Resolution of dental implant

In current dentistry, titanium implants are often used. The success rate of an implant is characterized by the presence of direct con-tact between the implant surface and sur-rounding bone. The bright field microscopy of histologically stained slices remains the most frequently used tool for qualitative and quantitative evaluation of the bone response to an implant material in preclinical studies. Other conventional methods such as trans-mission electron microscopy (TEM), scanning electron microscopy (SEM), and atomic force microscopy (AFM) are available to study the bone–implant interface. Likewise, such two-dimensional analysis techniques are not com-pletely representative of the whole specimen. As a result, also computed tomography (CT) is used frequently as a non-invasive non-destructive three-dimensional (3D) method to study radiopaque structures. On a microscopical level, the micro-CT technique is useful to determine bone volume and bone structural parameters . Still, the micro-CT technique for research approaches in bone has three major shortcom-ings. First, micro-CT systems contain various limitations, which restrain the image recon-struction and final quantification . Reconstruction artifacts, reconstruction cen-ter errors, mechanical imperfections (such as specimen axis wobble and rotation axis mis-alignment), and beam hardening can occur. Some of the limitations can be corrected using software algorithms, for example, for beam hardening and ring reduction (Stoppie et al. 2007; Banhart 2008; Zou et al. 2011). The occurrence of scattering at the titanium implant–bone interface is a second problem. For research purposes, scattering can be greatly reduced when polymer implant materi-als are used, which are equipped with a tita-nium surface coating (Shalabi et al. 2007). The third limitation is the resolution of conven-tional micro-CT, which is inferior to conven-tional two-dimensional (2D) microscopic images. To overcome these problems, higher spatial resolution and contrast sensitivity are required. Possibly, this can be achieved using CT systems at sub-micron resolution, such as synchrotron-CT (SRlCT), SEM-based nano-CT instruments, and stand-alone tube-based Table 1. Composition, size, density, solid content, and particles per milliliter of micro- and nanospheres

	Particle composition	Particle size ( lm/nm)	Density (g/cm3)	Solid content (% WT)	Particles//ml
	Polystyrene	100 lm	1.05	2	3.5*104
		10 lm	1.05	2	3.5*107
		1 lm	1.05	2	3.5*1010
	Polymethacrylate	100 lm	1.19	10	2.0*105
		50 lm	1.19	10	1.6*106
		20 lm	1.19	10	2.5*107
	SiO2	1 lm	1.8–2	2,5	7.9*1010
		500 nm	1.8–2	5	1.3*1012
	TiO2	1 lm	2.2	2.2	8.1*1010
		500 nm	2.2	2.5	7.3*1011

nano-CT systems (Banhart 2008; Stock 2009). However, comparative studies of the 3D bone– implant interface area between the micro- and nano-CT approaches have not been performed yet. In view of the above-mentioned facts, the goal of this study was dual. Firstly, it was aimed to determine the spatial resolution and sensitivity of micro- versus nano-CT using standardized samples composed of well-defined synthetic microspheres embedded in matrices of varying X-ray absorption coeffi-cients. Secondly, the objective was to vali-date the micro- versus nano-CT technique in vivo. For this purpose, titanium-coated poly-methyl methacrylate (PMMA) implants were placed in the mandible of Beagle dogs, a fre-quently used model for biomedical implant interface studies (Calvo Guirado et al. 2009; Junker et al. 2011). After harvesting and fix-ing the biological specimens, ex vivo micro-CT and nano-CT were performed and compared with descriptive histology and his-tomorphometry. were dried overnight at 37°C. Thereafter, the spheres were mixed with, respectively, sili-cone rubber and calcium phosphate cement (CPC) matrices. Silicone rubber samples were created as fol-lows. Each sphere type was mixed with a total volume of 0.7 ml of silicone compound (Elastosil RT 601 B and Elastosil RT 601 A, ratio 1:20; Wacker Silicones, Riemerling, Germany) and placed in