All PhD Theses

L. Camardella

Digital technology in orthodontics: Digital model acquisition, digital planning and 3D printing techniques


A scientific essay in Medical Sciences

DOCTORAL THESIS defended in public on 9th of April 2019


This PhD thesis is based on six studies that investigate main topics in the use of digital technology in orthodontics: the accuracy of digital model acquisition methods, the accuracy of digital planning tools with software programs and the accuracy of printed models using different 3D printing techniques.

Chapter 1 introduces the application of digital models in orthodontics, their different acquisition methods and their respective accuracy according to the literature. Indirect and direct scanning methods are described including their advantages and disadvantages. The use of digital planning in orthodontics with software programs is reported, emphasizing the virtual setup as an indispensable tool to simulate orthodontic treatments, and to provide more details for proper diagnosis and treatment planning of a malocclusion. The use of 3D printing to print dental models, indirect bonding trays or custom brackets is mentioned and the accuracy, advantages and disadvantages of the available 3D printing techniques are explained.

In chapter 2 and 3 two different indirect acquisition methods for digital models were studied, respectively plaster model scanning and PVS impression scanning.

In chapter 2, the accuracy and reliability of measurements performed using two different software programs on digital models acquired from two types of plaster model scanners are compared: a surface laser scanner and a computed tomography (CT) scanner. Two examiners used a sample of 30 pairs of models and performed measurements on plaster models with digital calipers. On digital models the measurements were done with Ortho Analyzer (OA) (3Shape) and Digimodel (DM) (OrthoProof) software programs, creating four different series of digital models: models from the laser scanner measured with OA (Laser OA), models from the laser scanner measured with DM (Laser DM), models from the CT scanner measured with OA (CT OA), and models from the CT scanner measured with DM (CT DM). Forty-two measurements, including tooth diameter, crown height, overjet, overbite, intercanine and intermolar distances and sagittal relationship, were obtained by examiner 1 and 25 selected parameters were measured by examiner 2 to evaluate the reliability of the measurement method. According to the paired t test, examiners 1 and 2 presented excellent interexaminer reliability, with only a few statistically significant differences in the parameters selected, which confirmed the good calibration process between the examiners. Compared with measurements on plaster models, Laser DM models presented three clinically relevant differences: the sum of the 6 upper teeth, the upper intercanine distance, and the right sagittal relationship. For the measurements on Laser OA models, only two parameters presented clinically relevant differences. For the CT OA and CT DM models, only one parameter showed clinically relevant difference. The measurements of dental diameters and dental crown heights on digital models were reliable compared to the measurements on plaster models. The measurements of the upper intercanine distance and the overbite showed the largest differences. These differences could have been caused by misinterpretation of the cuspid landmark due to some attrition on the models and by the subjectivity of the different measurement methods (digital calipers vs. software programs). In the comparisons of only the digital models, the crown height, transversal, and intermaxillary parameters did not show any clinically relevant difference, suggesting that it is easier to mark these points on digital models than on plaster models. Only four parameters in the sum of the mesiodistal diameters presented clinically relevant differences for the four groups of digital models. Finally, it was concluded that digital models generated from plaster models by using laser and CT scanning and measured using two different software programs are accurate and the measurements are reliable. Therefore, both fabrication methods and software programs can be used interchangeably in orthodontics.

Chapter 3 explores another digital model acquisition method: PVS impression scanning. In this study the accuracy and reliability of measurements on digital models obtained by laser scanning impressions 5, 10, and 15 days after they were made, using two different soft putty PVS materials, are evaluated. Thirty volunteers were selected to make impressions of their dentitions with alginate to create a plaster model and with PVS impression material to create a digital model by laser scanning of the impression. According to the manufacturer’s guidelines, the first PVS impression was made with the heavy putty material and then a soft putty material was used to record the anatomic details. The regular-viscosity soft putty was used for the maxillary arch and the light-viscosity soft putty for the mandibular arch to allow evaluation of possible accuracy differences between the 2 materials. The 30 pairs of digital model were divided into 3 groups of 10 pairs each, according to the time interval between taking the impressions and the scanning of the PVS impressions. T5 represented an interval of 5 days; T10 of 10 days; and T15 of 15 days. Three examiners measured 34 distances (tooth diameter, transverse distances (maxillary and mandibular intercanine and intermolar distances), and 2 interarch relationship measurements (overbite, overjet) on the plaster models with digital calipers and repeated these measurements on the digital models using Ortho Analyzer software. All plaster models of the sample were also scanned with the same laser scanner to acquire the respective digital models and enable comparisons by model superimposition of the digital models made from PVS impression scanning. The intra-examiner errors had low values for the measurements on plaster and digital models. The reproducibility analysis showed high ICC values for both plaster model measurements (r = 0.908) and digital models (r = 0.857). According to the paired t test, statistically significant differences were found for some measurements. From the 34 variables evaluated by each examiner, for examiner one, only 2 clinically significant differences in measurements were found; for examiner two 16; and for examiner three, 2 clinically significant different measurements. Therefore, examiners one and three had similar results, but for examiner two (an undergraduate student with less experience in measuring models) more clinically significant differences were found. On average, measurements on digital models with PVS impression scanning showed lower values compared with measurements on plaster models. The overbite was the only parameter with clinically significant differences for all examiners, with lower values for the digital models. Regarding the time interval between PVS impression taking and scanning, the paired t test showed no significant difference in the results among the 3 time periods (5, 10, and 15 days) compared with the plaster model measurements and by model superimposition. The type of soft putty had no influence on the accuracy of the digital models as the mean differences in maxillary arch superimpositions and mandibular arch superimpositions were not statistically significant.

