Polymerization composite shrinkage evaluation with 3D deformation analysis from μCT images

Abstract

Objectives

The aim of this study was to develop a method to experimentally determine and visualize the direction and amount of polymerization shrinkage.

Methods

We modified a composite to include 1.5 wt% traceable glass beads. A cylindrical cavity (6 mm diameter, 3 mm height) was restored with this traceable composite, with and without dentin adhesive, and digitized with high-resolution micro-computed tomography (μCT). Image segmentation was performed to extract the glass beads from the acquired 3D μCT images (uncured and cured). Afterwards, each glass bead was subjected to local rigid registration. The resulting displacement vectors were used to examine and calculate the changes.

Results

In unbonded restorations, the displacement vectors were oriented inwards to the center of mass, although not perfectly. Bonded restorations exhibited two contraction patterns: either toward one side of the cavity or toward the top-surface of the restoration. The displacement vector length values (mean/SD) for the bonded group (46.8 μm/10.0 μm) was significantly higher ( p < 0.01) than unbonded group (31.3 μm/8.5 μm), and the histogram curve was flatter (skew/kurtosis: 0.10/−0.56) as compared to the unbonded group (skew/kurtosis: 0.03/−0.26).

Significance

The proposed method can visualize real 3D displacement vectors generated by polymerization shrinkage. The bonding quality and cavity geometry are critical for the direction of polymerization contraction. This method has the potential to validate current models concerning the amount and orientation of shrinkage vectors.

Introduction

With currently available composite materials, it is inevitable that some (1.5–6 vol.%) shrinkage will occur during polymerization . Even newly developed ring-opening monomers like the siloranes still have a volumetric shrinkage of about 1 vol.% . Volume changes upon curing cause either marginal gaps or, in the case of enduring adhesion, stress within the tooth and the restorative material.

Throughout the years, numerous attempts have been made to analyze polymerization shrinkage and subsequent shrinkage stress. To date, only indirect methods that usually evaluate gap formation are used to compare the ability of different composite–dentin bonding agent (DBA) systems to form a tight marginal seal. However, these indirect methods have some inherent limitations. Marginal gap analysis with dye penetration seems to be a valid approach , although these studies are usually time-consuming and require the destruction of the samples by cutting them in order to observe deeper areas of the cavity . Sometimes, it is difficult to determine whether the dye penetrates into a gap or stains the hydrophilic DBA itself. Other methods such as SEM analysis can evaluate only superficial aspects. If native teeth are used, gaps due to drying artifacts cannot be differentiated from gaps due to shrinkage. If replicas are made, then the quality of the replica limits the discriminative power of the SEM evaluation. Finite element analysis (FEA) is an alternative approach to model the orientation of polymerization shrinkage by simulating the clinical situation, but it is limited by some assumptions required for computer modeling . Thus, there is no perfect measurement that matches the clinical situation since most setups are idealizations and simplifications of actual conditions.

The orientation of polymerization shrinkage is fundamental for predicting marginal adaptation and stress distribution . The extent and direction in which this shrinkage occurs can be portrayed as shrinkage vectors . In dental literature, controversial hypotheses state that light-cured resin composites shrink toward the light source or, similar to self-cured resin composites, shrink towards the center of mass . Based on these assumptions, different techniques of resin composite application were proposed in order to improve marginal integrity and reduce contraction stress. Examples include light-reflecting wedges to improve the proximal marginal adaptation of Class II restorations , the multiple increment technique , modulation of the light intensity and the use of low-modulus intermediate layers . However, there is little evidence regarding the direction of the polymerization shrinkage vectors of light-initiated resin composites.

X-ray micro-computed tomography (μCT) has been recently used to analyze the interface of the dentin-adhesive-composite and to examine the 3D marginal adaptation in light-cured resin composite restorations . The availability of high-resolution μCT now makes it possible to obtain actual 3D information from the cavity during polymerization.

The aim of this study was (i) to develop an experimental method in order to make the direction and amount of polymerization shrinkage vectors visible and (ii) to gain more insight into the consequences of curing contraction within the tooth cavity via 3D deformation analysis. Our hypothesis was that the shrinkage vector could be visualized by registration of μCT images recorded before and after curing.