an Eppendorf tube (Eppendorf AG, Hamburg, Germany). After overnight polymerization at 37°C, the sili-cone samples were removed from the tubes. CPC consisting of 0.5 g calcium phosphate mixture (85% a-tri-calcium phosphate (Cam Bioceramics B.V., Leiden, The Netherlands), 10% CaHPO₄ (Sigma, St. Louis, MO, USA), and 5% HCa₅O₁₃P₃ (J.T. Baker Chemical Co, Phillipsburg, NJ, USA) was blended for 20 s with each type of sphere. Thereafter, 190 ll of 2% Na₂HPO4 (Merck, Darmstadt, Ger-many) solution was added and thoroughly mixed for 20 s. The thus-obtained deformable paste was placed in cylindrical Teflon molds, and the CPC was hardened overnight at 37°C. Both silicone rubber and CPC samples were finally trimmed to a fixed volume of approximately 8 mm³ for further micro- and nano-CT analysis. To confirm sphere size, surface texture, and distribution of the nano- and micro-spheres in both samples, SEM analysis was performed. The specimens were coated using a 30-nm thin layer of Au–Pd, using a high-resolution sputter coater (208HR; Cressing-ton Scientific, Watford, UK), and imaged using a field emission SEM (JSM 6330F; JEOL Corporation, Tokyo, Japan) operated at an accelerating voltage of 3 keV. Implant design Custom-made PMMA implants (3.2 mm inby 4 mm in length) were used. All implants were coated with titanium (Ti), using a radio frequency magnetron sputtering unit (Edwards ESM 100, Crawford, UK). During the sputtering process, a commercially pure Ti target was used. The Ti target was sputter-cleaned before deposition. The target was considered clean when the plasma turned from pink to blue, resulting from blue light–emitting Ti in the plasma. Pressure was kept at 5.0 9 10 ³ mbar using argon gas (4 l/ h), and the power was set at 100 W. Thereaf-ter, the PMMA implants were sputter-coated for 3 9 5 min, with a rotation of the implant of 120° before each time step to ensure homogenous coverage. The coating thickness was determined to be 10 nm using a Nano-scope IIIA atomic force microscope (Veeco Industries, Santa Barbara, CA, USA). Before use in the animal study, the implants were autoclaved for 15 min at 121°C. Implantation procedure and implant retrieval Ten adult Beagle dogs (1–2 years old, weight 10–12 kg) were used. The research protocol was approved by the ethical committee of King Saud University (Riyadh, Kingdom of Saudi Arabia), and national guidelines for care and use of laboratory animals were observed. The animals were anesthetized, and after intubation, general anesthesia was maintained with Isoflurane (Rhodia Orga-nique Fine Limited, Avonmouth, Bristol, UK). In a first surgical session, three premolars (P2-4) were delicately removed in the left side of the mandible. Healing time for the extrac-tion sockets was 3 months. In a second surgical session, in the left side of the mandible of the 10 Beagle dogs, 10 implants were placed, that is, one implant per dog. To reduce peri-operative bleeding, local anesthesia was performed with 2% lido-caine-containing epinephrine (1 : 100,000). Before surgery, the local mucosa was cleaned with a 10% povidone-iodine and incised. A bone defect of 3.2 mm in diameter by 8 mm in depth was created by drilling, using low rotational drill speeds (800–1200 rpm) and continuous external cooling with saline. The implant was placed press-fit at the bottom of the defect. Finally, the gingiva was sutured and all implants were left in a submerged position for an implantation time of 4 weeks. After surgery, Finadyne (Schering corpora-tion, Kenilworth, NJ, USA) and a broad-spec-trum antibiotic (Gentamycin, SPIMACO, Qassim, Saudi Arabia, 4 mg/kg body weight intramuscularly) were given for 7 days. At the end of the implantation period, the animals were euthanized by an overdose of pentobarbital, and the implants with sur-rounding tissue were harvested. The speci-mens were fixed in 10% neutral-buffered formalin solution with subsequent dehydra-tion in a graded series of ethanol (70–100%). Thereafter, specimens were hemi-sectioned. The top part was used for micro-CT and subse-quent nano-CT