Accuracy of space analysis with emodels and plaster models

ORIGINAL ARTICLE Accuracy of space analysis with emodels and plaster models S. Russell Mullen, a Chris A. Martin, b Peter Ngan, c and Marcia Gladwin d Leesburg, Va, and Morgantown, WVa Introduction: The purposes of this study were to determine the accuracy and speed of measuring the overall arch length and the Bolton ratio, and the time to perform a Bolton analysis for each patient by using software (emodel, version 6., GeoDigm Corp, Chanhassen, Minn) compared with hand-held plaster models. Methods: Models from 3 patients selected from the files of the Department of Orthodontics at West Virginia University were included in this study. The mesiodistal width of each tooth from first molar to first molar was measured to the nearest.1 mm with digital calipers, and the Bolton ratio was calculated for each patient. The times required to make the measurements and to perform the analysis were recorded in seconds by using a stopwatch. This process was repeated to record the digital measurements with the software. To evaluate whether there was any magnification in the emodels, quarter-inch ball bearings were mounted on a modified study model. Measurements of the greatest diameter were taken on each ball bearing by using digital calipers and the emodel software. The difference between the 2 methods was calculated, and a paired t test was used to analyze the data. Results: There was no significant difference between the Bolton ratios calculated with the 2 methods. A significant difference in arch length calculations was found between the 2 methods, but it was within the range of error found in this study and was considered clinically insignificant. Significant differences were found in the time needed to make the measurements and the calculations between the 2 methods; the emodel software was an average of 65 seconds faster. The measurements on the ball-bearing mounted models were an average of.67 mm greater on the emodel software than direct measurements on the casts (range, to.16 mm). The difference was significant (P.45). Conclusions: These results suggest that, when performing a Bolton analysis, the emodel can be as accurate as, and significantly faster than, the traditional method of digital calipers and plaster models. A clinician who has switched to using emodel software can be confident in his or her diagnoses using it. (Am J Orthod Dentofacial Orthop 27;132:346-52) Space analysis is a critical step in orthodontic diagnosis decisions when determining whether extractions are necessary to accommodate a crowded dentition. This requires comparing the total mesiodistal (MD) widths of teeth in the dental arch to the space available in that arch. In addition, to achieve functional occlusion with proper overjet and overbite, the maxillary and the mandibular teeth must be proportional in size. Several methods of space analysis are available to orthodontists, including those of Black, 1 Wheeler, 2 Neff, 3 a Private practice, Leesburg, Va. b Associate professor, Department of Orthodontics, School of Dentistry, West Virginia University, Morgantown. c Chair, Department of Orthodontics, School of Dentistry, West Virginia University, Morgantown. d Professor, Department of Dental Hygiene, School of Dentistry, West Virginia University, Morgantown. Reprint requests to: Chris A. Martin, Department of Orthodontics, West Virginia University, School of Dentistry, Health Sciences Center North, PO Box 948, Morgantown, WV 2656-948; e-mail, camartin@hsc.wvu.edu. Submitted, April 25; revised and accepted, August 25. 889-546/$32. Copyright 27 by the American Association of Orthodontists. doi:1.116/j.ajodo.25.8.44 Howes, 4 and Bolton. 5 These methods require an accurate impression of the patient s dentition and the fabrication of plaster models for measurement. Bell et al 16 comparatively assessed direct measurements of dental study models and measurements of computer-generated 3-dimensional (3D) images of the same study models and found no statistically significant differences between the measurements. Quimby et al 17 tested the accuracy, reproducibility, efficacy, and effectiveness of measurements made on computer-based models and found that those measurements appeared to be generally as accurate and reliable as measurements from plaster models. Recently, electronic storage of models became available, allowing users to store and view 3D models on a computer. This concept could eliminate the problem of model storage in an orthodontic office and shorten the time necessary to perform space analyses. However, the accuracy of emodels (version 6., GeoDigm Corp, Chanhassen, Minn) to perform space analyses has not been reported in the literature. The purpose of this study was to compare the accuracy and time to perform the Bolton analysis with emodels and plaster models. 346

