Phase contrast in laboratory-based X-ray CT

Y. De Witte1, M. N. Boone1, B. Pauwels2, E. Herremans3, E. Van de Casteele2, L. Van Hoorebeke1

1 UGCT, Ghent University, Proeftuinstraat 86, 9000 Ghent, Belgium Yoni.DeWitte@UGent.be 2 SkyScan, Kartuizersweg 3B, 2550 Kontich, Belgium 3 MeBioS, K.U.Leuven, Willem de Croylaan 42, 3001 Heverlee, Belgium

When scanning weakly attenuating samples using an X-ray tube with a very small focal spot size, the acquired projections not only contain absorption contrast, but also phase contrast due to refraction of the X-rays. Although the resulting edge enhancement can be beneficial in radiography, the tomographic reconstruction of such mixed projections using a conventional algorithm yields cross-sections that are severely distorted by so-called phase contrast artefacts. These artefacts can be prevented by applying a phase retrieval method like the Modified Bronnikov Algorithm (MBA)1, or the similar simultaneous phase and amplitude extraction2, which allows reconstructing the object’s refraction function instead of its attenuation function. Alternatively, one can use the Bronnikov-Aided Correction (BAC)3 to reduce the phase contrast signal in the projections, which can then be reconstructed without introducing phase artefacts in the reconstructed attenuation function. It was already shown that these two complementary approaches can significantly improve the reconstruction quality of high-resolution CT scans of a variety of samples3,4. Since neither method requires changes in the scanner hardware or the acquisition, they can be readily applied to projection data from any high-resolution scanner. Here, we present results for the application of both methods on projection data of several samples acquired using the SkyScan2011 x-ray nanotomograph. As is illustrated in figure 1, the resulting cross-sections provide a more realistic representation of the sample and are much better suited for further quantitative analysis.

Figure 1: Reconstructed slice of a foam for the Inside Food project, without any special processing (left), using the phase retrieval method MBA (middle) and the

phase reduction method BAC (right).

References: 1. A. Groso et al., “Implementation of a fast method for high resolution phase

contrast tomography”, Opt. Express, 14(18), 8103-8110, 2006 2. D. Paganin et al., “Simultaneous phase and amplitude extraction from a

single defocused image of a homogeneous object”, J. Micr., 206(1), 33-40, 2002

3. Y. De Witte et al., “Bronnikov-aided correction for x-ray computed tomography”, J. Opt. Soc. Am. A, 26(4), 890-894, 2009.

4. M. N. Boone et al., “Practical use of the modified Bronnikov algorithm in micro-CT”, Nucl. Instrum. Methods Phys. Res. Sec. A, 267(7), 1182-1186, 2009

Studying Aquatic Insects Anatomy with the SkyScan 1172 high-resolution micro-CT

J. Alba-Tercedor & C.E. Sáinz-Cantero Caparrós

Department of Animal Biology. Faculty of Sciences. University of Granada. 18071-Granada. Spain. jalba@ugr.es

Aims Possibility to study details never seen before with classic microscopy methods on the anatomy of insects and the new possibilities offered by the micro-CT new techniques encouraged us to investigate theses possibilities with a Skyscan 1172 microtomograph. The material was preserved in alcohol so first it all we tried to scan directly specimens and the results were somewhat disappointing. Thus we investigated different ways until we could get satisfactory results. Here we present the experiences.

Methods Fresh alcohol preserved specimens of aquatic insects (Ephemeroptera and Coleoptera) were used for the observations with a Skys Scan 1172 microtomograph. Observations were performed direct fresh: specimen submerged in alcohol inside a Eppendorf microcentrifuge tube (A), or alternatively wrapping the specimen with laboratory parafilm (B). Other observations were performed drying specimens: air dried (C), or dehydrated in alcohol series and finally dried by critical point method (D). Also we tested to stain animals fixed in Karnowvky’s fixative and stained with osmium tetroxide (E) as described by Uzun et al. (2007). Samples were observed sticking them onto the SkyScan samples holder by using plastilin. No filter was used to scan images.

