Accuracy of models of partially edentulous arches obtained by three-dimensional printing: An in vitro study

Abstract

Aim:

The aim of this study was to evaluate the accuracy of models of partially edentulous arches obtained by three-dimensional (3D) printing.

Settings and Design:

This was an in vitro study.

Materials and Methods:

Fifteen partially edentulous models were evaluated, using two methods of measuring dimensions: virtual, using the Standard Tessellation Language files of the models and software (control group), and physical, through printing the models and digital caliper (test group). For both methods, measurements were made regarding the dimensions of the teeth (width and length – buccal/lingual or palatal/occlusal) and distances between the teeth.

Statistical Analysis Used:

For the variable of linear measurements (width and length) and distances between teeth of the same hemiarch, the Wilcoxon test was used, while for the variable between opposite hemiarches, the paired t-test was used.

Results:

In the evaluation of the linear measurements, a significant difference was observed only when the width of the molar tooth was analyzed (P = 0.014). When the buccal length was measured, all teeth had linear measurements provided by the virtual method that was lower than the physical (P = 0.000), as well as the lingual/palatal length in incisors (P = 0.003) and molars (P = 0.009) and in total (P = 0.001). As for the analyses between teeth, no difference was identified between the measurements provided by the virtual method compared to the physical one.

Conclusions:

The 3D printer used to print partially edentulous models provided linear distortions in the teeth but without changes in the distances between teeth of the same hemiarch and between teeth of opposite hemiarches.

INTRODUCTION

The use of the digital workflow in dentistry has gained increased acceptance in clinical practice. Performing intraoral scanning directly on the patient or scanning conventional plaster models allowed the digitization of the dental situation.[1] With this technology, the scanned dental arches can be stored as a three-dimensional (3D) surface file, in Standard Tessellation Language (STL).[2] Thus, it is possible to realize a fully digital workflow, incorporating computer-aided design, and computer-aided manufacturing (CAD/CAM).[3]

In this context, rapid prototyping, also known as 3D printing or manufacturing by addition, is often used in areas such as dental prostheses, oral and maxillofacial surgery, oral implant dentistry, orthodontics, endodontics, and periodontics. In oral rehabilitation, CAD/CAM systems and rapid prototyping have been used for years to manufacture inlays, onlays, crowns, fixed partial dentures, implant prostheses, and maxillofacial prostheses.[4] Recently, they have also been used in stages for the manufacture of removable prostheses, including removable partial dentures (RPDs). In these cases, the implementation of the digital workflow in the construction of the structure of the RPDs took place mainly in the scanning and planning of the prosthesis,[5] but it still has limitations related to the digital processing of the prosthesis base. Therefore, a partially digital workflow is usually used, with the need to print the working model for processing the prosthesis base.

However, the accuracy of printed models depends on factors, such as the acquisition and processing of images of the hard and soft tissues of the mouth and processes involved in the manufacture and postprocessing of these materials. Furthermore, models obtained by various printing techniques are affected by polymerization shrinkage.[1,6] For this reason, recent studies have evaluated the accuracy of full-arch dental models fabricated using different 3D printing technologies. A systematic review by Etemad-Shahidi et al.[6] indicated that the accuracy of full-arch dental models obtained by 3D printing ranged between <100 and >500 μm, with most of the evaluated models being clinically acceptable. Despite this, the authors emphasized that models considered clinically acceptable for orthodontic purposes, for example, may not be acceptable for dental prosthetic work or other procedures requiring high accuracy.

However, no studies were found in the literature that specifically evaluated the accuracy of printed models of partially edentulous arches for the purpose of making removable partial dentures with a partially digital flow, which requires accurate models for the laboratory processing steps of the prosthesis. In this sense, the present study aimed to evaluate the accuracy of a 3D printer for printing partially edentulous models. The null hypothesis (H0) consists of the lack of accuracy of the 3D printer for the printing of partially edentulous models. The alternative hypothesis (H1) indicates that the printing process leads to linear distortions in the partially edentulous model.

