Application of Texture and Fractal Dimension Analysis to Evaluate Subgingival Cement Surfaces in Terms of Biocompatibility

Biocompatibility is defined as “the ability of a biomaterial, prosthesis, or medical device to perform with an appropriate host response in a specific application”. Biocompatibility is especially important for restorative dentists as they use materials that remain in close contact with living tissues for a long time. The research material involves six types of cement used frequently in the subgingival region: Ketac Fil Plus (3M ESPE, Germany), Riva Self Cure (SDI, Australia) (Glass Ionomer Cements), Breeze (Pentron Clinical, USA) (Resin-based Cement), Adhesor Carbofine (Pentron, Czech Republic), Harvard Polycarboxylat Cement (Harvard Dental, Great Britain) (Zinc polycarboxylate types of cement) and Agatos S (Chema-Elektromet, Poland) (Zinc Phosphate Cement). Texture and fractal dimension analysis was applied. An evaluation of cytotoxicity and cell adhesion was carried out. The fractal dimension of Breeze (Pentron Clinical, USA) differed in each of the tested types of cement. Adhesor Carbofine (Pentron, Czech Republic) cytotoxicity was rated 4 on a 0–4 scale. The Ketac Fil Plus (3M ESPE, Germany) and Riva Self Cure (SDI, Australia) cements showed the most favorable conditions for the adhesion of fibroblasts, despite statistically significant differences in the fractal dimension of their surfaces.


Introduction
Biocompatibility is defined as "the ability of a biomaterial, prosthesis, or medical device to perform with an appropriate host response in a specific application" [1]. Because no material can be proven to be 100% biologically safe, technical tests and data are required to determine when the benefits outweigh the risks [2]. There is a sequence of studies assessing the safety of new materials. One of these is in vitro research, followed by investigations conducted on animals and, finally, clinical studies. This makes it possible to evaluate the biocompatibility of new materials and eliminate those with more significant cytotoxic potential [3].
Biocompatibility is especially important for restorative dentists as they use materials that remain in close contact with living tissues for a long time [2]. The increasing variety easily tested and are commonly used for cytotoxicity tests in dentistry are fibroblasts [28]. Normal fibroblast function is necessary to obtain periodontal tissue function.
There are certain limitations to the study and mathematical analysis of actual structures. These stem from the complicated and irregular construction of real objects visible at various scale levels. One of the methods used to analyze complex things is texture analysis (TA) and fractal dimension analysis (FDA).
A pixel is the smallest element of an image representing a specific color that is presented in digital photographs. Texture can be defined as an exemplary structure of an image that is created by pixels. A group of repetitive graphical attributes, such as brightness, smoothness, entropy, coarseness, linearity, or regularity, can characterize texture. Texture analysis provides quantitative, accurate, and sensitive detection of subtle changes in tested structures. TA comprises a series of mathematical techniques used to quantify and evaluate spatial variations in pixel grayscale intensities within a digital image [29]. Texture analysis is an essential tool used in various cases as object recognition, surface defect detection, pattern recognition, and medical image analysis. The classification of texture analysis methods highlights four categories: statistical methods, structural, model-based, and transform-based strategies [30]. TA is frequently used in dentistry for the analysis of computed tomography, magnetic resonance, or X-ray images [31][32][33].
A fractal is a geometrical object introduced by Mandelbrot that deals with self-similar forms compared to typical geometries taught in Cartesian and Euclidean mathematics [34]. While objects defined in Euclidean geometry have an integer topological dimension, fractals are characterized by an index of structural complexity called fractional dimension (FD). The difference between these two types and their FD is presented in Figure 1. Fractal dimension analysis (FDA) has provided a mathematical formalism for describing complex spatial and dynamical structures [35]. Self-similarity is commonly found in complex natural objects. Fractal dimension analysis based on image processing is broadly used in many areas of knowledge, including medicine, dentistry, technology, and materials science [34,36]. In materials, the fractal dimension is the most relevant parameter of surface topography [37]. It can provide information on subtle structural changes, or even mechanisms leading to its formation. The topography of a surface can be examined using such imaging methods as light microscopy, atomic microscopy (AFM), scanning tunneling microscopy (STM), and electron microscopy (SEM) [33]. FDA is broadly used in surface testing of such dental materials as lithium disilicate-based crowns [38], zirconia dental implants [39], or dental restorative composite [40].
This study aims to apply texture and fractal dimension analysis to evaluate subgingival cement surfaces in terms of biocompatibility.