Results No matter what method were used when samples where scanned in liquid Independently if they were stained or not, we always obtained very poor contrasted images to be able to observe the soft internal organs. In Fig.1 an Ephemeroptera nymph (Baetis rhodani) preserved in 70% alcohol, and observed with method B. Poor images of internal anatomy were obtained, only the gut content is conspicuous because its high contrasted mineral sand content.

Figure 1: Maximal Intensity Projection image of a alcohol preserved and wet scanned mayfly nymph (Baetis rhodani). (Source voltage: 80kv, Source Current: 100 µA, image pixel size: 3.63 µm).

In contrast, the reconstructions after getting images from dried specimens gave impressive and good results permitting to get accurate images of details and internal organs.

Fig. 2: 3D reconstructions of the penis of an air dried adult of the Ephemeroptera Ecdyonurus venosus: dorsal (a), ventral(b), lateral (c, d), apical (f) and section (e). (Source voltage: 49kv, Source Current: 100 µA, image pixel size: 2.0 µm).

[Escriba una cita del

documento o del resumen de

un punto interesante. Puede

situar el cuadro de texto en

cualquier lugar del documento.

Utilice la ficha Herramientas

de cuadro de texto para

cambiar el formato del cuadro

de texto de la cita.]

a b

c d

f e

Fig. 3: Maximal Intensity Projection images of a non stained critical point dried specimen of a Coleoptera of the genus Dryops: progressive lateral sections from the internal body cutting the digestive tube (a) until the external body wall (d) and a dorso-ventral view section (e) (Source voltage: 71kv, Source Current: 80 µA, image pixel size: 2.7 µm).

a

b

c

d

e e

Reconstructions of the penis of an adult of the Ephemeroptera species Ecdyonurus venosus, coming from an air dried preserved specimens (method C) of an old collection. Images were quite good and similar of what can be obtained with a scanning electron microscopy (SEM) technique, with the additional advantage of the possibility to make cuts to evidence the internal anatomy (Fig. 2). Extremely good results were obtained drying specimens by critical point (method D), as can be observed in figure 3, obtained from a Coleoptera of genus Dryops, from a collection preserved in alcohol. In fact by using a more complicated protocol, staining the specimens (methods E) similar results were obtained. In figure 3 a section of the nymph of the Ephemeroptera species Prosopistoma pennigerum was obtained were it is possible to see clearly the internal muscular system and organs, as well.

Fig. 3: Maximal Intensity Projection images of a stained critical point dried specimen of a nymph of the Ephemeroptera species Prosopistoma pennigerum (Source

voltage: 47 kv, Source Current: 100 µA, image pixel size: 1.82 µm).

Conclusion Up to now the result that we obtained by direct capture images of wet fresh preserved animal were not so good or even discouraging, in contrast with the very detailed contrasted conspicuous images obtained from dried specimens. Three dimensional reconstructions are impressive and in many cases quite similar to the results that can be get by SEM. However our experiences with the Maximal Intensity Projection utility of the CTan software results a very powerful tool to visualize the internal anatomy of the aquatic insects samples studied. It permits to dissect the animal from the external body wall to the more internal parts, obtaining impressive realistic images permitting to study details and perspectives never seen before. After failures getting good results with observations of wet samples of insect, it is clear that still it is needed a lot of work to evaluate existing fixation and staining methods, or to to develop new ones. In this sense is quite promise the recent paper by Metscher (2009).

References: 1. Uzun, H., Curthoys, I.S., Jones, A.S. “A new approach to visualizing the

membranous structures of the inner ear – high resolution X-ray micro-tomography.” Acta Oto-Laryngol. 127 (6), 568–573. 2007.

2. Metscher, B.D. “MicroCT for comparative morphology: simple staining methods allow high-contrast 3D imaging of diverse non-mineralized animal tissues”. BMC Physiology, 9:11 doi:10.1186/1472-6793-9-11. 2009.

