 4.10 7.02 6 3.93 NS U6 10.98 6 4.86 11.32 6 4.83 11.68 6 5.28 NS U7 2.93 6 5.97 3.47 6 6.03 4.29 6 6.48 NS Inclination U1 14.17 6 6.59 14.30 6 6.03 14.27 6 6.01 NS U2 10.30 6 5.96 10.63 6 6.29 10.79 6 6.71 NS U3 2.20 6 5.89 0.69 6 6.32 0.28 6 5.99 * U4 7.72 6 5.97 6.58 6 6.42 4.37 6 5.99 z U5 7.72 6 6.26 7.00 6 6.38 4.42 6 6.14 z U6 5.63 6 6.09 5.14 6 5.68 1.81 6 6.08 z U7 5.94 6 6.48 5.30 6 6.29 2.01 6 5.95 z Intermolar width (6-6)/arch depth 1.43 6 0.07 1.49 6 0.10 1.59 6 0.08 z Intercanine/intermolar width (6-6) 0.66 6 0.03 0.66 6 0.03 0.63 6 0.02 z Anterior curvature (sum of anterior tooth sizes [3-3]/intercanine width) 0.65 6 0.02 0.63 6 0.02 0.63 6 0.02 z Narrow (n 5 69) L1 L2 L3 L4 L5 L6 L7 L1 L2 L3 L4 L5 L6 L7 0.60 6 2.90 0.19 6 3.30 1.58 6 4.34 1.61 6 3.63 4.31 6 4.44 7.31 6 4.23 9.17 6 4.98 1.97 6 5.63 0.82 6 5.85 8.41 6 6.09 19.44 6 6.34 24.70 6 5.95 30.61 6 5.09 34.16 6 7.32 1.36 6 0.08 0.60 6 0.04 0.66 6 0.04 Mandible Middle (n 5 137) 0.53 6 2.29 0.39 6 3.31 1.12 6 5.52 1.54 6 4.23 3.83 6 4.64 7.23 6 4.01 10.4 6 5.46 1.45 6 6.28 1.16 6 6.02 7.09 6 5.39 18.57 6 6.43 22.41 6 7.27 28.64 6 6.01 33.39 6 6.25 1.45 6 0.08 0.59 6 0.03 0.64 6 0.03 Wide (n 5 100) 0.01 6 2.49 0.41 6 3.44 0.18 6 4.27 1.38 6 4.01 3.24 6 4.37 7.18 6 4.75 8.91 6 4.92 0.26 6 6.49 2.82 6 5.61 7.81 6 5.85 18.40 6 6.32 21.53 6 7.25 26.74 6 5.27 30.99 6 5.92 1.56 6 0.09 0.57 6 0.03 0.64 6 0.03 Signicance NS NS NS NS NS NS NS * * NS NS * z y z z z

NS, Statistically not signicant; U, Maxillary; L, mandibular; 1, central incisor; 2, lateral incisor; 3, canine; 4, rst premolar; 5, second premolar; 6, rst molar; 7, second molar. *P \0.05; yP \0.01; zP \0.001.

the arch form,9,40 to compare ethnic diversity,19,41 or to evaluate orthodontic treatment responses on the arch form.6-8,12,13,17,21,42-44 However, the literature lacks any description of a method that would distinguish 1 arch form from another or methods with coefcients of function for determining classications. Our study was meant to classify dental arch forms based not on intuition or biased speculation but by using an unbiased mathematical method. In addition, the most difcult problem in classifying the dental arch form was twofold. First, projected archform data sets were inconsistent in the origin; therefore, superimposition caused random rotation. Furthermore, it is natural in subjects with normal occlusion to have some asymmetry in the arch form.45 To compare diverse arch forms and eliminate the effect of innate asymmetry, location shift and rotation transformation were necessary. Since comparing and classifying the shape with any coefcient of previously developed geometric and mathematic functions was inappropriate, an alternative approach was developed. This was done by comparing the area from 1 arch shape to another shape as a new distance measure. Second, a relatively large individual variation of dental arch forms was found in both the maxilla and the mandible.