MATERIALS AND METHODS

This is a cross-sectional study, which was based on the guidelines of STROBE (The Strengthening the Reporting of Observational Studies in Epidemiology).[7] The digital archive bank of STL files from a Dental Prosthesis Laboratory, with the patients’ names blinded, was evaluated in the search for files of partially edentulous arches that presented at least 1 incisor, 1 canine, 1 premolar, and 1 molar. Files that showed significant coronal destruction and whose scanning was not performed satisfactorily were excluded, to present a fully complete model. This study was submitted to the Research Ethics Committee (CEP) of the Federal University of Rio Grande do Norte (UFRN), approved with protocol opinion number 4.745.226.

Materials and models’ fabrication

A total of 15 files were selected, included in the sample, and submitted to two methods: (V) virtual, control group and (P) physical, test group of this study. The physical models were made from a printer an Anycubic Photon Mono SE (Anycubic, Shenzhen, Guangdong, China) and printing resin for models (PrintaX, OdontoMega, Ribeirão Preto, São Paulo, Brazil), using LCD technology, with the following parameters: layer height (0.05), amount of fixation layer (6), base fixation/bottom exposure time (25–40 s), normal exposure time/exposure time (2–3 s), and with orientation of horizontal printing, being evaluated 3 months ± 15 days after printing; while the virtual ones were evaluated by the GOM Inspect 2019 software (GOM GmbH – Schmitzstraße, Braunschwei, Germany).

Analysis of the accuracy

Initially, the measurements were standardized, considering the tooth and the distances between teeth, for both methods (V and P). For the tooth, measurements of width at which the anterior teeth were measured from the mesioincisal to distoincisal angle and the posterior teeth, from mesioocclusal to distoocclusal angle. Visualized through lateral [Figure 1a] and occlusal view [Figure 1b]. As for the length, the incisors were measured from the cervical to the incisal edge; the canines, from the cervical to the point of union of the two buccal edges; the premolars, from the highest point of the buccal cusp to the highest point of the lingual or palatal cusp; on molars, from the highest point of the distobuccal cusp to the highest point of the mesiolingual or mesiopalatal cusp [Figure 1c-g].

Figure 1.

Measurements performed by the virtual and physical method. (a) Virtual model overview– lateral view, (b) Virtual model overview– occlusal view, (c) Measurement per tooth (incisor) in width (yellow line) and length (red line), (d) Measurement per tooth (canine) in width (yellow line) and length (red line), (e) Measurement per tooth (premolar and molar) in width (red line), (f) Measurement per tooth (premolar) in length (red line), (g) Measurement per tooth (molar) in length (red line), (h) Measurement of distance between teeth of the same hemiarch, (i) Measurement of distances between teeth of opposite hemiarches

For the distances between teeth, the distances between teeth of the same hemiarch and the distances between teeth of opposite hemiarches were measured, considering as a reference point the apex of the highest cusp (molars) and the apex of the palatine or lingual cusp (premolar). These distances between teeth still had the coding of the teeth involved, being: 1 (incisor), 2 (canine), 3 (premolar), and 4 (molar), in the distances between teeth of the same hemiarch: 1-3, 1-4, 2-3, 2-4, 3-4, and 4-4 [Figure 1h] and opposite hemiarches: 1-1, 1-3, 1-4, 2-2, 2-3, 2-4, 3-3, 3-4, and 4-4 [Figure 1i].

Then, an examiner (L. M. C. M. D.) was subjected to calibration. The examiner applied the virtual measurement method to 10 models, using the “straight line” measurement tool of the Tool of the GOM Inspect 2019 software (GOM GmbH – Schmitzstraße, Braunschwei, Germany), and after 15 days, the physical method, with the aid of a digital caliper, as described by Aly and Mohsen,[8] with a precision of 1 μm, considering the measurements described above. The agreement between the methods was obtained using the Interclass Correlation Coefficient (ICC), the ICC mean and the standard deviation at a significance level of 95%, considering as no agreement (0), poor (0.1–0.20), weak (0.21–0.40), moderate (0.41–0.60), substantial (0.61–0.80), almost perfect (0.81–0.99), or perfect (1).[9] The concordance between the two methods was evaluated, obtaining a Kappa value of 0.864 (distance between teeth of the same hemiarch), 0.855 (distance between teeth of opposite hemiarches), physical method (width: 0.975; buccal length: 0.961; occlusal length: 0.983), and virtual method (width: 0.980; buccal length: 0.987; occlusal length: 0.988).[10]