Subregional bone mineral density and bone volume fraction in whole human vertebrae:

concurrent analysis using DXA and micro-CT

E.Perilli1,2, A.M. Briggs3, J.D. Wark4, S. Kantor4, I.H. Parkinson1,2, N.L. Fazzalari1,2

1 Bone and Joint Research Laboratory, SA Pathology and Hanson Institute, Adelaide, South Australia, Australia, egon.perilli@health.sa.gov.au 2 Discipline of Anatomy and Pathology, University of Adelaide, Adelaide, South Australia, Australia 3 School of Physiotherapy, Faculty of Health Sciences, Curtin University of Technology, Perth, Western Australia, Australia 4 Department of Medicine, University of Melbourne, Bone & Mineral Service, Royal Melbourne Hospital, Melbourne, Victoria, Australia

Aims Vertebral fracture risk of patients is usually evaluated using dual-energy X-ray absorptiometry (DXA) of the lumbar spine, together with clinical factors. Posterior-Anterior (PA) projections are usually performed, with areal bone mineral density (BMD) measured within the whole vertebral body. However, bone distribution and thus bone strength vary within the vertebra1,2. Subregional BMD measurements using lateral-projection DXA scanning modality might be more informative about vertebral fragility. Nowadays, micro-computed tomography (micro-CT) allows three-dimensional structural characterization of entire bone segments, non-destructively and at high resolution3. To assess the capability of lateral-projection DXA to determine subregional variations in bone distribution, this study examined human vertebrae first by DXA and then by micro-CT.

Method Eight human cadaver spines were examined (mean age at death 78±10 years). These were immersed in a water bath and scanned by DXA in PA and in lateral projections (densitometer Hologic QDR4500, Hologic, Waltham, MA, USA). Subregional BMD analysis was performed in lateral projection of the L2 vertebrae, with the examination area divided via software in three subregions of interest (superior, central, inferior). The L2 vertebrae were then dissected and entirely scanned by micro-CT (17.4 μm pixel size, Skyscan 1076, Skyscan Kontich, Belgium). The micro-CT volume of interest comprised the trabecular bone of the entire vertebral body3. It was divided via software into three analogous subregions with equal height (superior, central, inferior), from which bone volume fraction (BV/TV) was assessed (software CTAnalyser, Skyscan).

Results For both DXA and micro-CT, the values of the whole vertebral body showed statistically significant differences compared to the subregions (p<0.01, one-way ANOVA for repeated measures, fig. 1 and 2). Statistically significant differences between the subregions were found for both BMD and BV/TV, with the inferior subregion having higher values than the superior subregion (p<0.05, paired t-test, fig.1 and 2).

The BMD measured by lateral-projection DXA over the whole vertebrae was significantly related to the total BV/TV measured by micro-CT (R2=0.59, p<0.05, fig. 3), whereas BMD measured in PA-projection DXA was not (R2=0.16, p=0.33, fig.3).

Superior subregion

Central subregion

Inferior

subregion

Lateral view

DXA analysis:

Figure 1: Schematic diagram of the three vertebral subregions, and bar plot of BMD measured by lateral DXA in these subregions (mean ± SD for each subregion). *Stat. sign. differences between all subregions (p<0.05, paired t-test). **Stat. sign. differences between inferior and central subregion, and whole vertebral body (p<0.05, paired t-test).

Superior

subregion Central

subregion Inferior

subregion

Coronal micro-CT cross-section

Axial micro-CT cross-section

Micro-CT analysis:

Figure 2: Micro-CT images a human vertebral body (17.4 μm pixel size), and bar plot of BV/TV measured in the three vertebral subregions (mean ± SD for each subregion). *Stat. sign. differences between inferior and central subregion (p<0.05, paired t-test). **Stat. sign. differences between inferior and superior subregion, and whole vertebral body (p<0.05, paired t-test).

Figure 3: Scatter plot of total “BV/TV vs. BMD”. (Left) BMD measured in lateral projection, (Right) BMD measured in posterior-anterior (PA) projection. While total BMD measured by lateral-projection DXA was significantly related to the total BV/TV, the BMD measured in the PA projection was not.