Procedure for Preparing the Cement Samples
The research material was prepared using six cements frequently applied in the subgingival region: Glass Ionomer Cements: Ketac Fil Plus (3M ESPE, Seefeld, Germany), Riva Self Cure (SDI, Australia), Resin-based Cement: Breeze (Pentron Clinical, Wallingford, CT, USA), Zinc polycarboxylate cements: Adhesor Carbofine (Pentron, Chodov, Czech Republic), Harvard Polycarboxylat Cement (Harvard Dental, Great Britain), Zinc Phosphate Cement: Agatos S (Chema-Elektromet, Poland). The cements were prepared in the form of 0.7 cm × 0.7 cm cubes based on a matrix made of the plastic presented in Figure 2. The materials were prepared in line with the manufacturers' recommendations and then sterilized with ethylene oxide gas. The 2.5 h EOG sterilization cycle was performed at 55 • C, and then the research material was subjected to a 12 h degassing process.

Cell Line
The Balb/3T3 cell lines were cultured using the DMEM medium (Lonza, Basel, Switzerland) with 10% fetal bovine serum (FBS) and 1% L-glutamine with a penicillin and streptomycin solution (Sigma-Aldrich ® , St. Louis, MO, USA). Normal human dermal fibroblasts NHDF (Lonza) were cultured in the FGM TM Fibroblast Growth Medium BulletKit TM (Lonza). Cell cultures were conducted at 37 • C, 5% of CO 2 with constant air humidity in the HERA cell CO2 Incubator 150i (Thermo Scientific, Waltham, MA, USA).

Determination of Cytotoxicity
Determination of cytotoxicity was conducted using normal Balb/3T3 murine fibroblasts (American Type Culture Collection ATCC ® , Old Town Manassas, VA, USA), which is one of the in vitro models used in the biological evaluation of medical devices [41][42][43][44]. The cells were seeded on 6-well plates in the amount of 1.0 × 10 5 cells per well. After 24 h of culture, the medium was changed, the material was applied to each well, and the culture was conducted for 24 h. Subsequently, an evaluation using the Olympus CKX53 contrast-phase inverted microscope (Olympus, Tokyo, Japan) was conducted. Morphology of cells was assessed under the specimens, around the materials, and in the whole well. The cytotoxic effect was determined according to a four-grading scale where changes in the culture over grade 2 (mild grade) are considered to be the cytotoxic effect [41]. A cell culture without contact with the evaluated materials constituted the control in the study.

Cell Adhesion
The NHDF cells were carried out on materials in 6-well plates in the amount of 6 × 10 4 cells and cultured for 48 h and five days. During this time, the cell cultures were gently rocked to evaluate the actual adhesion of cells to the surface of the material.
Subsequently, the cells were stained with a mixture of DAPI, 0.1µg/mL (Thermo Fisher, Waltham, MA USA), and propidium iodide, 0.5 mg/mL (Roche, Mannheim, Germany), and visualized with Eclipse80i fluorescence microscope (Nikon Corporation, Tokyo, Japan). Cells without contact with the evaluated material constituted the control in the study.

Taking Images
All photographs were taken using the Techrebal K10E stereoscopic microscope (Techrebal, Wilczyce, Poland). The eyepiece was replaced by the ZWO ASI178mm monochrome digital camera (ZWO CO., LTD., Suzhou, China). All photographs were taken using an 18× magnification and 3096 × 2080 resolution. The time of exposure was set to achieve histogram filling at the level of 90%, depending on the cement surface. The gain parameter (the sensitivity of CMOS matrix) was the same during all procedures, and it was set to 10 to reduce the noise. A camera in a 14-bit mode was used to achieve the widest dynamic range of photographs. The images were saved as 16-bit TIFF (Tagged Image File Format) files. Two graphical operations were applied to normalize all images for further analysis: auto-levels and high-pass filter to decrease the effect of non-homogeneous illumination of the examined samples ( Figure 3). Subsequently, the images were saved as 8-bit grayscale bitmaps. All graphical operations were performed using GIMP, version 2.