American Journal of Orthodontics and Dentofacial Orthopedics Volume 132, Number 3 Mullen et al 347 Fig 1. Measurement of plaster model with digital calipers. Fig 2. Measurement with emodel software. MATERIAL AND METHODS Pretreatment models from 3 patients in the Department of Orthodontics of West Virginia University s School of Dentistry were selected. The inclusion criterion was complete adult dentition from first molar to first molar in both arches. Alginate impressions of both arches of each patient were taken and sent to GeoDigm, which fabricated a plaster model and scanned it to produce the emodel. The plaster model was returned with the emodel for measurements. The plaster model and the emodel were therefore made from the same impressions and should have had identical measurements. To determine the accuracy of performing a Bolton analysis, the study was divided into 2 parts: the first part involved making measurements on the plaster models and the emodels of 3 patients to determine overall arch length (sum of MD widths of all teeth in 1 arch from first molar to first molar), the Bolton ratio, and the time to perform a Bolton analysis for each patient. In the second part, we determined whether there was any magnification in the process of creating an emodel. To determine the intraoperator error, arch length measurements were done at 2 separate times by each method. A second set of measurements (T2) was made on 5 randomly selected patients 2 months after the first set of measurements (T1). Measurements on plaster models and emodels Digital calipers (S225, Fowler, Boston, Mass) were used to make measurements on the plaster models; all measurements were rounded to the nearest.1 mm (Fig 1). The MD width of each tooth was measured at its greatest width, by holding the calipers perpendicular to the occlusal plane of the tooth. The Bolton analysis was done by summing the MD widths of all maxillary teeth, right permanent first molar to left permanent first molar; summing the MD widths of all mandibular teeth, right permanent first molar to left permanent first molar; and dividing the mandibular sum by the maxillary sum and multiplying by 1. The times required to take all the measurements and to perform the Bolton analysis with plaster models and digital calipers were recorded in seconds with a stopwatch. This process was repeated to record the digital measurements with the emodel software (Fig 2). The data were analyzed by using a paired t test on each data set to determine the error in the measurements. Measurements on ball-bearing mounted models To determine whether there was a magnification factor associated with creating an emodel, measurements were taken of quarter-inch ball bearings (Ball Supply, Avon, Conn), Anti-Friction Bearing Manufacturers Association grade 25 (machined to within.25 in) (Fig 3). The ball bearings were mounted on a modified study model and sent to the company for scanning with the same technique as for the orthodontic models. Five ball bearings were mounted: in the left second molar region, the left second premolar region, the central incisor region, the right second premolar region, and the right second molar region. Quarter-inch ball bearings were chosen because their diameter (6.35 mm) is similar to that of a tooth. The greatest diameter was measured on each ball bearing with digital calipers and on the computerized model with the emodel software. The measurements were added, and the values were recorded. The data were analyzed by averaging the measurements of the 5 ball bearings and using paired t tests to compare the emodel with the calipers. The P value was set at.5.

348 Mullen et al American Journal of Orthodontics and Dentofacial Orthopedics September 27 Fig 3. Ball bearings of known size mounted on modified plaster cast used to determine magnification factor. RESULTS Figure 4 shows the percentage differences in Bolton ratio between measurements from the plaster models and the emodel software. There was no significant difference between the Bolton ratio calculated with the 2 methods (.5 1.87; P.86; range, 3.75 to 5.9). Figure 5 shows the differences in seconds to calculate the Bolton ratio with the plaster vs the emodel. Significant differences were found between the time taken to perform the calculations on the plaster models; it was an average of 65.6 47. seconds slower than the calculation with emodels (P.1; range, 157 to 47 seconds). Figure 6 shows the difference in mandibular arch length measurements between the plaster models and the emodels. Significant differences were found between the 2 methods of measurements. The plaster models had an average of 1.5 1.36 mm greater arch length than the emodels (P.1; range, 4.67 to.89 mm). Figure 7 shows the difference in maxillary arch length measurements between the plaster models and the emodels. Significant differences were found. The plaster model had an average of 1.47 1.55 mm