analysis. For nano-CT, it was necessary to reduce the size of the specimen further, as the maximum size for nano-CT specimens was approximately 2.5 9 2.5 9 2.5 mm. Sectioning and reduction of the specimen were performed using a Leica EM TXP target preparation device (Leica Micro-systems, Wetzlar, Germany), with a 300-lm diamond-coated disk cutter blade. The bottom part was used to prepare histological sections. Micro- and nano-computed tomography The standardized samples, as well as the implant samples, were scanned using the Skyscan Micro-CT system model 1172 and nano-CT system model 2011 V 2.3 (Skyscan, Kontich, Belgium). Selection of these systems was based on the minimal image pixel size, 0.8 lm and 150 nm, respectively (Sasov et al. 2008). The SkyScan 1172 system was operated at 100 kV and 100 lA, and the exposure time was set to 260 ms. Scanning was performed by 180° rotation around the vertical axis and with a rotation step of 0.5°. During scanning, 0.5-mm aluminum filter was used. The nec-essary field of view (FOV) for the standard-ized and implant samples was determined, resulting in an optimal image pixel size of 3 and 4.37 lm, respectively. Subsequently, the projected images were reconstructed in 2000 9 2000 pixel-sized cross-sectional images using the Skyscan NRecon cone beam reconstruction software with beam-hardening compensation and ring artifact cor-rection set to 30% and 15, respectively. For nano-CT, samples were scanned 180° around the vertical axis in rotation steps of 0.3 degrees. The X-ray source was set to a voltage of 40 kV and current of 180 lA. The exposure time was 2.0 s using a detector of1280 9 1280 pixels. Raw data were recon-structed with the NRecon cone beam recon-struction software. Images with pixel size of 500 nm and 2 lm were generated from the standardized and implant samples, respec-tively. As a result, the FOV was 2.56 9 2.56 mm; that is, capable of including the entire specimen. Micro- and nano-computed tomography analysis Skyscan CTAn V.1.11 software was used for segmentation, registration, and quantification of all reconstructed images. The obtained images of the standardized samples were used to determine resolution and sensitivity limit. The digital ruler of the CTAn software program was used to measure the smallest visible sphere present in 10 representative images, with known image pixel size. The variation in radiodensity was detected using the gray scale histogram (range 0–255) values of the complete data set of the standardized sample. For the implant samples, bone regenera-tion, routinely, is detected in a 500-lm zone from the surface of the implant (Schouten et al. 2009; Junker et al. 2010a,b). Thus, in 2D cross sections of the reconstructed CT images, a standardized threshold of the gray values for bone within the total data was set. Then, the bone area (BA) percentage was ana-lyzed in the region of interest (ROI) zone of 500 lm surrounding the implant. Further, for the complete 3D reconstruc-tions, the bone volume (BV) percentage was calculated in a fixed volume of interest (VOI) zone of 500 lm surrounding the implant surface with a depth of 2 mm. Also, the bone-to-implant contact (BIC) percentage was calculated using an algorithm including automated adaptive segmentation of the bone and morphology-based operations as dilatation. Histology and histomorphometry The bottom part of the specimens was embedded in epoxy resin (EpoFix ; Struers, Rødovre, Denmark). After polymerization, thin sections of the implants were prepared using a modified SP1600 inner circular saw-ing microtome (Leica; van der Lubbe et al. 1988) and stained with methylene blue and basic fuchsin. The sections were made in a cross-sectional direction perpendicular to the longitudinal implant axis. Acquisition of microscopical images and histological evalua-tion were performed using an automated light microscope (Axio Imager Z1; Carl Zeiss Micro Imaging GmbH, Gottingen,€ Germany). With the aid of the digital image processing software module MosaiX of the Axiovision V4.6.3 software package