Medical knowledge | Dentistry » Luthardt-Sandkuhl - Accuracy of mechanical digitizing with a CAD CAM system for fixed restorations

Ralph G. Luthardt, Dr Med Denta Olaf Sandkuhl, Dipl Ingb Volker Herold, Dr Ing Habilc Michael H. Walter, Prof Dr Med Dentd Accuracy of Mechanical Digitizing with a CAD/CAM System for Fixed Restorations Purpose: Most research on the accuracy of dental computer-aided design/manufacturing (CAD/CAM) systems focuses on the marginal gap. A detailed analysis of the various components of CAD/CAM systems (digitizer, software, and milling machine) using this approach is nearly impossible. The purpose of this study was to determine the accuracy of the manual mechanical digitizer of the Precident-DCS system. Materials and Methods: Gauge blocks were aligned to the coordinate planes of the digitizer to determine the point and length measurement uncertainty. The values for each measuring point were compared for statistical differences concerning first versus second operator, pressure, and mode of sensing using analysis of variance. The measurement uncertainty was given by 95th percentiles. Results:

The mean one-dimensional point measurement uncertainty in the Y direction was 11 µm for the first, 8 µm for the second, and 37 µm for both operators. The three-dimensional point measurement uncertainty in the Y direction was 10 µm for the first, 33 µm for the second, and 60 µm for both operators. The point measurement uncertainty was significantly influenced by the pressure during sensing and by the operator as well. There were significant differences between the first and second recordings. The length measurement uncertainty in the Y direction for a gauge block of 20 mm was 52 µm for both operators. Conclusion: The reliability of the manually guided Precident-DCS digitizer is limited because of the significant influence of the operator and the mode of sensing (one or three dimensional). Compared with an automatic digitizing system, the measurement uncertainty of the manually guided digitizer exceeds the values published in the literature. Int J Prosthodont 2001;14:146–151 T

quality.1–3 Marginal fit is a paramount factor for quality assessment of fixed restorations4–6 Clinical trials have underlined the importance of marginal accuracy for clinical success.4,7 The CAD/CAM process for the fabrication of crowns and fixed partial dentures includes digitizing, CAD, and CAM. In performing a systematic analysis of the precision of dental CAD/CAM procedures, an examination of the accuracy of the various components of the system is required.8 Very little research has been done on errors caused by the digitizing process9,10 The three-dimensional, complex free-form surface data of manually prepared teeth have been acquired by optical or mechanical digitizers.10,11 Extraoral mechanical digitizing seems to be the more accurate procedure compared with intraoral optical digitizing techniques.6,12 Extraoral mechanical digitizing techniques are used by the contact probe device of the he idea behind computer-aided design/manufacturing (CAD/CAM) in dentistry is the

cost-effective manufacturing of restorations with similar or improved aSenior Instructor, Department of Prosthetic Dentistry, Dresden University of Technology, University Hospital Carl Gustav Carus Dresden, Dental School, Dresden, Germany. bEngineer, Department of Surface Machining, Friedrich-SchillerUniversity Jena, Technical Institute, Jena, Germany. cEngineer and Senior Instructor, Department of Surface Machining, Friedrich-Schiller-University Jena, Technical Institute, Jena, Germany. dProfessor and Head, Department of Prosthetic Dentistry, Dresden University of Technology, University Hospital Carl Gustav Carus Dresden, Dental School, Dresden, Germany. Reprint requests: Dr Ralph Luthardt, Department of Prosthetic Dentistry, Dresden University of Technology, Fetscherstraße 74, 01307 Dresden, Germany. Fax: + 0049-351-4585314 e-mail: Ralph.Luthardt@mailboxtu-dresdende The International Journal of Prosthodontics 146 Volume 14, Number 2, 2001 Luthardt et al Accuracy of