greater arch length than the emodels (P.1; range, 4.65 to 1.32 mm). Intraoperator error was determined by measuring the arch length of the maxillary and mandibular casts twice at T1 and T2 by using both the digital caliper and the emodel software. With the digital caliper for measurement, significant differences were found between the measurements at T1 and T2 for the mandibular plaster cast (P.5). The second set of measurements was an average of 1.4 mm smaller than the first set. No significant differences were found between the measurements at T1 and T2 for the maxillary plaster casts. The second measurement was an average of.94 mm smaller than the first set. With the emodel software, significant differences were found between the measurements at T1 and T2 for the mandibular casts (P.5). The second set of measurements was an average of 1.9 mm smaller than the first set. Significant differences were found between the measurements at T1 and T2 for the maxillary casts (P.5). The second measurement was an average of 3.2 mm smaller than the first set. To determine whether there was magnification with the emodel software, the greatest diameter of machined quarter-inch (6.35 mm) ball bearings was measured on plaster models. The greatest diameter was also measured on the emodels at normal screen magnification, and another set of measurements was made with the emodel at high screen magnification. The diameter measured by using the emodel at normal magnification was subtracted from the diameter measured with the calipers. The measurement was found to be an average of.67 mm greater on the emodel software (range, to.16 mm; P.45). The diameter measured by using the emodel at high magnification was subtracted from the diameter measured with the calipers. The measurement was an average of.54 mm greater on the emodel software (range,.2 to.9 mm; P.1). The diameter measured by using the emodel at high magnification was subtracted from the diameter measured by using the emodel at normal magnification. The measurement was an average of.13 mm greater on the emodel at normal magnification (range,.11 to.9 mm; P.55). When we evaluated the diameter magnification of the emodel at normal magnification and subtracted from the measurements taken on plaster, the maxillary arch measurement was an average of.58 mm greater with the emodel (range,.1 to.11; P.29). For the magnification of the mandibular arch, the measurement was an average of.76 mm greater with the emodel at normal magnification (range,. to.16; P.82). When we evaluated the diameter magnification of emodel at high magnification and subtracted from the measurements taken on plaster, the measurement of the maxillary arch was an average of.46 mm greater (range,.2 to.9 mm; P.23). For the magnification of the mandibular arch, the measurement was an average of.62 mm greater with the emodel at high magnification (range,.5 to.9; P.1). When we compared the 2 magnifications, we subtracted the diameter measured with the emodel at high magnification from the same measurements with the emodel at normal magnification. The measurement of the max-

American Journal of Orthodontics and Dentofacial Orthopedics Volume 132, Number 3 Mullen et al 349 1 8 6 Difference(%) 4 2 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15 16 17 18 19 2 21 22 23 24 25 26 27 28 29 3-2 -4-6 Fig 4. Difference in Bolton ratio between plaster models and emodels. 2 15 1 Time 5 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15 16 17 18 19 2 21 22 23 24 25 26 27 28 29 3-5 -1 Fig 5. Difference in seconds to calculate Bolton ratio between plaster models and emodels. illarry arch was an average of.12 mm greater with the emodels at normal magnification (range,.8 to.6 mm; P.64). The measurement of the mandibular arch was an average of.14 mm greater with the emodels at normal magnification (range,.11 to.9; P.73). DISCUSSION We found no significant difference between the Bolton ratios calculated using plaster models and emodels. The mean difference was.5 1.87. Calculation of the Bolton ratio with emodel was just as accurate as the traditional method of using calipers to measure the plaster models. According to Shellhart et al, 6 Bolton ratio discrepancies can vary by as much as 2.2 mm with needle-pointed dividers, making the difference between the emodels and the plaster models clinically insignificant. Nie and Lin 7 calculated the Bolton ratios of 6 patients using a 3D measuring machine, and found a standard deviation of 2.64. Tomassetti et al 8 found that, when a Bolton ratio was calculated 3 times with Vernier calipers on plaster models, 72.7% of the measurements fell within 1. mm of each other (range, to 2.8 mm). In that study, QuickCeph computer software was used to estimate the Bolton ratio, and it differed from measurements of the Vernier calipers by a mean of 1.84 mm. In addition, 52.4% of the measurements were within 1.4 mm, and 81.% were within 2.5 mm. The difference between the 2 methods was found to be insignificant. In our study, it took less time to calculate the Bolton ratio with the emodel software by an average of 65.6 seconds. The time to make the measurements was probably about the same, but, when calculating with plaster models, it is necessary to write