In the central subregion, the linear regression 'BV/TV vs. BMD' had the highest coefficient of determination (R2 =0.80, p<0.01), compared to the inferior and superior subregion (figure 4). The finding of BMD being lowest in the central subregion (fig. 1), whereas BV/TV being lowest in the superior subregion (fig. 2), can be explained by the inclusion of the vertebral endplates in DXA analysis. In the central subregion, which does not contain the endplates, the regression “BV/TV vs. BMD” shows the highest regression coefficient (figure 4, left).

Conclusion Differences in the bone parameters between the whole vertebra and the subregions were found by both lateral-projection DXA and micro-CT, both techniques showing higher values in the inferior compared to the superior subregions. This study shows that, in contrast to BMD assessed using PA-projection DXA scanning, measurements using lateral-projection DXA in the L2 vertebra are significantly related to BV/TV assessed via micro-CT. In particular, subregional BMD measurements are highly related to trabecular bone volume fraction in the central part of the vertebral body. These findings support lateral-projection DXA examination as a valuable modality for improving the evaluation of vertebral fragility.

References: 1. Briggs AM, Greig AM, Wark JD, "The vertebral fracture cascade in

osteoporosis: a review of aetiopathogenesis“ Osteoporos Int, 18(5), 575-584, 2007

2. Gong H, Zhang M, Yeung HY, Qin L, “Regional variations in microstructural properties of vertebral trabeculae with aging" J Bone Miner Metab, 23, 174-180, 2005

3. Briggs AM, Perilli E, Parkinson IH, Wrigley TV, Fazzalari NL, Kantor S, Wark JD, “Novel assessment of subregional Bone Mineral Density using DXA and pQCT and subregional microarchitecture using Micro-CT in whole human vertebrae: applications, methods, and correspondence between technologies” J Clin Densitom, 13(2):161-174, 2010

Figure 4: (Left) Scatter plot of “BV/TV vs. BMD” measured in the three vertebral subregions. The central subregion showed the highest determination coefficient in the regression (R

2=0.80, p<0.01). (Right) 3D

reconstruction of the trabecular bone volume of a human vertebral body obtained by micro-CT, subdivided into 3 subregions (voxel size 17.4 µm).

3D micro-CT

X-Ray Micro-CT Investigation of Sandstones with SkyScan 1172

D.A. Koroteev, A.N. Nadeev

Schlumberger Technology Company, 101000, 5a, Ogorodnaya Sloboda lane, Moscow, Russia, dkoroteev@slb.com

Aims We studied different regimes for scanning rock samples at the highest resolution using the SkyScan 1172 tomograph to find optimal conditions with respect to 3D

image quality and scanning duration. Critical characteristics of the reconstructed

images include high contrast of the grain-pore boundary and the boundary between

grains with different X-ray absorption coefficient. The ideal case is when a histogram

of a reconstructed 3D image represents a set of separate peaks responsible for pore space and grains of different density and element composition. This allows us to create a segmented digital 3D model which would reproduce the original sample with maximal accuracy.

Method We used the SkyScan 1172 X-ray microtomograph (version A) with a camera mode of the highest resolution (4,000×2,096 pixel2). The X-ray tube voltage varied from 50 kV to 100 kV. The tube’s power is constant and equals 10 W. Usually, we use a 3600 scan with a rotation step of 0.30 and an average within four frames. We performed scans with built-in Al and Al+Cu filters and a 0.125-mm Cu filter. Each time, the flat field correction was applied prior to scanning. A reconstruction of back-projection images is performed by NRecon shell with InstaRecon software. To process the reconstructed 3D images we used CTan and AvizoFire software. And for visualization purposes we used CTvol, CTvox, and AvizoFire technology.