The outcome of this study demonstrates that the acquisition of digital models by laser scanning of PVS impressions scanned within 15 days after impression taking resulted in an accurate digital model, except for the overbite parameter, regardless of the soft putty viscosity type. The accuracy of digital tools of software programs such as virtual setup and customized digital arch forms are discussed respectively in chapters 4 and 5. A virtual setup is a valuable tool for digital planning in orthodontics due to the possibility to simulate an orthodontic treatment. The evaluation of two different setups can be done by digital model superimposition using specific software programs.

Therefore, in chapter 4 the influence of different superimposition methods to compare the accuracy and predictability of diagnostic conventional and virtual setups are evaluated. Ten finished cases were selected to make both a conventional and virtual setup. In these setups second molars were not moved to allow using these molars as a stable reference for surface-based superimposition. The conventional and virtual setups were also compared to the digitized posttreatment models with two superimposition methods: the whole surface best fit (WSBF) method using only the outline of the dentition as a reference, and regional palatal rugae registration best fit (PRBF) method using the medial 2/3 of the third rugae of the palate and a small area dorsal to this rugae as a stable reference. The PRBF superimposition method was used to compare the maxillary virtual setup and the actual posttreatment models. Anterior, intermediate and posterior regions of the dentition were compared. According to the results, conventional and virtual setups were different when superimposed. However, considering the three regions studied, most of the mean differences of RMS were lower than 1.0 mm. Regarding the predictability of conventional and virtual setups, superimposition of the posttreatment models and both setups, using WSBF method, presented comparable differences and these differences were not statistically significant, indicating a similarity of both setup methods using the WSBF superimposition technique. However, there were statistically significant differences between the maxillary posttreatment and virtual setup models using WSBF and PRBF superimposition methods. The PRBF method showed larger differences between the models than the WSBF method. From this study it can be concluded that there are differences between diagnostic conventional and virtual setups and between both setup methods and the final result after orthodontic treatment. In addition, the model superimposition method (WSBF or PRBF) can influence the outcome of the superimposition of the setup models. It is important to establish stable structures as a reference to evaluate the accuracy and predictability of setup models. The arch form of the patient should be preserved or corrected according to the diagnosis and orthodontic treatment planning. A plaster model is the traditionally tool used to choose the best shape of the dental arch with an arch wire template. With the introduction of digital models, the accuracy of the arch form definition using software programs should be tested to evaluate if the arch form for orthodontic patients could be defined on plaster models with arch wire templates or on digital models with dedicated software with similar accuracy.

In chapter 5 we compared the accuracy of preformed wire shape templates on plaster models and customized digital arch form diagrams on digital models. Twenty pairs of dental plaster models were randomly selected and were scanned to create the respective digital models. Three examiners defined the arch form on the mandibular arch of these models by selecting the ideal preformed wire shape template on each plaster model or by making a customized digital arch form on the digital models using a digital arch form customization tool with the Ortho Analyzer software. Each digital arch form diagram created was individually exported as a report generated in PDF format by the software. The best-fit method, selecting the central anterior region as a reference, was used to superimpose both arch forms using Photoshop software. Differences between the superimposed arch forms were evaluated by splitting the diagrams into 6 segments (anterior, premolar and molar regions on the left and right sides). A difference was noticed in the magnification between the arch form size in the PDF report and the actual size of the models. On average, the arch sizes of the samples in the reports were 39.52% larger (range, 39.10% - 40.22%) than the real dimensions of the digital models. This magnification was corrected in each digital arch form to standardize a real proportion of 1:1 to enable a comparison by superimposition onto the arch forms selected on the plaster models. Fortunately, this magnification problem in the report was corrected in an updated version of the Ortho Analyzer software. The thickness of the line in both diagrams was 0.50 mm and the largest differences between the two arch forms in each region were registered after superimposition of the arch forms. An expansion of the customized digital arch form compared with the wire shape diagram for the plaster model was recorded as a positive value, whereas a contraction of the customized digital arch form was recorded as a negative value. Differences of 0 to 1.00 mm were considered clinically insignificant, and those larger than 1.00 mm were considered clinically significant. The results of this study showed that the largest differences between the diagram superimpositions in the anterior and premolar regions were clinically insignificant. The largest differences in the right molar region found by all examiners were clinically significant. When the molar regions on the left and right sides were compared, the largest differences in the first molar region for both sides were not clinically significant. However, for the second molar region, clinical significant differences were found by all examiners on the right side and for the measurements performed by examiner 2 on the left side. In general, the customized digital arch forms when compared with the arch form diagrams selected on the plaster models were expanded. The results of the intraclass correlation coefficients of the measurements between examiners showed a weak correlation in the premolar region and moderate correlations in the anterior and molar regions. These differences can be caused by the subjective method of arch form definition for both plaster and digital models by each examiner, especially in asymmetric arches in the premolar region. However, these differences will not have a clinical impact on the final arch form after orthodontic treatment. Moreover, the use of customized digital arch forms on the digital models enables creation of an arch form that fits more adequate in more areas of the dental arch, fitting especially better in the second molar area compared to the preformed wire shape diagrams selected on the plaster models. It can be concluded that the digital method of arch form definition can substitute or even improve the conventional arch form selection method used for plaster models. The replacement of plaster models by printed dental models is a next step in the transition of traditional into digital orthodontics. Therefore, the accuracy of printed models made with different 3D printing techniques must be tested. Chapters 6 and 7 explore the accuracy of printed models in orthodontics.