35 Mullen et al American Journal of Orthodontics and Dentofacial Orthopedics September 27 5 Lower cast measurement P-M 4 3 Difference (mm) 2 1 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15 16 17 18 19 2 21 22 23 24 25 26 27 28 29 3-1 -2 Fig 6. Difference in millimeters between plaster and emodel measurements on mandibular cast. 5 Upper cast measurement P-M 4 3 Difference (mm) 2 1 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15 16 17 18 19 2 21 22 23 24 25 26 27 28 29 3-1 -2 Fig 7. Difference in millimeters between plaster and emodel measurements on maxillary cast. down the measurements and make the calculation with either a calculator or a computer. When calculating the Bolton ratio with emodel software, it is not necessary to write the tooth measurements, and the calculation is done at the click of a button, giving the software an edge when it comes to time involved in calculating a Bolton ratio. Tomassetti et al 8 found a significant difference when comparing the time it takes to calculate a Bolton ratio using QuickCeph software with the Vernier calipers, with mean times of 1.85 minutes for the software and 8.6 minutes for the calipers. We also found a significant difference in the calculation of arch length and tooth structure in both arches with the 2 methods of measurement. The amount of tooth structure in the mandibular arch measured with emodel software was an average of 1.5 1.36 mm smaller than that measured on the plaster model. The amount of tooth structure in the maxillary arch measured with emodel software was an average of 1.48 1.55 mm smaller than that measured on plaster model. Bell et al 16 comparatively assessed the direct measurements of 22 study models and measurements of computer-generated 3D images using a photostereometric technique of the same study models and found no statistically significant difference between measurements of the dental casts

American Journal of Orthodontics and Dentofacial Orthopedics Volume 132, Number 3 Mullen et al 351 and the 3D images; the average difference between those measurements was.27 mm. This difference was within the range of operator errors (.1-.48 mm) and was not statistically significant (P.5). Quimby et al 17 tested the accuracy, reproducibility, efficacy, and effectiveness of measurements made on 5 computerbased models and found that those measurements appeared to be generally as accurate and reliable as those from plaster models. Motohashi and Kuroda 9 compared a 3D computer-aided design system with digital calipers to measure teeth and found no significant difference between the 2 methods at a level of 1%. Their method to scan the dental model with a laser and a computer was similar to that used by the company to fabricate the emodel. Kojima et al 1 evaluated a method for scanning dental casts into digital images using lasers and a camera; they found a measurement of.3 mm from a point on a premolar to a point on the contralateral premolar. Schirmer and Wiltshire 11 examined the difference between manual and computeraided space analysis. They measured MD widths of teeth using a Vernier gauge and found it to be highly accurate between 2 examiners. To take digital measurements, they photocopied the models and digitized them. They found significant differences between manual and digital measurements. The average discrepancies in arch length were 4.7 mm in the maxilla and 3.1 mm in the mandible. These differences were greater than, but similar to, our findings. Schirmer and Wiltshire 11 also found the digitized measurements to be smaller than the manual measurements. They attributed this to the difficulty of measuring a 3D model in 2 dimensions, because of the convex structure of the teeth, the curve of Spee, and differences in inclinations of the teeth. In our study, the intraoperator errors for the plaster casts were 1.4 mm in the mandibular arch and.93 mm in the maxillary arch. These values were within the ranges described in previous studies. 