Results We performed a series of scans on a natural sandstone sample of cylindrical form with an 8-mm diameter. Resolution was 2.2 um/pixel. Porosity of the sample was measured by means of classical porosimetry. The experimental value of the porosity is used as a reference point for choosing the most adequate threshold value to split pores from grains during the binarization procedure. To make images from different configurations comparable, we converted them from a 16-bit TIF format to an 8-bit JPG format with the same gray-scale range; see Fig. 1 for cropped examples. The images at 50 kV appear darker, but this does not affect the high contrast value. In Fig. 2 we provide the histograms of the uncropped images. Peaks on the histograms correspond to air (left) and grains (right). There is quite a significant difference in contrast between the first and second peaks for all cases. Al and Al+Cu filter cases at 50 kV demonstrate a strong dip between the peaks, while for the 100-kV cases the dip is much smoother. The gray-scale value corresponding to the dip might be considered as a threshold value to separate pores from grains. This characterizes 50 kV cases as most useful for binarization purposes. However,

even zero-filter scans might be useful, as they result in quite clear separation in histogram peaks and are about 2–4 times faster than scans with filters. Porosimetry data is applicable here for correcting the threshold value. But in the no-filter scans, the outer layer of the sample should be removed from the consideration due to noncorrectible beam hardening in this zone. If we zoom in to the right ends of the right peaks to investigate the gray-scale range responsible for small, bright areas (Fig. 1), we see that all cases are equivalent and the histogram features do not allow us to determine the value for further segmentation directly.

50 kV, no filter. 100 kV, no filter. 50 kV, Al filter.

100 kV, Al filter. 50 kV, AlCu filter. 100 kV, AlCu filter.

Figure 1: Reconstructed slices of sandstone sample at different acquisition regimes.

Figure 2: Histograms. Figure 3: Sandstone saturated with

clay. Upper highlighted pore filled with clay, lower one is empty.

We also performed scans of sandstones partially saturated with clay which has similar absorption properties to the quartz grains. But the shape and porous microstructure of the clay allows us to split it from the quarts grains, as shown in Fig. 3.

Another interesting issue discovered during the investigation of sandstones is anisotropy in the binarized model. Using CTan functionality we have noted that binarized models with open porosity corresponding to measured value, demonstrate quite noticeable anisotropy in the direction colinear with the axis of the sample rotation. To investigate the nature of this phenomenon we prepared a cubic sandstone sample and performed three scans at three orthogonal orientations. Stereology analysis of different spherical volumes of interest for all three cases delivered anisotropy mainly along the vector colinear to the rotation axis, and the degree of anisotropy was about 1.6. With help from the SkyScan researchers we found an artificial way to exclude the anisotropy by applying a smoothing procedure for 2D slices before thresholding, although this operation reduces the adequacy of the reconstructed porous space with respect to the original one.

Conclusion We carried out a series of XmCT experiments with sandstone samples. The main issues to conclude are

the lower the voltage and the stronger the filter, the better the contrast achieved between the grains and pores

in case of time limitations, scans with no filter are quite reasonable even for high-density materials such as sandstone. In this case, the outer layers of the sample should be removed from consideration

there is no characteristic feature on the histogram allowing us to separate dominating quarts grains from others automatically

some substances, such as clay, might be separated from grains by its characteristic morphological structure

quite noticeable artificial anisotropy might be introduced at the scanning or reconstruction stage. To overcome this problem, presmoothing should be applied to gray-scale images before binarization.

INVESTIGATION OF THE INFLUENCE OF SURFACE ROUGHNESS

MODIFICATION OF BONE TISSUE ENGINEERING SCAFFOLDS ON THE MORPHOLOGY AND MECHANICAL PROPERTIES

G. Kerckhofs1,2, G. Pyka1,2, S. Van Bael2,3, J. Schrooten1,2 and M. Wevers1

1 Department of Metallurgy and Materials Engineering, Katholieke Universiteit Leuven, B-3001 Leuven, Belgium; greet.kerckhofs@mtm.kuleuven.be,

gregory.pyka@mtm.kuleuven.be, jan.schrooten@mtm.kuleuven.be, martine.wevers@mtm.kuleuven.be