With the examiner precalibrated, the data collection of the variables of this study began, using the same methods and means reported for calibration. The models were randomly randomized considering the method (V and P). Then, the collection was submitted in two different time periods: (T1) 15 models and (T2), after 15 days (wash-out period), the same models were evaluated to blind the examiner so that he would not remember the answers previously collected.

Statistical analyses

The statistical software IBM SPSS (Statistics V22.0; IBM Corp - Armonk, New York, United States) was used for data tabulation and analysis. The nonnormality (P < 0.05) was observed when the linear measures (per tooth) and distances between teeth of the same hemiarch of the models were measured by the physical and virtual methods. Therefore, the Wilcoxon test (P < 0.05) was used to evaluate the effect of the impression on the accuracy of the linear measurements provided by the two tested methods, both when the tooth was considered the subject and the distances between teeth of the same hemiarch. When evaluating the distances between teeth of opposite hemiarches, the data showed normal distribution (P > 0.05), allowing the paired t-test to be performed, with a statistical significance of P < 0.05.

RESULTS

Fifteen partially edentulous models were evaluated, consisting of 60 teeth, 15 of which were incisors, canines, premolars, and molars (each). Measurements consisted of 63 distances between teeth of opposite hemiarches, distributed in 9 possibilities of crossing between the evaluated hemiarchs, and 26 distances between teeth of the same hemiarch in 6 measurement possibilities.

In the evaluation of the linear measurements, per tooth, between the physical and virtual methods, a significant difference was only observed when the width of the molar tooth was analyzed (P = 0.014). When the buccal length was measured, all teeth had linear measurements provided by the virtual method that was lower than the physical ones, impacting the total assessment, which disregards each tooth individually (P = 0.000). This was also observed when measuring the lingual/palatal length in incisors (P = 0.003) and molars (P = 0.009) and in total (P = 0.001) [Table 1]. As for the analysis between teeth of opposite hemiarches [Table 2] and of the same hemiarch [Table 3], no difference was identified between the measurements provided by the virtual method compared to the physical method.

Table 1.

Linear measurements of the partially edentulous models by the physical and virtual method per tooth and the total

Measurements	n	Method/tooth
		Incisor, med (Q25–75)			P	Canine, med (Q25–75)				P
		Physical method		Virtual method		Physical method		Virtual method
Width	15	5.80 (5.06–7.55)		5.82 (5.23–7.69)	0.638		6.69 (6.38–6.98)	6.82 (6.49–7.40)		0.320
Length
Vestibular	15	7.98 (7.31–9.70)		8.55 (7.68–9.66)	0.004*		9.24 (8.41–9.98)	9.55 (8.52–10.39)		0.002*
Lingual/palatine	15	8.84 (8.10–9.64)		9.11 (8.23–9.79)	0.003*		9.03 (8.35–9.81)	8.67 (8.25–9.74)		0.865
Occlusal	15	999 (999–999)		999.00 (999.00–999.00)	1.000		999.00 (999.00–999.00)	999.00 (999.00–999.00)		1.000
Measurements	Method/tooth
	Premolar, med (Q25–75)		P	Molar, med (Q25–75)		P	Total, med (Q25–75)		n	P
	Physical method	Virtual method		Physical method	Virtual method		Physical method	Virtual method
Width	6.77 (6.41–7.46)	6.65 (6.44–6.97)	0.211	10.31 (9.65–11.24)	10.26 (8.86–10.90)	0.014*	6.95 (6.26–9.21)	6.96 (6.29–8.75)	60	0.132
Length
Vestibular	7.77 (7.09–8.36)	8.04 (7.67–8.68)	0.005*	6.95 (5.70–7.78)	7.02 (5.88–8.01)	0.020*	7.94 (7.11–9.13)	8.31 (7.47–9.49)	60	0.000*
Lingual/palatine	4.74 (4.08–5.47)	4.82 (4.11–6.74)	0.147	5.04 (4.50–6.23)	6.76 (4.68–7.38)	0.009*	7.0 (4.93–8.98)	7.75 (5.43–9.17)	60	0.001*
Occlusal	5.02 (4.39–6.10)	4.82 (4.58–6.15)	0.551	7.73 (7.26–9.16)	8.02 (6.68–9.11)	0.532	504.82 (6.57–999.00)	504.90 (6.52–999.00)	60	0.877

