To analyse the marginal fit of 4-unit fixed dental prostheses (FDPs) and the accuracy of three-dimensional cast-datasets using both approaches to Computer Aided Design (CAD)/Computer Aided Manufacturing (CAM): direct and indirect digitalization.
A titanium model of a 4-unit FDP was digitized by an intraoral scanning device (iTero, Align Technology, Carlstadt, US; DD, n = 12). Additionally 12 conventional impressions were taken and referring master casts were digitized by a laboratory scanner (CS2, Straumann, Basel, Switzerland; ID, n = 12). Frameworks were fabricated (CARES CADCAM GmbH, Straumann, Markkleeberg, Germany) from base metal alloy (coron, Straumann; DD-C: n = 12; ID-C: n = 12) and zirconia (zerion, Straumann; DD-Z: n = 12; ID-Z: n = 12) from the same datasets. The marginal fit of the resulting frameworks and the accuracy of the underlying datasets from DD and ID were evaluated. Data were analyzed by unpaired two sample Student’s t -test with Levene-test ( p < 0.05).
Frameworks from group DD-C showed significantly better marginal fit than ID-C (DD-C: 56.90 ± 27.37 μm, ID-C: 90.64 ± 90.81 μm). For zirconia frameworks no differences between both digitalization methods (DD-Z: 127.23 ± 66.87 μm, ID-Z: 141.08 ± 193.17 μm) could be observed. Base metal alloy frameworks exhibited significantly better marginal fit than zirconia frameworks (DD: p < 0.001; ID: p = 0.022). Regarding the accuracy group DD showed significantly higher “trueness” than ID.
Direct and indirect digitalization lead to clinically acceptable marginal fit of 4-unit FDPs from base metal alloy and zirconia. Higher accuracy of datasets from DD leads to better marginal fit of frameworks from base metal alloy but not for ones from zirconia.
Dental restorations produced by Computer Aided Design (CAD)/Computer Aided Manufacturing (CAM) have experienced increasing importance over the last decades . The transformation of the clinical situation into a three-dimensional dataset, thus the digitalization, builds the initial step of the fabrication-process and is followed by CAD and CAM of the restoration . Generally, with direct and indirect digitalization two accesses to the digital workflow are available .
The method of indirect digitalization in the dental laboratory starts with a conventional impression using materials like hydrocolloid, polyether or polysiloxane, which all exhibits an adequate stability and precision . However, some drawbacks can be associated with this well-known procedure as the applied impression-technique, impregnated debris, tearing of the impression material and indistinct preparation margins have an effect on the quality of conventional impressions . All these aspects can negatively influence the fit of the final dental restoration. The indirect digitalization process itself is carried out in the dental laboratory by either digitizing the impression itself or the gypsum cast by optical, mechanical or computed tomography measuring techniques . The other alternative to access the digital workflow is the direct digitalization in the patient’s mouth. All intraoral scanning devices work on basis of optical principles for data acquisition. Scanning systems based on the confocal measurement principle for digitalization – as used in this study – offer the possibility to capture intraoral structures without any powder application.
However, the question arises, if direct and indirect digitalization may lead to the same accuracy of three-dimensional datasets and the resulting dental restorations. Thus there are two approaches to evaluate the two different digitalization methods.
The first approach means to evaluate the accuracy of the three-dimensional datasets of the virtual casts, by comparing them to a highly accurate three-dimensional reference dataset, using special inspection software. By superimposition of the datasets, the three-dimensional spatial differences between reference- and test-datasets can be analyzed . The parameters “trueness” and “precision” can be applied to describe the accuracy of three-dimensional digital models . “Trueness” refers to the closeness of agreement between the arithmetic mean of a large number of test results and the true or accepted reference value. The term “precision” refers to the closeness of agreement between test results, and is normally expressed as standard deviation. However, these approaches evaluate only the accuracy of the digitalization itself, but give no information about the total fabrication process and the final outcome.
The second possibility is the evaluation of the marginal and internal fit of the resulting dental restorations from both workflows by using the replica-technique . With this method, the complete workflow, including the digitalization, design- and fabrication process is analyzed. For the longevity of fixed prosthodontics especially the marginal fit is decisive. Marginal gaps of 120 μm or less seem to be clinically acceptable .
With modern CAD/CAM systems a wide range of industrially prefabricated materials, like base metal alloys, zirconia or polymers can be processed. Nevertheless, currently insufficient knowledge is available about the accuracy and synchronization of the different workflow steps after indirect and direct digitalization and the influence of different materials on the fit of dental restorations.