2 Prometheus, Division of Skeletal Tissue Engineering, Katholieke Universiteit Leuven, B-3000 Leuven, Belgium

3 Department of Mechanical Engineering, Division of Production engineering, Machine design and Automation, Katholieke Universiteit Leuven,B-3001 Leuven,

Belgium; simon.vanbael@mech.kuleuven.be

Aims

Bone tissue engineering (TE) is a multidisciplinary field of science focusing on healing large bone defects by designing and manufacturing constructs that combine open porous biomaterials (= scaffolds) with osteogenic cells to support cell seeding and in vivo bone formation [1]. In TE, the tendency is to evolve from the use of open porous foams with a random structure to scaffolds with a complex, but highly controllable designed morphology useful for the production of a new generation of bone implants [2]. Selective laser melting (SLM), a relatively young rapid prototyping (RP) technique, offers the opportunity to produce micro-porous structures with global morphological properties that are not random, but highly controlled through robust computer design [3,4]. Achieving controlled surface properties is also essential in the design and production of biocompatible scaffolds [2,5], since the strut surface roughness (SSR) influences cell behaviour within a scaffold [5]. Despite the advantage of SLM to allow a high control of the morphology at the mesoscale, at this moment functional constraints caused by working close to the technical limits of the production device prevent production of 3D porous scaffolds with a desired and controlled surface morphology at a cell-relevant level (microscale). Therefore, a modification of the as-produced SSR is needed to support the desired cell response. It is obvious that any surface roughness modification performed after production will change the topography of the struts surface as well as the local and global mechanical properties of the structure. Therefore, in the present research the influence of the applied struts surface roughness modification (SSRM) procedures on the morphology (both meso- and microscale) and the mechanical behaviour of the scaffolds has been determined. Materials and methods

In this study, the struts surface of open porous Ti6Al4V scaffolds produced by SLM has been modified by chemical and electrochemical polishing. Scaffolds were designed by creating a CAD design (fig. 1a) using Magics software [Materialise NV, Haasrode, Belgium] and were produced on a non-commercial SLM machine equipped with a Yb:YAG fibre laser with a beam spot size of 80 µm and a maximum power of 300 W on the powder bed. The unit cell applied for the design (fid. 1b) had a diamond shape-like structure (fig.1a, 1c and 1d). The porous samples were built, in a closed and argon flushed chamber, layer-by-layer using a metal powder layer thickness of 30 µm. The diameter and height of the samples presented in fig.1a, 1c

and 1d was 6.0 ± 0.5 mm and 12.0 ± 0.5 mm respectively. The designed strut size was 100 µm and the pore size 1 mm. SEM images (fig. 1e) show a large and highly inhomogeneous roughness caused by non–melted powder grains attached to the strut surface on the as-produced Ti6Al4V scaffolds.

a) b) c) d) e)

Figure 1. Examples of the Ti6Al4V porous scaffolds: a) CAD design of the scaffolds, b) the unit cell, c) cylindrical sample (side view) and d) cylindrical sample (top view) and e) SEM pictures of the unit cell showing highly inhomogeneous roughness caused by non–melted powder grains attached to the strut

surface . Scale bars = 2 mm.

An appropriate roughness reduction procedure was applied by combining chemical and/or electrochemical polishing of the scaffolds, which apart from removing the inhomogeneities of the struts, allows to obtain a cell-friendly strut topology. In a first step, the produced samples (fig. 2a) were polished chemically (fig. 2b) in order to remove the attached non-melted powder grains. During chemical polishing samples were immersed for 10 minutes in a chemical solution with the following composition: HF + H2O (hydrofluoric acid, HF, Riedel-de Haën, Germany, p.a. 48%). In a second step, electrochemical polishing was applied to obtain the desired surface morphology (fig. 2c). For electrolytic polishing, the electrolyte had the following composition: CH3COOH + H2SO4 + HF (hydrofluoric acid HF, Riedel-de Haën, Germany, p.a. 48%; acetic acid CH3COOH, Acros Organics, Belgium, p.a. glacial; sulfuric acid H2SO4, Fisher Scientific, United Kingdom, p.a. >95%). The combination of chemical and electrochemical polishing gives the opportunity to modify the strut surface in a controlled way.