combined with a motorized microscope stage, a high-resolu-tion image (0.27 lm pixel size) of the com-plete ROI was obtained. For the final image quantification, a com-puter-based image analysis system (Leica QWIN Pro image analysis software; Leica Imaging Systems, Cambridge, UK) was used. For calibrating the Leica QWIN quantifica-tion program for further histomorphometric analysis of the microscopic images, the known pixel size was imported. The ROIs of corresponding planes of the 2D micro-CT and histological slices were quantified in lm². The BA percentage was determined in a 500-lm ROI zone from the implant surface, and the BIC percentage was calculated using the perimeter of the implant and total of contact parts of bone tissue with the implant surface. Statistical analysis The mean gray values of both silicone and CPC matrices when using micro- and nano-CT were compared using an unpaired t-test (InStat, version 3.05; GraphPad Software Inc., San Diego, CA, USA). Statistical analysis of BA, BV, and BIC percentage measurements of the different samples using both CT tech-niques and histology were conducted using paired t-tests (SPSS version 20, IBM Statis-tics, Armonk, NY, USA). For repeated testing of BA and BIC, Bonferroni correction was applied. Results Standardized samples Using SEM, all four types of micro- and nanospheres were detected by visual inspec-tion and observed to be distributed evenly in both silicone rubber and CPC matrices. The rough surface of CPC made it more difficult to distinguish 1-lm and 500-nm spheres in CPC (Fig. 1). The micro-CT system allowed the detec-tion of the 10- and 100-lm microspheres inside the silicone rubber as well as CPC matrix. On the other hand, due to limitations in spatial resolution, the one-micron spheres could only be detected using the nano-CT system, while the 500-nm spheres could not be detected at all. The sensitivity limits were also deter-mined. After normalization, the gray value ranges in the histograms of the radio absorption in micro-CT were 31–124 (mean 49.0 19.3) for silicone matrices and 55–187 (mean 104.3 21.7) for CPC Cuijpers et al High-resolution CT of implant–bone interface

matrices. When using nano-CT, gray value ranges in the radio absorption histograms were 39–138 (mean 55.43 25.7) for silicone matrices and 50–187 (mean 89.3 25.8) for CPC matrices. Due to the more uniform radiodensity of the silicone matrix, the microspheres could be easily discriminated in the obtained histo-gram. The more irregular texture and varia-tion in radiodensity in the CPC allowed for less sensitivity (Fig. 2). There was no significant difference in the mean gray values of both silicone and CPC samples when using micro- and nano-CT. P-values were 0.804 and 0.593 for silicone and CPC, respectively. Implants At the end of the 4-week implantation time, one implant was found to be lost. All other implants could be harvested, and there were no visible signs of an excessive inflammatory response or adverse tissue reaction. Micro- and nano-CT evaluation Using micro-CT, the total field of view (8.7 mm), containing the implant with a sur-rounding 500-lm-wide bone area, could be analyzed at 4.37 lm resolution in terms of pixel size. No scattering of the thin Ti layer applied on the PMMA implant surface was found. The PMMA implant and the bone were visible at high magnification (Fig. 3a,b). Morphological details of the bone were restricted to overall structure including the shape of the trabeculae. The extraction of the 3D VOI from the total sample is shown in Fig. 4a,b. Using nano-CT, even at 2 lm resolution, the limited FOV (2.56 mm) included the entire ROI. Similar to micro-CT, no scatter-ing artifacts were found. Compared with micro-CT, in the nano-CT, a much higher histomorphological details of bone and other tissues could be detected (Fig. 3d,e). Even the structure of the osteons and small blood ves-sels was clearly visible (Fig. 3g). For both CT techniques and histology, the BA, BIC, and BV percentage results and obtained paired t-test data are listed in Table 2.