The study in chapter 6 compared measurements on plaster models made from alginate impressions and printed models made from digital datasets acquired by intraoral scanning. In this study, 28 volunteers were selected and alginate impressions and intraoral scans were made to make both plaster models and digital models of their dentition. The digital models were printed with a stereolithographic (SLA) 3D printer with a horseshoe-shaped design, as commonly used for clear aligner fabrication. Two calibrated examiners measured distances with a digital caliper (mesiodistal diameter, crown height, upper and lower intercanine and intermolar distances, overjet, overbite and right and left interarch sagittal relationship) on the plaster and printed models. The intra-examiner error comparison showed an excellent accuracy of measurements for both examiners. The result of the paired t test showed no clinically relevant differences in the measurements of teeth dimensions (mesiodistal diameter and crown height) between the plaster and printed models. In addition, the interarch relationship (overjet, overbite, and sagittal relationship) did not reveal any clinically relevant difference. However, the transversal dimensions, especially the upper and lower intermolar distances, presented a clinically relevant reduction in the printed models. A possible explanation of these clinically relevant differences in transversal distances may be by model shrinkage during the post cure phase with ultraviolet light. This post cure procedure is needed for printed models with the SLA technique, as the model is not completely cured during printing. Therefore, it was concluded that the printed models with the SLA technique using a horseshoe-shaped base cannot replace conventional plaster models made from alginate impressions in orthodontics, due to their clinically relevant reduced transversal dimensions in the posterior region.

Puzzled with the results of the study in chapter 6, we performed another study to evaluate the influence of different designs of model bases, using 2 types of 3D printing techniques to print models from the intraoral scans of 10 volunteers (chapter 7). Three types of model base design (regular base, horseshoe-shaped base, and horseshoe-shaped base with a bar connecting the posterior region) were used and these digital models were printed with two different 3D printing techniques (SLA and Polyjet printers). The printed models were compared by measuring transversal parameters (distances between the canines, first and second premolars, and first and second molars) and by model superimposition, after laser scanning of all printed models. The printed models with the regular base were considered the “gold standard” for two comparisons methods: model superimposition and measuring. According to the transversal measurements results, the SLA models with horseshoe-shaped base presented progressive differences with smaller values from the anterior to the posterior regions of the arches, compared to the other base designs. Both bases, regular and horseshoe-shaped with bar, presented similar transversal distances with the SLA printing technique. Polyjet models had greater accuracy of the transversal parameters independent of the model base design used. According to the model superimposition, only the models with horseshoe-shaped base made with the SLA 3D printing method presented statistically significant differences compared to the other base designs. Printed models with the Polyjet technique with different base designs did not show any statistically significant difference when the model superimposition method was used. The disadvantage of the SLA process mentioned in the literature, is the necessity to post cure the printed parts with ultraviolet light to improve the stability of the printed object. Dental models printed with Polyjet printing technique are fully cured during the building process. It can be suggested that the post curing period could affect the accuracy of SLA models without a posterior connection bar or a regular base. The presence of a posterior connection bar in the horseshoe-shaped base models or the use of a regular base design avoided the transversal contraction as seen in the models with a horseshoe-shaped base when the SLA printing technique was used.

In chapter 8 the results of the six studies and the available results found in the literature are discussed. The future of orthodontics is also discussed. It can be assumed that the use of digital technology will have benefits for the orthodontists. Some digital tools that are available nowadays might lead to better results and will increase the predictability of orthodontic treatment. The possibility to combine digital data such as 3D photos, digital models and CBCTs are promising. A fully digital workflow to be used during an orthodontic treatment is described in this thesis. Of course the application of digital technology in orthodontics needs time to be implemented in a clinical routine. A financial investment and an investment in learning time are indispensable for the implementation of intraoral scanners, software programs and 3D printers. A learning curve should also be expected for the orthodontists, orthodontic assistants and dental labs to benefit all the advantages of digital orthodontics. This thesis can be useful for the orthodontists who intend to embrace the digital technology in their clinical practice. In the future the use of digital technology in orthodontics, as presented in this thesis, will certainly increase.