Fractal Dimension Analysis
All fractal analyses were performed in ImageJ, version 1.53e (Image Processing and Analysis in Java-Wayne Rasband and contributors, National Institutes of Health, USA, public domain license, https://imagej.nih.gov/ij/ accessed 1 October 2021), and plugin FracLac, version 2.5 (Charles Sturt University, Australia, public domain license, accessed 1 October 2021).
In our study, we decided to use a modified algorithm of the counting box method, which makes it possible to analyze monochromatic images, such as 8-or 16-bit images. In the case of grayscale images, we applied the intensity difference algorithm to calculate fractal dimension. The analyzed image is divided into boxes as in the counting box method. The difference between the maximum pixel intensity and the minimum pixel intensity is calculated in each box (δI i,j,ε , where i, j-the location of the analyzed box in the ε scale) In the next step, 1 is added to the intensity difference to prevent its value from becoming a 0 I i,j,ε = δI i,j,ε + 1 Finally, the fractal dimension of the intensity difference is described using the following formula All operations are shown in Figure 4. Ii,j,ε= δIi,j,ε + 1 (2) Finally, the fractal dimension of the intensity difference is described using the following formula All operations are shown in Figure 4.

Texture Analysis
The region of interests (ROIs) were normalized (µ ± 3σ) to share the same average (µ) and standard deviation (σ) of optical density within the ROI for texture analysis. Selected image texture features (entropy and difference entropy from the co-occurrence matrix, and long-run emphasis moment from the run-length matrix) in ROIs were calculated where Σ is the sum, Ng is the number of grey levels in the microphotograph, i and j stand for a grey level of pixels 5-pixel distant one from another, p is the probability, log is the common logarithm.
where Σ is the sum, Nr is the number of series of pixels with grey level i and length k, Ng is the number of grey levels for the image, Nr is the number of pixels in series, p is probability [45,46]. These three equations were subsequently used for the texture index construction [47]. Finally, the texture index (TI) and bone index (BI), which represent the ratio of the measure of the diversity of the structure observed in the microphotograph to the measure of the presence of uniform longitudinal structures, was calculated The index defined in this way [48] was taken as a measure of cement surface structure.

Statistical Analysis
Statistica version 13.3 (StatSoft, Cracow, Poland) was used to perform all statistical tests. A value of 0.05 was deemed to be statistically significant. The Shapiro-Wilk test was used to confirm the normality of distribution. Due to a normal distribution, parametric tests were performed. The analysis of variance (ANOVA) and the least significant difference post hoc were applied to reveal the fractal dimension differences between the examined surfaces of cements. A correlation matrix was used to estimate the correlation between FD's measured surface and the adhesion of fibroblasts. Qualitative properties of cells were converted to quantitative parameters. For amounts of cells: lack of cells-0, few cells-1, cells on surface-2, culture of cells-3; for cytotoxicity: none-0, moderate-1, severe-2. Correlation coefficient were calculated two times between FD, TI, BI and amounts of cells (after 48 h and 5 days). The value of this coefficient was lower after 5 days than after 48 h Table 1 presents a summary of the mean value of the fractal dimension for individual cements with the results of cell adhesion and cytotoxicity. Our study revealed an average correlation coefficient (r = 0.47) between the fractal dimension and the number of fibroblasts after 48 h. After 5 days of fibroblast incubation, the correlation coefficient dropped to 0.31. Figure 5 shows examples of light microscope images of the surfaces of the investigated cements subjected to a subsequent analysis of textures and the fractal dimension. Table 1. Mean values of FD for individual cements and results of fibroblast adhesion tests after 48 h and 5 days of culture as well as evaluation of cytotoxicity of the tested cements (SD-standard deviation, FD-fractal dimension). Qualitative properties of cells were converted to quantitative parameters. For amounts of cells: lack of cells-0, few cells-1, cells on surface-2, culture of cells 3; for cytotoxicity: none-0, moderate-1, severe-2.