Histological evaluation Histological sections showed excellent pre-servation of the bone morphology when embedded in epoxy resin. Because of sample loss and tissue processing, 8 of the 9 samples were finally suitable for final histological analysis. The outline of the Ti-coated PMMA implant was always visible. Pre-existing and newly formed bone was clearly visible. In areas where no bone contact existed, locally a thin layer of stromal cells was lining the implant interface (Fig. 3f). Moreover, histo-logical details as osteons and small blood ves-sels were clearly visible due to sub-micron resolution (Fig. 3h). Different bone cell types were clearly visi-ble when using magnifications with pixel sizes up to 0.27 lm. Figure 4c,d visualizes the extraction of the 2D ROI from the histo-logically stained slice for analysis. Quantifi-cation of the BA and BIC of the histological slices is listed in Table 2. Comparative analysis of CT and histomorphometric data Statistical comparison of the micro-CT, nano-CT, and histomorphometric paired data showed no significant difference in BA (P = 0.869) and BIC (P = 0.881) for micro-CT versus histology. In contrast, BA and BIC as assessed in nano-CT were always signifi-cantly lower compared with micro-CT (P = 0.040 and P = 0.064) as well as histology (P = 0.051 and P = 0.052). BV could not be determined in 2D histomorphometric analy-sis. For BV, a significantly lower value was found in nano-CT compared with the micro-CT analysis. Notable was also the large diversity in standard deviations for nano-CT (from 35 up to 295%) compared with micro-CT and light microscopy. Discussion This study aimed to evaluate spatial resolu-tion and sensitivity of micro- versus nano-CT, including a comparative validation of both techniques versus light microscopical imaging as the gold standard. Overall, it was found that nano-CT provided better resolu-tion, which proved to result in a detail of observation close to histological assessment. However, in terms of quantification, micro-CT showed to result in a more reliable assessment of bone healing parameters matching histomorphometry. Although the image acquisition principles between nano-and micro-CT are similar, there are limita-tions in sample size. For nano-CT, the achievable volume of interest is smaller, which affects the reliability of the obtained data. In this case, the measurement turned out to be too low. In addition, the individual measurements showed a large statistical scattering resulting in higher diversity of the standard deviations. Therefore, a direct com-parison between both CT techniques reveals a more optimal resolution for nano-CT, but a superior capacity for quantification for micro-CT. Thus, both techniques should be regarded as complementary on the basis of details and accuracy. Considering the current study design, sev-eral remarks have to be made. First, the com-parative steps were performed using standardized samples made of silicone rubber and ceramic material. This was done pur-posely, as these two matrices exhibit com-pletely different radio absorption coefficients, hardness, and texture. As a consequence, they provide the optimal means to assess resolution and sensitivity of the used CT sys-tems. Resolution in nano-CT was found to be significantly better, but sensitivity appeared to be similar. Probably, the use of synchrotron micro-CT with a high-intensity monochromatic X-ray beam, combined with phase contrast–enhanced imaging, is a better approach to increase sensitivity (Stiller et al. Cuijpers et al High-resolution CT of implant–bone interface

Table 2. Mean and SD of the bone area (BA), bone–implant contact (BIC), and bone volume BV percentage in micro-CT, nano-CT, and histology. Shown are number of measurements (N), compared groups, mean difference (MD), 95% confidence interval (CI), and P value for BA, BIC, and BV 2009; Yue et al. 2010; Appel et al. 2011) and result in a better sensitivity.