12-14 The difference in the repeated measurements with emodel software was slightly larger than the difference between the plaster models. The difference between the 2 emodel measurements might be partly attributed to the change in software from version 6. (T1) to version 7. (T2). The technique for making the measurements was similar for both versions, but even a small change could produce the difference seen in our study. This difference in the measurements between the digital calipers and the emodel software could be attributed to several factors. One was the difficulty of finding the greatest MD width of the teeth with the emodel software. The user can rotate the cast on the screen to accurately assess the points chosen as the greatest diameter, but this process is still difficult. The resolution of the emodels is high, but it is difficult to choose the exact contact point between 2 teeth. In some cases, the interproximal area between the teeth is not well defined enough for certainty that the greatest MD diameter is being measured. An example of this can be seen in the interproximal area between the canine and the incisor in Figure 2. In other cases, the interproximal areas are well defined and easy to see, but it can still be difficult to get a measurement at the tangent perpendicular to the greatest diameter. This is illustrated in the incisor area of Figure 2. When estimating the contact areas, the operator will tend to underestimate the measurement, leading to a discrepancy of about 1.5 mm less tooth structure per arch. The difference of 1.5 mm could be clinically insignificant, because, according to Proffit, 15 a tooth-size difference of less than 1.5 mm is not considered significant. For the machined ball bearings, which had no contact with each other, the measurements on emodel software were an average of.54 mm greater than those on the plaster models. This can be attributed to the difficulty of finding the point with the greatest diameter. When measuring something without contact with adjacent objects, the operator will tend to draw a tangent a little larger than the object, making the measurements a little larger on the emodels than on the plaster models. CONCLUSIONS The accuracy of a space analysis such as the Bolton ratio was found to be similar with either the emodels or the plaster models. The difference between the Bolton ratio calculations was statistically insignificant. The difference between the arch length calculations was statistically significant but within the range of error found in this and other studies, and was considered clinically insignificant. The times taken to make the measurements and the calculations were statistically and clinically significant; the emodel software was an average of 65 seconds faster. Emodel software for measuring a patient s dentition and calculating the Bolton ratio is just as accurate and faster than using digital calipers with plaster models. A clinician who has switched to emodel software can be confident in his or her diagnoses using it. REFERENCES 1. Black GV. Descriptive anatomy of the human teeth. Philadelphia: S. S. White Dental Manufacturing; 192.

352 Mullen et al American Journal of Orthodontics and Dentofacial Orthopedics September 27 2. Wheeler RC. A textbook of dental anatomy and physiology. Philadelphia: W. B. Saunders; 1961. 3. Neff CW. Tailored occlusion with the anterior coefficient. Am J Orthod 1949;35:39-14. 4. Howes AE. Case analysis and treatment planning based upon the relationship of the tooth material to its supporting bone. Am J Orthod Oral Surg 1947;33:499-533. 5. Bolton A. Disharmony in tooth size and its relation to the analysis and treatment of malocclusion. Angle Orthod 1958;28: 113-3. 6. Bell A, Ayoub AF, Siebert P. Assessment of the accuracy of a three-dimensional imaging system for archiving dental study models. J Orthod 23;3:219-23. 7. Quimby M, Vig K, Rashid R, Firestone A. The accuracy and reliability of measurements made on computer-based digital models. Angle Orthod 24;74:298-33. 8. Shellhart WC, Lange DW, Kluemper GT, Hicks EP, Kaplan AL. Reliability of the Bolton tooth-size analysis when applied to crowded dentitions. Angle Orthod 1995;65:327-34. 9. Nie Q, Lin J. Comparison of intermaxillary tooth size discrepancies among different malocclusion groups. Am J Orthod Dentofacial Orthop 1999;116:539-44. 1. Tomassetti JJ, Taloumis LJ, Denny JM, Fischer JR Jr. A comparison of 3 computerized Bolton tooth-size analyses with a commonly used method. Angle Orthod 21;71:351-7. 11. Motohashi N, Kuroda T. A 3D computer-aided design system applied to diagnosis and treatment planning in orthodontics and orthognathic surgery. Eur J Orthod 1999;21:263-74. 12. Kojima T, Sohmura T, Wakabayashi K, Nagano M, Nakamura T, Takashima F, et al. Development of a new high-speed measuring system to analyze the dental cast form. J Dent Mater 1999;18: 354-65. 13. Schirmer UR, Wiltshire WA. Manual and computer-aided space analysis: a comparative study. Am J Orthod Dentofacial Orthop 1997;112:676-8. 14. Richmond S. Recording the dental cast in three dimensions. Am J Orthod Dentofacial Orthop 1987;92:199-26. 15. McCann J, Burden DJ. An investigation of tooth size in Northern Irish people with bimaxillary dental protrusion. Eur J Orthod 1996;18:617-21. 16. Johal AS, Battagel JM. Dental crowding: a comparison of three methods of assessment. Eur J Orthod 1997;19:543-51. 17. Proffit W. Contemporary orthodontics. St Louis: Mosby; 2. p. 118-2.