Mechanical Digitizing with CAD/CAM Procera system (Nobel Biocare)4,5,10 and the manually guided probe of the Precident-DCS system (DCS Production).6,11,13 The sapphire ball at the tip of the measuring probe of the Procera reader is held to the surface of the rotating die and moved up and down at a constant speed. While the whole die surface is registered in one reading by the Procera digitizer, the tungsten probe of the Precident-DCS system has to be guided manually across the surface step by step. The digitizer software accepts measuring values within a specified range of pressure. Contrary to the Procera reader, the die surface can be sensed by any point of the probe tip of the Precident-DCS digitizer during the circumferential sampling of data. Persson et al10 found a maximum shape-related error of ± 10 µm for the contact probe device of the Procera system. Data by the manufacturer give a digitizing accuracy of 3 to 5 µm for the manually guided probe of the Precident-DCS

system.2,11 The mean marginal discrepancy is 61 ± 37 µm for Procera titanium crowns,4 or 56 ± 21 µm and 63 ± 13 µm, respectively, for Procera AllCeram premolar and molar crowns.7 However, the marginal fit means of titanium copings fabricated with the Precident-DCS system range from 21 ± 15 µm to 82 ± 25 µm6 and from 111 to 165 µm.11 The mean marginal discrepancy of titanium copings is 93 to 135 µm, and 55 to 88 µm for crown frameworks manufactured with zirconia ceramics.14 Whereas data on the Procera digitizer10 and the Cerec system (Sirona)9,15 are available, the manual digitizer of the Precident-DCS system has not yet been investigated. To determine the reason for the differences in the marginal gap of crowns manufactured with the Precident-DCS system and to realize the full potential of the system, the errors that could occur in each phase in the production process must be known. The accuracy of coordinate-measuring machines of different construction and automation can

be compared by the measurement uncertainty.16 This study was undertaken to determine the 1-D and 3-D point measurement uncertainty and the 3-D length measurement uncertainty of the mechanical digitizer of the Precident-DCS system. Other variables affecting the measurement uncertainty, which include the operator and applied pressure, were also examined. allowing the manufacturing of five-unit fixed partial denture frameworks. The 3-D shape of the prepared abutment teeth is acquired by the operator through a manually guided movement of the probe on the master cast (Fig 1). The method of data acquisition was described by Samet et al11 and Besimo et al.6 The specification of measurement uncertainty is intended to permit comparison of the accuracy of coordinate-measurement machines (digitizing systems) of different construction. Point and length measurement uncertainty were determined according to ENISO 10360-217 and VDI/VDE 261716 regarding the variables (1) operator, (2) pressure during

sensing, (3) mode of sensing (1-D, 3-D), and (4) differences between the first and second recordings. The assessment of the point measurement uncertainty is done by repeated measurements of a particular point. The assessment of the length measuring uncertainty is done by determining the distance between two points located on parallel surfaces some distance apart. For the assessment of the measurement uncertainty, the calibration software of the digitizer was used. The calibration software displays the three coordinates X, Y, and Z, representing the position of the mechanical probe given in digitizer units inside the measurement range. For the determination of the measurement uncertainty, gauge blocks were placed on the spheroid joint–based holder of the digitizer and were aligned parallel to the XY plane. The alignment was checked by guiding the probe on the measuring surface in both axes. The measurement lines followed the borders of the mechanically limited measuring range of the

digitizer. The alignment was regarded to be correct if the differences in the coordinate line perpendicular to the measurement plane were less than 2 digitizer units between start and end All experiments were performed on two days. During the measurements, the gauge block remained in the same position. Each experiment was performed by two experienced operators, both being first as well as second operator in each step. Regarding the pressure manually applied to the mechanical probe, the operators were requested to sense with (1) a pressure just strong enough to exceed the lower threshold (minor pressure) or (2) the maximum force just below the upper threshold (high pressure). The other operator recorded the values for the coordinate line under investigation as displayed by the calibration software. The 1-D measurement uncertainty was determined in a measurement line parallel to a coordinate line of the digitizer The 3-D measurement uncertainty was assessed by measurements at an angle to