Materials and methods

Specimen preparation and experiment design

The dimethacrylate-based flowable resin composite (Tetric ® EvoFlow, Ivoclar, Vivadent AG, Schann/Liechtenstein, Switzerland) was selected to obtain shrinkage values that could be clearly identified with the given μCT resolution. In order to visualize the movement of the material, radiolucent spherical glass fillers with the mean particle diameter of 40–90 μm (Sigmund Linder GmbH, Warmensteinach, Germany) were chosen as traceable markers. The glass beads were silanized to ensure a durable connection to the composite. The total amount of glass beads added to the composite was approximately 1.5 wt%. Sixteen intact molar teeth were selected and their cusp tips were removed to obtain a flat surface. The flat surface ensured consistent and unimpeded access for light curing. An occlusal tooth cavity with 3 mm depth and 6 mm diameter was prepared. This cylindrical cavity was selected in order to eliminate as much elastic deformation of the tooth as possible and to comply with one of the models evaluated by Versluis et al. . The teeth were divided into two groups. In the first group, the dentin surface was not pre-treated with a DBA, thus serving as a negative control, while a self-etching DBA was applied to the second group. The tooth restored with the traceable resin composite was covered with a radiolucent, dark cap to avoid hardening of the resin composite during μCT measurements. The restoration was digitized before and after light curing (90° direction, 40 s, 950 mW/cm 2 light intensity, 8 mm light-tip diameter, LED SmartLight ® PS, Dentsply/Caulk, DE, USA). The materials used in this study are described in Table 1 .

Table 1
Composition of the modified dental composite and adhesive used in this study.
Brand name Composition Batch no. Manufacturer
Modified dental composite
Tetric ® EvoFlow (flowable resin) Matrix: dimethacrylates (38 wt%)
Fillers: barium glass, ytterbium trifluoride, highly dispersed silicon dioxide, mixed oxide and copolymer (62 wt%)
Others (<1 wt%)
Particle sizes of the inorganic fillers: 40–3000 nm
LOT: J21884 Ivoclar Vivadent AG, Bendererstrasse 2, FL-9494
Schaan Principality of Liechtenstein
Glass beads (as traceable markers) SiO 2 (72.50 wt%), Na 2 O (13.00 wt%), CaO (9.06 wt%), MgO (4.22 wt%), Al 2 O 3 (0.58 wt%)
Diameter: 90% of 40–90 μm, 8% more than 100 μm
Art. no.: 5211 Sigmund Linder GmbH, Warmensteinach, Germany
Adhesive
Adper™ Prompt L-Pop (self-etch adhesives) Liquid 1 (red blister): methacrylated phosphoric esters, bis-GMA, initiators based on camphorquinone, stabilizers
Liquid 2 (yellow blister): water, 2-hydroxyethyl methacrylate (HEMA), polyalkenoic acid, stabilizers
LOT: D2691 319369 3M, ESPE
St. Paul, MN

X-ray micro-computed tomography measurement

A high-resolution μCT apparatus (μCT 40, Scanco Medical AG, Basserdorf, Switzerland) was used to evaluate the movement of material due to polymerization shrinkage of the light-curing composite. The μCT utilized an acceleration voltage of 70 kVp and a cathode current of 114 μA. The samples were scanned with 8 μm resolution using an integration time of 300 ms and were never removed from the μCT attachment. The 3D data before and after curing were subjected to image segmentation and registration. The flow chart in Fig. 1 explains how the digital 3D data were obtained for image registration. The purpose of the overall process was to determine the center of the same spherical particle before and after curing and to describe the movement of the sphere as a displacement vector. The start point of the displacement vector was identical to the center of the sphere before curing, while the end point of the displacement vector was identical to the center of the same sphere after curing.

Fig. 1
Flow chart describing how to obtain digital 3D data sets before and after polymerization. The numbers in brackets indicate the sequence of the performed steps.

Data processing

The deformation vector field was obtained in a three-step approach. In the first step, the volume of the restoration was identified in the μCT data set. The subsequent evaluation was limited to this subimage only, which was called volume of interest (VOI). The second step identified the traceable markers and labeled each marker individually in the VOI of the uncured composite. The last step determined the displacement vectors of the individual markers. All displacement vectors were collectively defined as the displacement vector field.