This in vitro study assesses the fit of CAD/CAM-generated frameworks of 4-unit FDPs after direct and indirect digitalization. Therefore frameworks from base metal alloy and from zirconia, fabricated from the same datasets, were compared to evaluate the influence of different materials and production processes. The null hypothesis was, that the method of digitalization (direct or indirect) and the used material (base metal alloy or zirconia) does not influence the marginal fit of CAD/CAM-fabricated frameworks of 4-unit fixed dental prostheses (FDP). Additionally the present study evaluated the accuracy of the underlying datasets by use of inspection software. The second null hypothesis was that direct and indirect digitalization leads to the same accuracy of the datasets when compared to a highly accurate reference dataset.
Materials and methods
Reference model and reference dataset
In this study a titanium model with a prepared premolar (FDI 14) – and molar (FDI 17) – abutment was used. The template for the titanium model was an upper jaw typodont model (Basic Study Model, KaVo Dental GmbH, Biberach, Germany). The preparation design of the single abutments was a 360° chamfer with a convergence angle of 6°. The titanium master model was produced in a milling center (KaVo, Biberach). The model was digitized by an industrial CT-system (X-ray Detector: Perkin Elmer PE XRD 1620, X-ray Tube: Feinfocus FXE 225.99, Fraunhofer EZRT, Fürth, Germany) using the helical method to obtain a highly precise reference dataset. To assess the voxel size, a control measurement with a calibrated ball bar was conducted. Data were converted into STL (surface tessellation language) format. To optimize the data, special methods for the correction of artifacts were conducted . Irrelevant surface-areas were removed from the dataset (Qualify 10.0, Geomagic, Morrisville, US). The resulting STL-dataset was exported and defined as reference (REF) for the present study.
Fabrication of the frameworks
In total 48 frameworks were fabricated on the basis of two different digitalization techniques ( n = 24 per group): (1) direct digitalization and (2) indirect digitalization. Both groups included frameworks of two different CAD/CAM materials ( n = 12 per group): (1) base metal alloy (coron, Straumann, Basel, Switzerland) and (2) zirconia (zerion, Straumann). Table 1 shows the settings for both CAD/CAM materials, according to the specifications of the manufacturer. The complete CAM process of the 48 frameworks took place in a centralized fabrication center (Straumann CARES CADCAM GmbH, Markkleeberg, Germany). Used was a 3+1 axes milling unit (Wissner Gesellschaft für Maschinenbau mbH, Göttingen, Germany). Zirconia frameworks were sintered in a sintering furnace (Nabertherm GmbH, Lilienthal, Germany) with a total process time of 9.5 h and a temperature maximum of 1450 °C.
|Base metal alloy (coron, Straumann)||Zirconia (zerion, Straumann)|
|Chemical composition||50–70% cobalt, 20–40% chrome, 4–10% tungsten||Yttria stabilized zirconia ceramic (Y-TZP)|
|Elastic modulus||230 GPa||210 GPa|
|CAD-parameters||Cement spacer 30 μm, varnish 60 μm starting 0.5 mm above the margin, radius correction 110%, substructure thickness 0.4 mm||Cement spacer 30 μm, varnish 60 μm starting 1.5 mm above the margin, radius correction 90%, substructure thickness 0.6 mm|
|Batch No.||23022954, /−984, /−961, /−985, /−962, /−499, /−500, /−501, /−502, /−426||45011120, /−142, /−143, 43007149, /−156, /−123, /−184|
The titanium master model was digitized twelve times using the iTero intraoral scanner (Align Technology, Carlstadt, US) without any powder application. The raw data were sent to the manufacturer for downsizing and post processing, according to clinical practice. Then the data were sent back to the dental laboratory of the Department of Prosthodontics (LMU, Munich, Germany), where the design of the 24 frameworks was performed (Straumann CARES Visual design software, Straumann, Basel, Switzerland). From each dataset two frameworks were ordered, resulting in 12 frameworks from base metal alloy (DD-C) and 12 from zirconia (DD-Z). Additionally the 12 underlying datasets were exported as test datasets (DD 1–12) in STL-format.
Twelve conventional monophasic impressions (Impregum Penta; 3M ESPE, Seefeld, Germany) were taken from the reference model the following manufacturer’s recommendations. To ensure homogeneity and a minimum layer of 3 mm in each direction, the impression material was mixed by an automix system (Pentamix 2; 3M ESPE) and custom impression trays (Palatray XL; Heraeus Kulzer, Hanau, Germany) were used. After the material was set, the impressions were removed carefully from the titanium model and visually evaluated. Obvious defaults (like bubbles, separation from the tray, etc. ) engendered an iteration of the impression. In accordance to clinical routine disinfection for 10 min (Impresept, 3M ESPE) was performed. Manufacturing of gypsum-casts (Octa-Scan; Heraeus Kulzer, Hanau, Germany) was performed 24 h after impression taking. The indirect digitalization of the 12 master-casts was performed using an optical scanning device (Straumann CARES Scan CS2) 48 h after cast fabrication. Subsequently the design of the 24 frameworks, 12 for production of frameworks from base metal alloy (ID-C) and 12 for the production from zirconia (ID-Z) was performed (Straumann CARES Visual design software). Additionally the 12 underlying datasets were exported as test datasets (ID 1–12) in STL-format.