the orientation of the coordinate lines (Fig 2). For the determination of the point measurement uncertainty, specific points on the first gauge block (100 Materials and Methods A mechanical sensor using a tungsten probe, which is movable in three axes, characterizes the PrecidentDCS digitizer. The movements of the axes, as well as the forces applied to the probe, were measured using internal units (digitizer units). The maximum measurement range is about 70 mm  40 mm  20 mm, Volume 14, Number 2, 2001 147 The International Journal of Prosthodontics Accuracy of Mechanical Digitizing with CAD/CAM Luthardt et al Fig 1 (left) Acquiring the 3-D surface data with the tungsten probe of the Precident-DCS digitizer. Fig 2a (below left) Method of 1-D measurement of the length measurement uncertainty. Fig 2b (below) Method of 3-D measurement of the length measurement uncertainty. X X direction, 1-D Y Z X direction, 3-D X mm  35 mm  10 mm) were sensed ten times by each operator

both with minor and high pressure: cases by not more than the measurement uncertainty u (|La – Lr| ≤ u); 47.5% of the measurements are lower/higher than the median. Data showed no normal distribution Therefore, the calculation of the interval ± 2 standard deviations was not feasible Because of the multimodal distribution of the measurements, the measurement uncertainty was calculated using the 95th percentile (2.5% to 975%) The results for each measuring point were tested for significant differences with regard to operator, pressure during sensing, mode of sensing (1-D, 3-D), and differences between the first and second recordings by analysis of variance (ANOVA). The mean of the measurement uncertainty was given by the 95th percentile divided by 2. Additionally, the mean measurement uncertainty was converted into µm using the conversion value calculated of the mean length measurement uncertainty in digitizer units, divided by the known length of the gauge in µm (1 digitizing

unit = 4.9 µm) • In the Z direction (left and right borders of the measurement range [1-D, 3-D]) • In the Y direction (left and right borders of the measurement range [1-D, 3-D; Fig 3, points a, g]) • In the Y direction (in 2,000 digitizer unit steps on the X axes, starting at x = 2,000 [1-D; Fig 3, points b, c, d, e, f]) The length measurement uncertainty is related to the determination of the distance between two points located on parallel surfaces. For the determination of the length measurement uncertainty, a second (length 20 mm) and a third (length 30 mm) gauge block were used. The gauge blocks were aligned to the YZ coordinate plane (X direction) and to the XZ coordinate plane (Y direction), respectively. The measurement surfaces of the gauge block were sampled alternately (Fig 2). The measuring point was always located in the center of the measured surface. According to EN-ISO 10360-2,17 the indicated value La deviates from the true value Lr in 95% of all The

International Journal of Prosthodontics Results The 95th percentiles of the 1- and 3-D point measurement uncertainty in the Z and Y directions 148 Volume 14, Number 2, 2001 Luthardt et al Accuracy of Mechanical Digitizing with CAD/CAM varied depending on the operator and pressure applied (Table 1). The mean point measurement uncertainty ranged from 5 to 191 µm (1-D), and from 10 to 45 µm (3-D). The calculated overall point measurement uncertainties ranged from 37 to 61 µm because of the scattering of the single measurement uncertainties. The 1-D point measurement uncertainties in the Z direction were significantly different (P < 005) for the operator and the pressure No significant difference (P > 0.05) was found for sensing at the right border with high pressure with either operator. The means for the left border were not statistically different when compared with the first measurement, whereas the mean values for the right border were statistically different. The 1-D

point measurement uncertainties in the Y direction were significantly different (P < 0.05) for the operator and the pressure. No significant difference (P > 005) was found for the locations x = 4,000, x = 8,000, and right border, when sensing with high pressure with either operator. The mean values for the 3-D point measurement uncertainty were statistically different for both locations when compared to the first measurement. Statistical analysis of the differences between the first and second experiments for measurements in the Y direction was omitted for reasons of the experimental design (setting zero for the second experiment). The 1- and 3-D point measurement uncertainties were significantly different (P < 0.05) for the operator, the pressure, and the mode of sensing (1-D, 3-D). The 95th percentiles of the 1- and 3-D length measurement uncertainties in the X direction varied depending on the operator, whereas the length measurement uncertainty in the Y direction showed an