*Statistical significance. Med (Q25–75): Median (quartile 25–75), P: Wilcoxon test

Table 2.

Distances between teeth of opposite hemiarches in partially edentulous models measured by physical and virtual method

Distances between opposite hemiarches	Measurement method, average (SD)
	n	Physical	Virtual	P
1–1	1	17.44 (0)	17.72 (0)	-
1–3	2	33.22 (0.32)	32.88 (1.32)	0.718
1–4	7	49.67 (7.47)	49.11 (7.23)	0.656
2–2	1	34.21 (0)	34.07 (0)	-
2–3	2	43.22 (0.59)	42.94 (0.31)	0.405
2–4	6	48.61 (10.34)	48.77 (10.30)	0.533
3–3	9	41.48 (8.30)	40.57 (7.01)	0.107
3–4	19	49.84 (7.28)	50.70 (6.30)	0.502
4–4	16	54.18 (5.25)	54.20 (5.11)	0.957
Total	63	48.11 (9.40)	48.18 (9.18)	0.867

1: Incisor, 2: Canine, 3: Premolar, 4: Molar, SD: Standard deviation, P: Paired T-test

Table 3.

Distances between teeth of the same hemiarch in partially edentulous models measured by physical and digital method

Distances between the same hemiarch	Measurement method, med (Q25–75)
	n	Physical	Virtual	P
1–3	1	-	-	-
1–4	2	25.90 (16.75–22.09)	32.21 (17.29–31.02)	0.180
2–3	2	16.95 (12.58–12.84)	17.04 (12.75–12.81)	0.655
2–4	4	37.78 (21.89–44.40)	38.85 (22.06–44.82)	0.465
3–4	15	21.99 (19.83–23.35)	20.10 (16.30–23.20)	0.221
4–4	2	22.47 (16.31–17.40)	19.62 (14.01–15.42)	0.180
Total	26	22.02 (17.87–27.54)	20.33 (17.06–23.96)	0.459

1: Incisor, 2: Canine, 3: Premolar, 4: Molar, P: Wilcoxon test, Med (Q25–75): median (quartile 25–75)

DISCUSSION

The null hypothesis (H0), which consists of the lack of accuracy of the 3D printer for the materialization of partially edentulous models, and the alternative hypothesis (H1), which indicates that the printing process leads to linear distortions in the partially edentulous model, were fully accepted. In the present study, it was observed that the linear measurements of molar teeth width and buccal and/or lingual/palatal length of all teeth underwent statistically significant changes. However, no statistically significant difference was identified in the evaluations between teeth, either from the same hemiarch or between opposite hemiarches.

In this context, the accuracy of models for making partially removable prostheses is important, since the printed models can be used to design and/or plan the metallic structure. This model is critical to the adaptation to the patient’s oral tissues to avoid tooth movement and discomfort, which may result in the patient not using this type of prosthesis.[11] The correct functioning depends on an acceptable adaptation of the RPDs.[12] Changes to the printed model, therefore, may compromise the accuracy of these steps.

As in previous studies, the virtual models in STL were used as a control group, while models made by 3D printing were used as a test group.[13,14] In the present study, linear measurements of the width and length of the groups of teeth evaluated were performed due to the importance of these measures, especially for the planning stage of the components of removable partial dentures, which are important for issues such as the guide plane and undercut areas. The increased molar width found in this study may result in a maladaptation of a retention or opposition clip, for example, if made from one of these models, may be larger than necessary. Furthermore, changes in the buccal and lingual/palatal length of the teeth can also interfere with the guide plane of dental elements, staples, and larger connectors. In this study, these lengths in the printed models showed significant distortions, which can influence the adaptation and, consequently, the success of the rehabilitation treatment.