In this study, the cubic spline curve was applied. However, several previous studies advocated the use of the higher-order polynomial curve and considered it the best representation of the dental arch.12-16 In nding a single standard arch form to explain all, it is generally true that the higher the order of a polynomial curve, the better its t. But in describing and clustering dental arches, it is not always so, since dental arches are not perfectly symmetrical, and the data points are not always positioned on a smooth line. A more general question is how good a t is it with a smooth curve? By increasing the complexity of the curve, it is possible to t the data more closely. By using as many parameters as a higher-order polynomial curve, we can t the data point exactly. But it is possible that doing this might cause overtting, which produces irregular bends on the curve. For this reason, the third-order polynomial would be a good match between the data interpretability and the model complexity. That is to say, to solve (1) the asymmetric condition, (2) a random rotation into left or right, and (3) the problem for the absence of the origin of the data points ([x,y] 5 [0,0]), we had to take several steps. First, we interpolated the data points using piece-wise linear function and performed location shift and rotation transformation. Second, after clustering,

American Journal of Orthodontics and Dentofacial Orthopedics

July 2011  Vol 140  Issue 1

94

Lee et al

a symmetric third-order polynomial function was constructed to explain the shape. Thus, using the cubic spline function was related to solving the asymmetry problem and producing a smooth curve without an irregular bend that might have been incorporated via tting a higher-order polynomial function. Recently, the trend in practice has been toward more individualized care. Along with this, clinicians should allow for relatively large individual variations that are normal biologic phenomena. In this study, we had a large sample size compared with previous studies: a total of 306 subjects with naturally occurring normal occlusion. The data included the second molar measurements. In addition, removing the outliers to adjust the homogeneity of the clusters was against our study purpose. On the contrary, we tried our best to include as much normal variation as possible. Inclusion of as much as possible normal variation is a merit in this study. In this respect, a method to control for the outliers and to preserve individual information simultaneously is essential. To do this, we used the PAM clustering method rather than the conventional K-means clustering method. Further questions remain: ie, how many arch-form types exist, and can we discriminate the number of arch-form types? These questions are difcult, and the current literature on commercial arch-form templates answers them only partially. In this study, cluster analysis helped to determine the number of arch-form classications so that the best t can be achieved with maximum between-group distances. Cluster analysis has gained popularity recently in biomedical science and is expected to be applicable to ensure both individual variation and practical convenience. The main usefulness of the silhouettes lies in the interpretation and validation of the classication results.32 What we have focused on is not the measurements or ratio variables between them, but the form or shape itself. We have tried to classify the form using a method as invariant as possible for the linear measurements, such as tooth size, and intercanine to intermolar width or depth ratio. Doing so might have a possible disadvantage that the classication could depend largely on the tooth size. Using the intercanine (or intermolar) width or depth variable, for example, also means that the position of the canines (or molars) in the dental arch is important. In addition, the tooth position along the arch is inuenced by the mesially located tooth: ie, tooth sizes. It would be impossible to concentrate on form and shape. Developing a new distance measure seemed necessary to overcome this shortcoming and to objectively distinguish form and shape by themselves without incorporating any linear or ratio variables in the clustering model.

Tooth size affected the arch-form type in that the greater the sum of the teeth, the greater the tendency to be a wider arch form. This could be the reason that earlier studies paid so much attention to tooth size, especially the sum of the 6 anterior teeth, in describing a geometric explanation for dental arch forms.1,46 We could not exclude the role of tooth size completely. The inuence exerted by arch width on the type of arch form seemed natural. Basal arch width and inclination of the posterior teeth both were signicantly different between arch-form types. This emphasizes the role of basal bone anatomy on dental arch formation, as suggested by Ronay et al.22 The amount of mandibular incisor inclination was a signicant factor that could have inuenced the arch shape as described by Mutinelli et al.4 There was no signicant difference in the interarch width ratio except for the interrst molar width ratio among the normal occlusion subjects as shown in Table IV; this can be explained in 2 ways. First, there is a certain interarch width ratio that is necessary to achieve normal occlusion. The interarch relationship varies relatively little over time.40 The second explanation is that the rst molars have some freedom in rotation and in the amount of buccal overjet, since they are the rst posterior teeth to erupt into the mouth and have no antecedents. Initially, the data were sorted by sex but, after a pilot analysis, they were pooled, since no sexual dimorphism has been shown for either the maxillary or mandibular arch forms. Unlike the toothsize comparison result that usually reports sexual dimorphism, the distribution pattern of arch shapes was not signicantly different between the sexes; this agreed with previous reports.24,47 Ethnic diversity should be considered also.48 This study was based on subjects from the Korean Standard Occlusion Study. It is possible that arch forms can vary signicantly between ethnic groups. Various studies have looked not only at the reference points of the teeth differently but also at the methods for determining the points. Some studies used the arch form, connecting incisal edges and cusp tips as landmarks,10,21,45 whereas others used the reference points on the calculated centroids of the occlusal surfaces.49 Whether the reference points used were the contact point,5 the most facial portion of the proximal contact area,23 the facial axis point on the 3-dimensional virtual model,12,22,24 or the simulated bracket bonding with glued glass bead,16 the diversity in the selection of the reference points seems to be based on the purpose of the investigation. Although Trivino et al16 classied 8 arch forms and subdivided them into small, medium, and large sizes, the method of classication followed clinical intuition for the most part.