Subimage selection

The selection of the restoration volume was performed interactively using the software InsightSNAP ( www.itksnap.org ). Restoration volumes selected from images acquired before and after curing were stored and used for subsequent image processing steps.

Sphere segmentation

This procedure identified and separated the glass spheres from the remainder of the restoration in images taken before curing. As the gray value of the radiolucent glass spheres was smaller than that of the composite material, a threshold was applied to segment sphere candidates. Unique labels were assigned to each connected region that had been segmented. In order to restrict registration to spherical structures, the inertia tensor and its eigenvalues were determined for each labeled component. For ideal spheres, identical eigenvalues were expected; thus regions for which the eigenvalues differed significantly were considered non-spherical and excluded from the remaining image processing steps.

Registration of individual spheres

Since the hardness of the glass beads significantly exceeded that of the composite, deformation of spheres during polymerization was not expected, and a local rigid registration (block matching) was performed to determine the displacement of segmented spheres during the polymerization process ( Fig. 2 ). The procedure of block matching was described in detail by Rösch et al. .

Fig. 2
Workflow of block matching to determine the deformation vectors. (A) The VOI was selected from the 3D data stack of the μCT image. (B) The glass beads were segmented using a gray level threshold followed by the exclusion of non-spherical objects, and each individual sphere was labeled. The labels were color-coded for visual control. (C) Segmented glass beads superimposed onto the corresponding gray value image after polymerization before and (D) after the block matching registration.

Briefly, the overall procedure was as follows: the largest sphere of the uncured situation was identified in the sorted list of spheres. Then, a spherical mask containing both the glass bead and some surrounding material in the uncured situation was determined. At the beginning of block matching, the coordinates of the sphere center were transferred to the cured situation, and the gray value cross-correlation was optimized iteratively with respect to sphere displacement between the cured and uncured situations. The calculation of cross-correlation was restricted to the area corresponding to the spherical mask. One by one, the displacement vectors of each identifiable sphere were determined and stored. The combination of all individual displacement vectors was described as the displacement vector field.

Deformation field visualization

The deformation field was visualized using VTK ( www.vtk.org ) version 5.0.4. The individual translation vectors were visualized starting from the midpoint of the selected sphere. For the printed figures, the length of the vectors was scaled by a factor of 3 in order to enhance visibility.

For quantitative analysis of the overall amount of deformation independent of the direction of the deformation, the vector length values were summarized as a histogram. The skewness and kurtosis were calculated to characterize the distribution of the deformation vector lengths. In addition, the mean and standard deviation (SD) of each sample was subjected to independent Student’s t -test ( α = 0.05) to compare the deformation change between unbonded and bonded group ( N = 8).

Materials and methods

Specimen preparation and experiment design

The dimethacrylate-based flowable resin composite (Tetric ® EvoFlow, Ivoclar, Vivadent AG, Schann/Liechtenstein, Switzerland) was selected to obtain shrinkage values that could be clearly identified with the given μCT resolution. In order to visualize the movement of the material, radiolucent spherical glass fillers with the mean particle diameter of 40–90 μm (Sigmund Linder GmbH, Warmensteinach, Germany) were chosen as traceable markers. The glass beads were silanized to ensure a durable connection to the composite. The total amount of glass beads added to the composite was approximately 1.5 wt%. Sixteen intact molar teeth were selected and their cusp tips were removed to obtain a flat surface. The flat surface ensured consistent and unimpeded access for light curing. An occlusal tooth cavity with 3 mm depth and 6 mm diameter was prepared. This cylindrical cavity was selected in order to eliminate as much elastic deformation of the tooth as possible and to comply with one of the models evaluated by Versluis et al. . The teeth were divided into two groups. In the first group, the dentin surface was not pre-treated with a DBA, thus serving as a negative control, while a self-etching DBA was applied to the second group. The tooth restored with the traceable resin composite was covered with a radiolucent, dark cap to avoid hardening of the resin composite during μCT measurements. The restoration was digitized before and after light curing (90° direction, 40 s, 950 mW/cm 2 light intensity, 8 mm light-tip diameter, LED SmartLight ® PS, Dentsply/Caulk, DE, USA). The materials used in this study are described in Table 1 .