Cement
lack of cells- few cells- cluster of cells (3) few cells-

Amount of cells after 5 days
lack of cells- few cells- lack of cells- lack of cells- cells on surface (2) cluster of cells (3) Cytotoxicity severe- moderate- moderate- moderate- none- none-(0)
Altered cells morphology was observed under and near the sample, such as lysis, rounded, vacuolated cells. In the rest well cell, morphology was normal, comparable to that of cells in the control culture that had no contact with the test materials.
Riva Self Cure (SDI, Australia) and Ketac Fil Plus (3M ESPE, Germany) showed a 0 degree of cytotoxicity and did not change the cell culture.
Cytoxicity of Adhesor Carbofine (Pentron, Czech Republic) was rated 4 on a 0-4 scale. After the contact with this material, the cell culture showed features of a disrupted culture.
According to PN-EN ISO 10993-5: 2009 Biological evaluation of medical devices-Part 5: Tests for In vitro cytotoxicity, medical devices with a cytotoxicity grade above 2 are considered to have a cytotoxic effect [41]. Morphology of Balb/3T3 after direct contact with the materials is presented in Figures 6 and 7 and Table 3.  The investigated materials showed different adhesive properties. Riva Self Cure (SDI, Australia) and Ketac Fil Plus (3M ESPE, Germany) had the most favorable properties, promoting adhesion of the NHDF cells to the surface. Cells were also observed on the surface of Agatos S (Chema-Elektromet, Poland) but in a smaller amount. No cells were observed on the surface of Breeze (Pentron Clinical, USA) and Adhesor Carbofine (Pentron, Czech Republic). The conducted research also revealed that there is a relationship between the cytotoxicity of materials and their adhesive properties. In the case of Riva Self Cure (SDI, Australia) and Ketac Fil Plus (3M ESPE, Germany) that showed no cytotoxic effect, Riva Self Cure (SDI, Australia) appeared to have a little better adhesive properties than Ketac Fil Plus (3M ESPE, Germany) as cell clusters were observed after 5 days of cultivation, which may indicate that the material promotes coating by the cell layer. This could suggest that the NHDF cells proliferate on this surface (Figure 8).

Texture Analysis
Microphotographs of cement surfaces revealed significant differences among tested materials (Table 4 and Figure 9). The highest values of a texture index and Bone Index were noted in the case of Ketac Fil Plus (3M ESPE, Germany), Adhesor Carbofine (Pentron, Czech Republic), and Riva Self Cure (SDI, Australia). The lowest values of these features were noted in the case of Breeze (Pentron Clinical, USA) and Agatos S (Chema-Elektromet, Poland).  The R-Squared statistic indicates that the model as fitted explains 8.5% of the variability in TI (p < 0.05) at the fifth day of incubation. The correlation coefficient equals −0.29, indicating a relatively weak relationship between the variables. The more developed the surface, the lower the cytotoxicity (in other words, more living cells adhere to the more developed surface: TI = 1.87042 + 0.0389976 × Number of Cells 2 ). A similar relationship, and a weak one, was found for descriptions of surface structure using BI: R-Squared = 7%; correlation coefficient = −0.27; BI = 1/(1.37535 − 0.0278543 × Number of Cells 2 ; p < 0.05).