July 2011  Vol 140  Issue 1

American Journal of Orthodontics and Dentofacial Orthopedics

Lee et al

95

In clinical orthodontic practice, before selecting a preformed archwire for a patient, the form of the arch is estimated by the clinicians eye or with the aid of a certain arch-form template. However, it is somewhat arbitrary and causes us to question how we could test goodness of t for form of the dental arch. The problem is that any preformed archwire we choose is completely arbitrary and does not take the underling anatomic conguration into account. If there is a difference between the maxilla and the mandible in arch shapes, it becomes more problematic. To eliminate this problem, it could be envisioned that, in the future, every orthodontic clinic could be equipped with an intraoral 3-dimensional scanner synchronized with an arch-form molding machine; this would solve all of the discussed problems. It is hoped that the arch-form classication method will provide a practical guide in designing and fabricating preformed archwire forms. In addition, it would be of interest to study with more sophisticated equipment such as 3-dimensional virtual images that can simulate orthodontic attachment bonding without interference by positional and structural obstacles. We thank I. H. Yang for data input in the Dephi language and H. S. An for helpful comments for improving this article.
REFERENCES 1. Hawley CA. Determination of the normal arch and its application to orthodontia. Dent Cosmos 1905;47:541-52. 2. Currier JH. A computerized geometric analysis of human dental arch form. Am J Orthod 1969;56:164-79. 3. Brader AC. Dental arch form related with intraoral forces: PR 5 C. Am J Orthod 1972;61:541-61. 4. Mutinelli S, Manfredi M, Cozzani M. A mathematic-geometric model to calculate variation in mandibular arch form. Eur J Orthod 2000;22:113-25. 5. Battagel JM. Individualized catenary curves: their relationship to arch form and perimeter. Br J Orthod 1996;23:21-8. 6. BeGole EA. A computer program for the analysis of dental arch form using the catenary curve. Comput Programs Biomed 1981; 13:93-9. 7. BeGole EA, Fox DL, Sadowsky C. Analysis of change in arch form with premolar expansion. Am J Orthod Dentofacial Orthop 1998; 113:307-15. 8. BeGole EA, Lyew RC. A new method for analyzing change in dental arch form. Am J Orthod Dentofacial Orthop 1998;113:394-401. 9. Henrikson J, Persson M, Thilander B. Long-term stability of dental arch form in normal occlusion from 13 to 31 years of age. Eur J Orthod 2001;23:51-61. 10. de la Cruz A, Sampson P, Little RM,  Artun J, Shapiro PA. Long-term changes in arch form after orthodontic treatment and retention. Am J Orthod Dentofacial Orthop 1995;107:518-30. 11. Hechter FJ. Symmetry and dental arch form of orthodontically treated patients. Dent J 1978;44:173-84. 12. Miyake H, Ryu T, Himuro T. Effects on the dental arch form using a preadjusted appliance with premolar extraction in Class I crowding. Angle Orthod 2008;78:1043-9.