Table 1
Composition of the modified dental composite and adhesive used in this study.
Brand name Composition Batch no. Manufacturer
Modified dental composite
Tetric ® EvoFlow (flowable resin) Matrix: dimethacrylates (38 wt%)
Fillers: barium glass, ytterbium trifluoride, highly dispersed silicon dioxide, mixed oxide and copolymer (62 wt%)
Others (<1 wt%)
Particle sizes of the inorganic fillers: 40–3000 nm
LOT: J21884 Ivoclar Vivadent AG, Bendererstrasse 2, FL-9494
Schaan Principality of Liechtenstein
Glass beads (as traceable markers) SiO 2 (72.50 wt%), Na 2 O (13.00 wt%), CaO (9.06 wt%), MgO (4.22 wt%), Al 2 O 3 (0.58 wt%)
Diameter: 90% of 40–90 μm, 8% more than 100 μm
Art. no.: 5211 Sigmund Linder GmbH, Warmensteinach, Germany
Adhesive
Adper™ Prompt L-Pop (self-etch adhesives) Liquid 1 (red blister): methacrylated phosphoric esters, bis-GMA, initiators based on camphorquinone, stabilizers
Liquid 2 (yellow blister): water, 2-hydroxyethyl methacrylate (HEMA), polyalkenoic acid, stabilizers
LOT: D2691 319369 3M, ESPE
St. Paul, MN

X-ray micro-computed tomography measurement

A high-resolution μCT apparatus (μCT 40, Scanco Medical AG, Basserdorf, Switzerland) was used to evaluate the movement of material due to polymerization shrinkage of the light-curing composite. The μCT utilized an acceleration voltage of 70 kVp and a cathode current of 114 μA. The samples were scanned with 8 μm resolution using an integration time of 300 ms and were never removed from the μCT attachment. The 3D data before and after curing were subjected to image segmentation and registration. The flow chart in Fig. 1 explains how the digital 3D data were obtained for image registration. The purpose of the overall process was to determine the center of the same spherical particle before and after curing and to describe the movement of the sphere as a displacement vector. The start point of the displacement vector was identical to the center of the sphere before curing, while the end point of the displacement vector was identical to the center of the same sphere after curing.

Fig. 1
Flow chart describing how to obtain digital 3D data sets before and after polymerization. The numbers in brackets indicate the sequence of the performed steps.

Data processing

The deformation vector field was obtained in a three-step approach. In the first step, the volume of the restoration was identified in the μCT data set. The subsequent evaluation was limited to this subimage only, which was called volume of interest (VOI). The second step identified the traceable markers and labeled each marker individually in the VOI of the uncured composite. The last step determined the displacement vectors of the individual markers. All displacement vectors were collectively defined as the displacement vector field.

Subimage selection

The selection of the restoration volume was performed interactively using the software InsightSNAP ( www.itksnap.org ). Restoration volumes selected from images acquired before and after curing were stored and used for subsequent image processing steps.

Sphere segmentation

This procedure identified and separated the glass spheres from the remainder of the restoration in images taken before curing. As the gray value of the radiolucent glass spheres was smaller than that of the composite material, a threshold was applied to segment sphere candidates. Unique labels were assigned to each connected region that had been segmented. In order to restrict registration to spherical structures, the inertia tensor and its eigenvalues were determined for each labeled component. For ideal spheres, identical eigenvalues were expected; thus regions for which the eigenvalues differed significantly were considered non-spherical and excluded from the remaining image processing steps.

Registration of individual spheres

Since the hardness of the glass beads significantly exceeded that of the composite, deformation of spheres during polymerization was not expected, and a local rigid registration (block matching) was performed to determine the displacement of segmented spheres during the polymerization process ( Fig. 2 ). The procedure of block matching was described in detail by Rösch et al. .

Nov 30, 2017 | Posted by in Dental Materials | Comments Off on Polymerization composite shrinkage evaluation with 3D deformation analysis from μCT images
Premium Wordpress Themes by UFO Themes