Mean   SD	N	Compared groups	MD (95% CI)	P value
BA micro-CT	86.3	6.8	3	BA micro-CT vs. BA nano-CT	21.4	(2.3, 40.5)	0.04
BA nano-CT	65.5	11	7	BA micro-CT vs. BA histology	0.2	(  3.0, 3.5)	0.869
BA histology	84.1	8	3	BA nano-CT vs. BA histology	18.3	(  39.1,2.6)	0.064
BIC micro-CT	84	4	3	BIC micro-CT vs. BIC nano-CT	21.5	(  0.1,43.1)	0.051
BIC nano-CT	62.8	11.7	7	BIC micro-CT vs. BIC histology	0.3	(  4.4, 5.0)	0.881
BIC histology	82.5	5.9	3	BIC nano-CT vs. BIC histology	19.1	(  38,8, 0.5)	0.052
BV micro-CT	81.4	5.5	3	BV micro-CT vs. BV nano-CT	26.7	(  58.4, 111.8)	0.157
BV nano-CT	55.4	7.4

For the comparative evaluation of CT and histological imaging, Ti-coated PMMA implants were placed in the mandibles of Bea-gle dogs. This implant design was chosen as the most reliable model to quantitate bone healing parameters, as the use of solid tita-nium implants affects the CT assessment negatively, due to electron-scattering artifacts (Shalabi et al. 2007). Even when specific combined metal filters are used to reduce scattering at the implant surface (Cha et al. 2009). A polymer implant behaves superior due to the diminished occurrence of X-ray scattering, while the surface coating of tita-nium serves as a close representation of a solid titanium implant. PMMA was chosen as the implant core because Perilli et al. (2007) reported that this polymer is very suitable to study structural parameters of human bone biopsies with micro-CT without obtaining undesirable artifacts. Besides these aspects related to the CT imaging, a second reason to apply a PMMA-based implant was based on the ease of histo-logical processing. Previously, it was shown that polycarbonate plastic implants, vapor-coated with a layer 120–250 nm of titanium, were applicable for implant research purposes. Such implants were integrated in rabbit bone at 12 weeks of implantation and subsequently could even be processed for electron micros-copy (Linder et al. 1983). Albrektsson & Hans-son (1985) corroborated such results in a later study, using a 100-nm titanium or zirconium coating, which enabled light and electron microscopy to study the host bone response in high details up to protein attachment level. Thus, our current study followed such princi-ple as well. Because our implant was made of PMMA, for subsequent histological analyses, embedding in epoxy was preferable over rou-tine MMA embedding. The final validity of CT measurements was performed in comparison with such histological analysis. Of course, no direct comparison could be made as the implant was split in an upper part for CT and lower part for histology. Still, Britz et al. (2010) showed that the matching of the same images, but acquired via different tech-niques, can also not always be performed at a high accuracy. Histological processing of the samples can result in minor loss of form; moreover, there will always be intrin-sic differences in thickness and localization between histological sections in contrast to the CT-reconstructed cross sections. Others compared the use of micro-CT and histol-ogy for the evaluation of cortical bone structure and were capable of obtaining a very close correlation between the three-dimensional micro-CT parameters versus two-dimensional histological sections (Parti-celli et al. 2011). Herein, we assumed that bone morphology was similar between upper (CT) and lower (histology) parts of the implant for regions less than 500 lm apart. The present study confirmed the cor-relation between micro-CT and histology for the analysis of bone parameters, as also had been determined earlier (Stoppie et al. 2005, 2007). In contrast, the use of nano-CT was found to be less accurate, resulting in lower values compared with micro-CT and light microscopical histomorphometry. The explanation for this observation is that micro-CT is able to study larger samples (~centimeter range) compared with nano-CT (~millimeter range). In addition, when nano-CT is used at nanometer resolution, analy-sis of only a very limited VOI is possible. This will always result in over- or underes-timation errors during quantification. As a second consequence, also the diversity of the standard deviation is large. Only a very limited number of publica-tions are available describing the use of stand-alone nano-CT systems for the analysis of bone tissue (Salmon & Sasov 2007; van Hove et al. 2009). Salmon et al. explored the resolution limitations of nano-and micro-CT when studying the ultrastruc-ture of bone. The juvenile murine fibula was chosen as a useful site for nano-CT imaging due to the requirements of a very small specimen size for attaining optimum resolution. A range of histomorphometric parameters were generated including bone area, osteocyte lacuna characteristics, and cellular resorption and formation factors. Nevertheless, no validation of obtained parameters versus other measuring tech-niques was implemented.