Discussion
In this study, glass ionomer materials (Ketac Fil Plus (3M ESPE, Germany), Riva Self Cure (SDI, Australia)) showed no changes in fibroblast growth. The results are consistent with the majority of publications where glass ionomer cement showed low cytotoxicity [28,49,50]. Some research shows that the cytotoxic effects of GI cement are time-dependent. Mallineni et al. noticed that freshly-prepared GI materials are mildly cytotoxic but their cytotoxicity decreases over time. The reason for this phenomenon is fluoride release, which has a therapeutic value but causes cytotoxicity [4]. A study conducted by Lang et al. confirmed that viability and proliferation of fibroblast cells changed with time and showed no differences when compared to the control group by the 21st day of the study [51]. A negative biological influence of GI cement was demonstrated in a study by Milhem et al. It revealed that GI cement showed greater cytotoxicity than other materials, including composites [52].
The results of this study indicate that the highest cytotoxicity is shown by Zinc Polycarboxylate Cements, Adhesor Carbofine (Pentron, Czech Republic) and Harvard Polycarboxylat Cement (Harvard Dental, Great Britain). The cytotoxic effect of this group of cements has been confirmed by studies. A study conducted by Schmid-Schwap et al. suggests that the release of zinc ions and acidity may be the cause of cytotoxicity [53].
In this study, the resin-based material and zinc phosphate cements Breeze (Pentron Clinical, USA) and Agatos S (Chema-Elektromet, Poland) are characterized by moderate cytotoxicity. Some studies have shown that the resin-based materials are more cytotoxic than others. Monomers such as TEGDMA, HEMA, BisGMA, and UDMA reduce cell viability and generate a breakdown of the mitochondrial membrane in fibroblasts [54]. A study conducted by De Souza Costa et al. showed that light-cured resins are less cytotoxic than chemically-cured systems. The results depend on the curing efficiency of the polymerization lamp and the type of the resin system [3].
Fractal analysis was used by Salareno et al. to investigate the effect of air polishing on the surface of composites. They concluded that the use of glycine in air polishing generates the least roughening surface, and it correlates with the disappearance of the surface fractal character [37]. Talu et al. studied the influence of artificial saliva storage on 3-D surface texture characteristics of dental nanocomposites by using multifractal analysis. The exposure of artificial saliva storage changes anisotropic surface texture to more isotropic [55]. Glass ionomer surface roughness was investigated by Reddy et al. They used one-way ANOVA and Tukey's significant difference tests to compare the surface of the material after conditioning it in citric acid of a different pH. The effects of pH on the surface texture of a glass ionomer cement depend on the used material and cause deterioration of a type II glass ionomer cement [56]. FDA is also used in the analysis of the surface of prosthetic restorations. Schestatsky et. al. demonstrated that the CAD/CAM technique produced smoother but more complex topography features (higher FD) than the pressing technique [35].
FDA is also used in dentistry to test organic tissues, such as teeth or bones. In a work by Nezefat et al., molar teeth with their enamel, the dentin, and cementum were tested using a power spectral density analysis and fractal dimension through AFM images. Hayek et al. tested radiographic estimation of bone quality using FDA. They noticed that it was a useful and non-invasive tool for examining bone density before the implantation, and proposed an image-based classification of bone density in the posterior regions of each jaw [57]. FDA can also be used as an objective method for detecting bone destruction induced by periodontitis. In the study conducted by Belgin et al., it was demonstrated that the FD values in patients with periodontitis were significantly lower than in the group without the disease [58].
Among the materials studied, Ketac Fil Plus (3M ESPE, Germany), Adhesor Carbofine (Pentron, Czech Republic), and Riva Self Cure (SDI, Australia) had the most extensive surface areas. This is indicated by a high value of TI and BI indexes. At the other end are the materials Agatos S (Chema-Elektromet, Poland) and Breeze (Pentron Clinical, USA). The latter cement, in particular, has a very smooth, homogeneous surface.

1.
The Ketac Fil Plus (3M ESPE, Germany) and Riva Self Cure (SDI, Australia) cements provided the most favorable conditions for fibroblast adhesion, despite the statistically significant differences in the fractal dimension between them. Insofar as the surface texture is considered, Ketac Fil Plus (3M ESPE, Germany) cement should be indicated here.

3.
Moderate positive correlation was observed between fractal dimension (FD) and the amounts of cells after 48 h. The correlation coefficient was decreased to a weak positive linear after 5 days. 4.
The study revealed a moderate negative linear correlation between texture index (TI), Bone index (BI), and the amounts of cells after 48 h of incubation. The correlation coefficient decreased to weak negative linear after 5 days.

1.
Flat surfaces of cements were investigated to simplify taking microscopic photos for analysis. In real restorations, the surface is most commonly a cylinder; this may affect fibroblast culturing.

2.
Some of the examined cements: Ketac Fil Plus (3M ESPE, Germany) and Riva Self Cure (SDI, Australia) may release fluoride ions which may affect cytotoxicity.