increased scattering for the first operator (Table 2). The mean length measurement uncertainty in the X direction ranged from 12 to 14 µm (1-D), and from 14 to 17 µm (3-D). Corresponding to the results of the point measurement uncertainty, the calculated 1- and 3-D overall length measurement uncertainties were 92 µm and 60 µm, respectively, in the X direction, and 53 µm (1-D) in the Y direction because of the scattering of the single measurement uncertainties. The length measurement uncertainties in the X direction were significantly different (P < 0.05) for the operator and the mode of sensing (1D, 3-D) No significant differences (P > 005) were found for the length measurement uncertainty in the Y direction. X Y Z Y direction, 1-D g f e d c b a Fig 3 Method of estimating the 1-D measurement point measurement uncertainty in the Y direction with measurement points a = left border of the measurement range; b, x = 2,000; c, x = 4,000; d, x = 6,000; e, x = 8,000; f, x =

10,000; g = right border of the measurement range. shape mathematically.10,11 Thus, if real tooth geometries had been used, the procedure applied, with direct analysis of the measurements, would not have been possible. The method used provides the opportunity to evaluate the digitizer as an isolated part of the Precident-DCS system. The measurements were not influenced by other components. The measurement uncertainty of the digitizer could have been affected by handling variables (operator and applied pressure) and system variables (mechanical or electric modules of the coordinate-measuring machine and mechanical stiffness of the holder). The measurements were carried out by experienced operators to minimize individual errors. Therefore, the number of operators was limited to two. In everyday work, the operator tries to guide the probe with constant pressure directed to the center of the die. This procedure should be simulated by the realistic approach of digitizing with subjective

constant pressure, on one hand with minor pressure just higher than the lower threshold, and on the other hand with high pressure just below the upper threshold. System variables were not varied Persson et al10 evaluated the accuracy of the Procera reader measuring a high-precision manufactured square gauge, which was measured in a coordinate-measuring machine. According to these experiments,10 the gauge was aligned to the coordinate line of the measuring device. This alignment of the test equipment is a common method described in ENISO 10360-2. 17 To compare the read data file (Procera) and the ideal calculated file (coordinatemeasuring machine), a special computer program must be used. In contrast, in the present experiment, Discussion A more clinical study design cannot be used because it is almost impossible to describe a tooth Volume 14, Number 2, 2001 149 The International Journal of Prosthodontics Accuracy of Mechanical Digitizing with CAD/CAM Table 1 Luthardt et al

Point Measurement Uncertainty in Z and Y Directions at Specific Points (95th Percentile in Digitizer Units [DU]) First examiner Minor pressure High pressure Location 1-D Z direction left border Z direction right border Y direction left border Y direction X = 2000 DU Y direction X = 4000 DU Y direction X = 6000 DU Y direction X = 8000 DU Y direction X = 10,000 DU Y direction right border 3-D Z direction left border Z direction right border Y direction left border Y direction right border Second examiner Minor pressure High pressure Both examiners minor + high pressure –5 –3Ba –7 –6Bb 5913 5917B* 5916 5917B* 5915 5917B* 5911 5913B* 5911 5913B* 5909 5911B* 5902 5905B* 0 6Aa –1 8Ab 5908 5909A* 5906 5913A* 5902 5910A* 5902 5905A* 5897 5900A* 5895 5900A* 5892 5893A* –9 –6Da –7 –4Cb 5904 5909D* 5904 5909D* 5908 5913C* 5905 5910D* 5906 5909C* 5902 5905C* 5894 5897C* 5 12Ca 6 4Ab 5895 5897C* 5902 5906C* 5905 5908A* 5904 5907C* 5898 5901A* 5898 5903A* 5891 5893A* –8