In addition, measurements between teeth were evaluated which are also relevant during the planning of prosthetic components. The measurement between teeth of opposite hemiarches is relevant because it may be related to larger connectors which occupy the entire patient’s arch, just as the distance between teeth of the same hemiarch may be related to components such as the saddle, which may occupy only a hemiarch. In this study, it was observed that the evaluations between teeth did not present statistically significant changes, which may indicate these structures alone would not be influenced by the distortions found in the models. This is advantageous data when referring to the use of printed models for removable partial dentures since the dimensional accuracy of the complete arch is necessary for the adequate seating of the metallic structures that conventionally involve the two hemiarches.

Therefore, several factors may have influenced the distortions found in this study. A systematic review published by Etemad-Shahidi et al.[6] indicated that 3D printing techniques affect the accuracy of printed models, with stereolithography (SLA) and digital light projection (DLP) being the printing technologies with the highest accuracy. The study by Tsolakis et al.[15] showed that DLP 3D printers have greater accuracy for printing dental models than LCD (liquid crystal display). The study by Venezia et al.,[16] in turn, found that SLA technology had less distortion when compared to LCD and DLP. In this study, an LCD printer was used, and this printing technique could be one of the factors that contributed to the observed distortions.

In addition, another factor that can influence the accuracy of printed models is the material of choice. The study by Al-Qarni and Gad[17] evaluated three types of printed resin, and it was observed that the material used influenced the accuracy of 3D printing. In this study, impression resin for models (PrintaX, OdontoMega, Ribeirão Preto, SP, Brazil) was used, which may also have influenced the accuracy of dental models. Furthermore, print orientation can also impact accuracy. Revilla-León et al.[18] evaluated impression of occlusal devices with orientations of 0°, 45°, 70°, and 90°, and it was observed that the impression orientation of 0º presented the highest precision among the tested groups. To avoid this bias, all models in this study were printed with the 0° orientation (horizontal).

Other factors that can also impact the accuracy of printed models are time and storage conditions. Lin et al.[13] indicated that 2-week storage time after impression affected the accuracy of full-arch models, while Yousef et al.[19] observed that fully dentate models stored under light exposure for 3 months were less accurate than those stored in closed boxes for the same period of time. In this sense, measuring the models after 3 months ± 15 days of storage, combined with exposure to light, may have contributed to the distortions found in the present study.

Despite the statistically significant differences observed in this study, it is not possible to state that this will have clinical implications of misfit in prostheses made from these 3D printing models, requiring clinical studies to make RPDs from printed models and evaluate the fit, and clinically rehabilitative success. A systematic review[5] indicated that the digital technique for making structures for removable partial dentures is accurate, with clinically acceptable misfits (<311 μm).

As a limiting aspect of this study, there is the use of only one 3D printer, a printing resin, and a printing orientation of only 0°. Therefore, future studies are needed to assess the impression accuracy of partially edentulous models made from other materials and impression methods, as well as clinical trials to assess whether the distortions found in dental models have a clinical impact.

CONCLUSIONS

Although this in vitro study was carried out with only one 3D printer, one type of resin, and one impression orientation, the results can contribute to the planning of removable partial dentures, since it was verified that the distortion in the size of the teeth can influence the clasps of retention. However, as there was no change between the hemiarches, components such as the saddle and the major connector would not suffer interference. Based on the results found and the limitations of this study, it can be concluded:

1. Partially edentulous models made from 3D printing may present significant linear distortions of the teeth, especially in the width of the molar teeth and buccal and/or lingual/palatal length of the dental elements

2. 3D printing does not seem to significantly change the measurements between teeth of the same hemiarch and of opposite hemiarches of the printed partially edentulous models.

Financial support and sponsorship

CAPES - Coordination for the Improvement of Higher Education Personnel (N°88887.531281/2020-00).

Conflicts of interest

There are no conflicts of interest.