13. Felton JM, Sinclair PM, Jones DL, Alexander RG. A computerized analysis of the shape and stability of mandibular arch form. Am J Orthod Dentofacial Orthop 1987;92:478-83. 14. Wakabayashi K, Sohmura T, Takahashi J, Kojima T, Akao T, Nakamura T, et al. Development of the computerized dental cast form analyzing systemthree dimensional diagnosis of dental arch form and the investigation of measuring condition. Dent Mater J 1997;16:180-90. 15. Noroozi H, Nik TH, Saeeda R. The dental arch form revisited. Angle Orthod 2001;71:386-9. 16. Trivino T, Siqueira DF, Scanavini MA. A new concept of mandibular dental arch forms with normal occlusion. Am J Orthod Dentofacial Orthop 2008;133:10.e15-22. 17. Mutinelli S, Cozzani M, Manfredi M, Bee M, Siciliani G. Dental arch changes following rapid maxillary expansion. Eur J Orthod 2008; 30:469-76. 18. Valenzuela AP, Pardo MA, Yezioro S. Description of dental arch form using the Fourier series. Int J Adult Orthod Orthognath Surg 2002;17:59-65. 19. Kasai K, Kanazawa E, Aboshi H, Richards LC, Matsuno M. Dental arch form in three Pacic populations: a comparison with Japanese and Australian aboriginal samples. J Nihon Univ Sch Dent 1997;39:196-201. 20. Braun S, Hnat WP, Fender DE, Legan HL. The form of the human dental arch. Angle Orthod 1998;68:29-36. 21. Taner TU, Ciger S, El H, Germec D, Es A. Evaluation of dental arch width and form changes after orthodontic treatment and retention with a new computerized method. Am J Orthod Dentofacial Orthop 2004;126:464-75. 22. Ronay V, Miner RM, Will LA, Arai K. Mandibular arch form: the relationship between dental and basal anatomy. Am J Orthod Dentofacial Orthop 2008;134:430-8. 23. Kook YA, Nojima K, Moon HB, McLaughlin RP, Sinclair PM. Comparison of arch forms between Korean and North American white populations. Am J Orthod Dentofacial Orthop 2004;126: 680-6. 24. Camporesi M, Franchi L, Baccetti T, Antonini A. Thin-plate spline analysis of arch form in a southern European population with an ideal natural occlusion. Eur J Orthod 2006;28:135-40. 25. Lee SJ, Lee S, Lim J, Ahn SJ, Kim TW. Cluster analysis of tooth size in subjects with normal occlusion. Am J Orthod Dentofacial Orthop 2007;132:796-800. 26. Jensen E, Kai-Jen Yen P, Moorrees CF, Thomsen SO. Mesiodistal crown diameters of the deciduous and permanent teeth in individuals. J Dent Res 1957;36:39-47. 27. Howes AE. Model analysis for treatment planning: a portion of a symposium on case analysis and treatment planning. Am J Orthod 1952;38:183-207. 28. Lee SJ, Ahn SJ, Kim TW. Clinical crown angulation and inclination of normal occlusion in a large Korean sample. Korean J Orthod 2005;35:331-40. 29. Kim SJ, Park SY, Woo HH, Park EJ, Kim YH, Moon SC, et al. A study on the limit of orthodontic treatment. Korean J Orthod 2004;34:239-45. 30. Lee SJ, Kim TW, Nahm DS. Transverse implications of maxillary premolar extraction in Class III presurgical orthodontic treatment. Am J Orthod Dentofacial Orthop 2006;129:740-8. 31. Andrews LF. Straight wire: the concept and appliance. San Diego: L.A. Wells; 1986: p. 14-31. 32. Rousseeuw PJ. Silhouettes: a graphical aid to the interpretation and validation of cluster analysis. J Comp App Math 1987;20: 53-65. 33. Kaufman L, Rousseeuw PJ. Finding groups in data: an introduction to cluster analysis. New York: John Wiley & Sons; 1990.

American Journal of Orthodontics and Dentofacial Orthopedics

July 2011  Vol 140  Issue 1

96

Lee et al

34. Ruppert D, Wand MP, Carroll RJ. Semiparametric regression. New York: Cambridge University Press; 2003. 35. Lee SJ, Ahn SJ, Lim WH, Lee S, Lim J, Park HJ. Variation of the intermaxillary tooth-size relationship in normal occlusion. Eur J Orthod 2011;33:9-14. 36. Seo SH, An H, Lee SJ, Lim WH, Kim BR. Mixed dentition analysis using a multivariate approach. Korean J Orthod 2009;39: 112-9. 37. Interlandi S. New method for establishing arch form. J Clin Orthod 1978;12:843-5. 38. Lu KH. An orthogonal analysis of the form, symmetry and asymmetry of the dental arch. Arch Oral Biol 1966;11:1057-69. 39. Biggerstaff RH. Three variations in dental arch form estimated by a quadratic equation. J Dent Res 1972;51:1509. 40. Harris EF. A longitudinal study of arch size and form in untreated adults. Am J Orthod Dentofacial Orthop 1997;111:419-27. 41. Collins BP, Harris EF. Arch form in American blacks and whites with malocclusions. J Tenn Dent Assoc 1998;78:15-8. 42. Mutinelli S, Cozzani M, Manfredi M, Siciliani G. Dental arch analysis system. Prog Orthod 2004;5:200-11.

43. Begole EA. A computer program for the analysis of dental arch form using the cubic spline function. Comput Programs Biomed 1979;10:136-42. 44. Shapiro PA. Mandibular dental arch form and dimension. Treatment and postretention changes. Am J Orthod 1974;66:58-70. 45. Cassidy KM, Harris EF, Tolley EA, Keim RG. Genetic inuence on dental arch form in orthodontic patients. Angle Orthod 1998;68: 445-54. 46. Bonwill WGA. The scientic articulation on the human teeth as founded on geometrical, mathematical and mechanical laws. Dent Items Interest 1899;21:617-43. 47. Haralabakis NB, Sifakakis I, Papagrigorakis M, Papadakis G. The correlation of sexual dimorphism in tooth size and arch form. World J Orthod 2006;7:254-60. 48. Worms FW, Speidel TM, Isaacson RJ, Meskin LH. Ponts index and dental arch form. J Am Dent Assoc 1972;85:876-81. 49. Ferrario VF, Sforza C, Miani A Jr, Tartaglia G. Maxillary versus mandibular arch form differences in human permanent dentition assessed by Euclidean-distance matrix analysis. Arch Oral Biol 1994;39:135-9.

July 2011  Vol 140  Issue 1

American Journal of Orthodontics and Dentofacial Orthopedics

Вам также может понравиться