12* –9 10* 5899 5915* 5902 5922* 5902 5916* 5902 5914* 5897 5915* 5897 5910* 5889 5902* –7 –2Fb –36 –5Eb 5912 5915F* 5908 5914E* 3 6Eb 1 6Db 5898 5903E* 5898 5901D* –1 6Hb –7 –5Fb 5904 5911H* 5888 5893G* 13 8Gb 4 7Db 5882 5902G* 5886 5893F* –4 12* –12 11* 5893 5917* 5888 5914* *A statistical analysis was impossible for technical or statistical reasons. Multifactorial ANOVA was used elsewhere Within each location, mean values with the same upper-case superscripted letter were not statistically significantly different (P > 0.05); within each location, mean values with the lower-case subscripted letter a were not statistically significantly different from the first measurement (P > 005), whereas b indicates statistical difference. Table 2 Length Measurement Uncertainty in X and Y Directions at Specific Points (95th Percentile in Digitizer Units [DU]) Location 1-D X direction Y direction 3-D X direction Y direction First examiner Second examiner Both

examiners 4122–4128A 6100–6138A 4104–4109B 6116–6123A 4098–4136* 6104–6126* 4118–4124C 4104–4109D 4101–4126* *A statistical analysis was impossible for technical or statistical reasons. Multifactorial ANOVA was used elsewhere. Within each location, mean values with the same upper-case superscripted letter were not statistically significantly different (P > 0.05) the measurement uncertainty was calculated directly without best-fit adjustment. Consequently, the calculated measurement uncertainty is not influenced by errors of best-fit adjustment. Errors that could occur in each phase of the production process must be known.10 Because the errors during impression taking and fabrication of the models are not controllable for clinical reasons, it is necessary to determine and estimate the dimensional errors that might occur in data acquisition and the following processes. The results for the 1-D point measurement uncertainty of the Precident-DCS digitizer

indicate that The International Journal of Prosthodontics the measurement of single points by one operator is possible with high accuracy in the horizontal (6.7 to 11.3 µm in the Y direction) and axial directions (48 to 19.1 µm in the Z direction) The determination of the overall 1-D measurement uncertainty revealed that the mechanically guided digitizer of the Precident-DCS system is influenced by the user and the pressure applied during sensing. The data acquisition of master dies is done by a circumferential sensing of the die.13 In practice, this circumferential sensing leads to variations in pressure. Therefore, the significant differences of the applied pressure limit the accuracy of digitizing 150 Volume 14, Number 2, 2001 Luthardt et al Accuracy of Mechanical Digitizing with CAD/CAM complex, free-form geometries. The significant operator-dependent differences limit the reliability of fit These results confirmed findings of in vitro measurements of the vertical

marginal gap, which had indicated a reduced reliability concerning the shape of copings after repeated digitizing of the same sample die. One reason for differences in shape after repeated digitizing may be deviations of the measured data files caused by variations in pressure during sensing. The significant influence of the mode of sensing indicates that in the step-by-step data acquisition process, errors could increase because of interruptions. These findings were confirmed by the results for the length measurement uncertainty. However, the 1- and 3-D length measurement uncertainties in the X direction ranged from 11.9 to 167 µm Nevertheless, the overall data were significantly influenced by the operator and the mode of sensing. Compared with the Procera digitizer,10 the Precident-DCS system showed an increased measurement uncertainty. Benz and Schwarz9 found that repeated measurements with the Cerec-1 device delivered pixel Z data within a range of ± 49 µm. The mechanical

digitizer of the Precident-DCS system showed a resolution of about 49 µm, which is superior to the results measured with the Cerec-1 device However, the accuracy of 3 to 5 µm determined in experiments undertaken by the manufacturer2 might be the resolution of the digitizer. The lack of a standard method for accuracy assessment frequently leads to inaccurate definition of accuracy and resolution.12 Pfeiffer and Schwotzer15 found that the measurement accuracy of the Cerec-2 device in applicationrelevant volumes should be ± 25 µm and ± 90 µm in the full measurement range. The mean 1- and 3-D measurement uncertainty of the Precident-DCS digitizer is superior in comparison with the accuracy reported for the Cerec device.15 However, the overall length measurement uncertainty is similar to our results. Nevertheless, it has to be taken into consideration that the Cerec device is used for intraoral data acquisition without conventional impression techniques, while the accuracy of the