a) b) c) Figure 2. SEM images of the: a) raw strut with attached powder grains, b) strut after chemical polishing,

c) strut after chemical and electrochemical polishing,

For determining the influence of the SSRM procedure on the struts roughness, a non-destructive SEM image-based measurement protocol was applied. Based on the profile lines on the surfaces of the struts, the following roughness parameters have been determined:

- the arithmetic average deviation:…………………………………..…..

n

i

ia yn

R1

1

- the root mean square deviation of the roughness profile from the mean line:

…………………………………………………………………………...

n

i

iq yn

R1

21

- difference between highest peak and deepest valley:.……………… VPT RRR

where n = number of data points in X direction, y = the surface height relative to the mean plane, RP = the highest point and RV = the lowest points in the evaluation length.

The quantification of the scaffold morphology as-produced and after surface roughness modification was done by micro-CT, determining the influence of the surface roughness modification on the porosity, the available surface, the specific surface and the strut thickness. For this purpose, the Philips HOMX 161 x-ray system with AEA tomahawk CT software was used. The applied acquisition parameters are presented in table 1. The morphological parameters were determined using commercially available image analysis software, namely CTAn (SkyScan NV, Kontich, Belgium) [3].

Table 1. Micro-CT acquisition parameters used for imaging the Ti6Al4V SLM scaffolds

Voltage Current Filter material Voxel size

Rotation step, angle

Frame averaging

90 kV 0.39 mA 1 mm Al 12.6 µm 0.5° over 187° 32 frames

The characterisation of the mechanical behaviour of the tested micro-porous structures prior to and after surface roughness modification is performed by continuous mechanical compressive loading, determining the E-modules, the strength and the strain at maximum stress. The samples were placed on an in house developed in-situ loading stage with maximum available load 3kN [3]. As a first step, a preload of 0.01kN was applied and afterwards compression at a constant rate of 0.2mm/min was maintained until final failure. Obtained load and displacement data were used to analyse the mechanical properties of the as-produced and surface modified scaffolds.

The main goal of all these experiments is the characterisation of the 3D SLM porous structures prior to and after surface roughness modification in order to optimise the design, the modeling, the production of the porous structures and the surface roughness modification and hence to improve the properties of the scaffolds.

Results

- Surface roughness measurement

Quantitative analysis of the surface roughness in function of the applied surface roughness modification was done on the basis of SEM images. Obtained profile lines of the strut surfaces determined on the basis of the pixels distribution in the SEM images are presented in figure 3. It can be seen that the profiles clearly reflect the strut surface topology and can be used for determining the surface roughness of the examined samples.

(a) (b) Figure 3. SEM images of a typical strut of the porous Ti6Al4V scaffolds with fitted profile lines

generated during analysis: (a) prior to and (b) after surface roughness modification

As can be seen in fig. 3a, SEM investigation revealed a high and inhomogeneous roughness of the strut surface especially at the bottom of the struts, which was caused by attached non-melted powder grains. In order to characterise inhomogeneities of the scaffold strut surfaces, the roughness of the top and bottom of the strut were analysed separately. Obtained results (fig. 4) showed a difference between the roughness of the top and bottom surface of the strut for the samples prior to surface roughness modification. The average as-produced scaffold roughness (Ra) ranged between 9 and 13 µm for both the top and bottom side of the struts, but the difference between highest peak and deepest valley (Rt) for the top and bottom side of the struts was large (36 µm).

Figure 4. Strut surface roughness data of the Ti6Al4V scaffolds prior to and after surface roughness modification.