Production of the replicas
To analyse the marginal and internal gap of the frameworks, the replica-technique described by Molin and Karlsson was applied, without any prior manual adjustment of the frameworks ( Fig. 1 ). To obtain replicas of the gap between abutments and frameworks, both retainers of the 4-unit FDP were filled with light-body A-silicon (Virtual Light Body, Ivoclar Vivadent, Schaan, Liechtenstein, LOT: NL4150). Thereafter, the substructure was placed onto the abutment teeth of the titanium master model and axially loaded with finger pressure. After setting of the light-body material, the frameworks were removed with caution to ensure that the thin silicon layer remained on the master model. Subsequently a heavy-body material (Virtual Putty, Ivoclar Vivadent, LOT: 4041) was circumferentially put on the light-body silicon. After setting of the heavy-body material the thin replica-layer was removed from the master model. The replica specimens were sectioned perpendicularly with a scalpel in oro-vestibular and mesio-distal direction, resulting in four cross-sections and therefore eight measurement locations per substructure ( Table 2 ).
Microscopic acquisition of the replicas
The replicas were analyzed using a reflected light microscope (Axioscope 2; Zeiss, Oberkochen, Germany) at 50× magnification (ocular 10×/23, lens 5×/0.13). Every sectioned specimen was captured by a digital single lens reflex camera (Nikon D100; Tokio, Japan) through the microscope and the pictures were directly transferred to a computer. To acquire a complete cross-section eight to ten digital images were necessary ( Fig. 2 A ). The subsequent merging of the single images to obtain one complete image of the whole section was performed by Adobe Photoshop CS software (Adobe Systems, Inc., San Jose, CA, USA).
Measurement of the marginal and internal fit
Special software (Optimas 6.5, Media Cybernetics, Silver Spring, MD, USA) was used to quantify the marginal and internal fit of the frameworks. The microscopic images of the replicas were imported and a series of points on the outer and inner border of the light-body silicone was marked. The software automatically connected these single points to two continuous lines. From each line, perpendiculars were constructed by the software and the length of every perpendicular was determined beginning at the starting point until crossing with the opposing line. In every plane about 5000 perpendiculars were measured ( Fig. 2 B–D).
Furthermore the section-planes were divided into four measurement areas: 1. Marginal Opening (MO), 2. Chamfer Area (CA), 3. Axial Wall (AW) and 4. Occlusal Plateau (OP). In total, 1536 measurements were performed on the 48 frameworks. Means for every measurement location and area were calculated for the groups of ID and DD.
Determination of the accuracy
The spatial differences of datasets of direct (DD, n = 12) and indirect (ID, n = 12) digitalization in relation to the REF-dataset were analyzed using Qualify 12.1.2 inspection software (Geomagic). Therefore, all datasets were reduced to the field of interest by elimination of all artifacts and areas below the preparation margin. Alignment was performed on the basis of a best-fit-algorithm with automatic outlier elimination. After the alignment of each test- with the REF-dataset, a three-dimensional analysis of the spatial divergences in x , y and z -axis was performed by the inspection software and visualized in a color-coded illustration. The mean positive and negative deviations between each test- and the REF-dataset, as well as the corresponding standard deviation were calculated using the inspection software. Additionally, absolute values for each measurement point of the test datasets were calculated based on the positive and negative deviations. Subsequently for each group (DD and ID) one “overall mean” value was calculated for positive, negative and absolute values. These values were classified as indicators for the “trueness” whereas for determination of the “precision” the standard deviation was used .
Normality of data distribution was tested using Kolmogorov–Smirnov and Shapiro–Wilk tests. Unpaired two sample Student’s t -test (for mean) combined with Levene-test (homogeneity of variances) based on the assumption of normal data distribution were used to analyse the marginal fit of both digitalization methods, to analyse differences of the materials tested within one digitalization method and for analysis of the accuracy of DD and ID in relation to REF. For the data analysis the Statistical Package for the Social Science Version 20 (SPSS Inc., Chicago, US) was used. The level of significance was set at 5% ( p < 0.05).