Precident-DCS system is dependent upon the accuracy of the clinical impression. Future research should be done investigating the accuracy of new intraoral and extraoral optical highspeed data-acquisition systems. The accuracy of these systems should be investigated in comparison with mechanical measuring systems. The measurement of the point or length measurement uncertainty allowed the examination of different variables (operator, system) without the influence of clinical or technical factors. A reduced reliability of the manually guided Precident-DCS digitizer was shown by the significant influence of the operator and the mode of sensing (1-D, 3-D). Compared with other systems, the Volume 14, Number 2, 2001 measurement uncertainty of the manually guided digitizer (Precident-DCS) exceeds the values found with the automatically digitizing Procera system. Therefore, innovative digitizers should use procedures that are not influenced by manual factors or personal skills.

Acknowledgment This investigation was supported by research grant B 403-97-005 from the Thuringian Ministry of Sciences, Research and Culture, Erfurt, Germany. References 1. Nathaniel P CAD/CAM dentistry Ann R Aust Coll Dent Surg 1996; 13:99–107. 2. Willer J, Rossbach A, Weber HP Computer-assisted milling of dental restorations using a new CAD/CAM data acquisition system. J Prosthet Dent 1998;80:346–353 3. Russell MM, Andersson M, Dahlmo K, Razzoog ME, Lang BR A new computer-assisted method for fabrication of crowns and fixed partial dentures. Quintessence Int 1995;26:757–763 4. Karlsson S The fit of Procera titanium crowns An in vitro and clinical study. Acta Odontol Scand 1993;51:129–134 5. Odén A, Andersson M, Krystek-Ondracek I, Magnusson D Fiveyear clinical evaluation of Procera AllCeram crowns J Prosthet Dent 1998;80:450–456. 6. Besimo C, Jeger C, Guggenheim R Marginal adaptation of titanium frameworks produced by CAD/CAM techniques. Int J Prosthodont

1997;10:541–546. 7. May KB, Russell MM, Razzoog ME, Lang BR Precision of fit: The Procera AllCeram crown. J Prosthet Dent 1998;80:394–404 8. Andersson M, Carlsson L, Persson M, Bergman B Accuracy of machine milling and spark erosion with a CAD/CAM system. J Prosthet Dent 1996;76:187–193. 9. Benz C, Schwarz P How precise is the optical Cerec impression? [in German]. Dtsch Zahnarztl Z 1991;46:632–634 10. Persson M, Andersson M, Bergman B The accuracy of a high-precision digitizer for CAD/CAM of crowns J Prosthet Dent 1995;74: 223–229. 11. Samet N, Resheff B, Gelbard S, Stern N A CAD/CAM system for the production of metal copings for porcelain-fused-to-metal restorations. J Prosthet Dent 1995;73:457–463 12. Hewlett ER, Orro ME, Clark GT Accuracy testing of three-dimensional digitizing systems Dent Mater 1992;8:49–53 13. Luthardt RG, Rieger W, Musil R Grinding of zirconia-TZP in dentistryCAD/CAM-technology for the manufacturing of fixed dentures. In: Sedel L, Rey C (eds) 10th

International Symposium on Ceramics in Medicine Bioceramics, 5–8 Oct 1997, Paris. Oxford: Elsevier, 1997:437–440. 14. Luthardt RG, Herold V, Sandkuhl O, Reitz B, Knaak JP, Lenz E High-performance ceramics for crowns [in German]. Dtsch Zahnarztl Z 1998;53:280–285. 15. Pfeiffer J, Schwotzer A Three dimensional optical measurement of teeth [in German]. Technisches Messen 1996;63:254–261 16. Verband Deutscher Ingenieure/Verband Deutscher Electrotechniker VDI/VDE-2617 Accuracy of coordinate measuring machines Characteristic parameters and their checkingMeasurement task specific measurement uncertainty length measurement uncertainty. Berlin: Beuth, 1986 17. International Standards Organization EN-ISO 10360-2 Coordinate metrologyPart 2: Performance assessment of coordinate measuring machines. Berlin: Beuth, 1995 151 The International Journal of Prosthodontics