The surface roughness of the scaffolds after polishing showed a smaller standard deviation, especially for Rt, which allowed to assume that the scaffold morphology after surface roughness modification was more homogenous. A higher strut roughness reduction at the bottom side can be explained by the different current density distribution present during electrochemical polishing, which caused a higher dissolution rate of the rougher surface areas. Assessment of the sample volume reduction in function of the applied surface roughness modification was done on the basis of measurements of the changes in sample weight, as shown in Figure 5. - Morphological characterization

Micro-CT based characterisation of the Ti6Al4V scaffolds prior to and after surface roughness modification was done in order to determine the influence of the surface roughness modification on the global scaffold morphology. Changes in porosity, average strut thickness and sample volume as well as strut thickness distribution were investigated (table 2 and figure 6).

Table 2. Morphological characteristics of the scaffolds prior to and after surface roughness modification detrmined on the basis of micro-CT based image analysis.

Tested Sample

Sample volume Porosity Avg. strut thickness

micro-CT based characterisation

[mm3] [%] [µm]

As-produced 57.47 ± 1.31 86.31 ± 0.16 213.81 ± 0.57

Surface modified 33.40 ± 2.19 92.03 ± 0.74 169.05 ± 7.58

Figure 6 shows changes in strut thickness distribution of the surface modified samples. Micro-CT image based investigation provides the possibility to analyse, visualise and quantify (2D and 3D) the changes in strut surface morphology in a non destructive way.

- Mechanical characterization

In order to determine the influence of the surface roughness modification procedure of the Ti6Al4V scaffolds on the mechanical properties, compression tests of the as-produced and surface modified samples was performed (table 3). It can be clearly seen that the applied surface roughness modification introduced changes in the mechanical behaviour of the porous structures. In figure 6, it can be seen that the applied surface roughness modification

Figure 6. Strut thickness distribution prior to and after surface modification

Figure 5. Average sample volume reduction in function of the applied surface modification

leads to a more smooth strut surface, but it also decreases the strut diameter, thus also significantly decreasing the mechanical properties. This is caused by the changed scaffold unit cell dimensions due to the reduction in strut size and the increase in the pore size during polishing. Since pores can be described as spaces between struts of the scaffold, changes of the pore size, revealed by micro-CT characterisation, are caused automatically by reduction of the strut thickness. To compensate for this strut thickness reduction related loss in mechanical strength, the effective strut thickness that determines the mechanical properties of the scaffolds should be accounted for both in the design and SLM-production to ensure desired mechanical properties after controlled surface roughness modification.

Table 3. Mechanical properties of the scaffolds prior to and after surface roughness modification

Conclusion

This study showed that thorough characterisation of the changes in scaffold morphological and mechanical properties due to surface roughness modification can be easily obtained by the SEM and micro-CT image-based analysis combined with mechanical testing. The 2D SEM image-based protocol can be applied to determine the strut surface morphology and can become a valuable tool for determining the roughness of complex porous structures. This will result in an optimisation of the design, the production and surface roughness modification protocols related to obtaining controlled morphological and mechanical properties of porous (metallic) materials. In a next step in vitro biological experiments are needed to evaluate the relation between these properties and cell behaviour.

Compression tests, performed on the as-produced and surface modified scaffolds, showed changes in mechanical properties as function of the applied surface roughness modification. Analysis of the mechanical properties combined with micro-CT based characterisation of the scaffolds, when related to the applied surface roughness modification, can be used for the optimisation of the design, the modelling and the production and hence to improve the properties of scaffolds that can be applied for bone regeneration. References [1] Lenas, P., et al. Developmental engineering: A new paradigm for the design and

manufacturing of the cell-based products. Part I: from three-dimensional cell growth to biomimetics of in vivo development. Tissue Engineering: Part B, 15 (2009)

[2] Karageorgiou V. et al. Porosity of 3D biomaterial scaffolds and osteogenesis. Biomaterials, 26 (2005) 5474–5491.

[3] Kerckhofs, G., et al. Mechanical characterization of porous structures by the combined use of micro-CT and in-situ loading. World Conference on Non-Destructive Testing 2008. Shanghai, 25-28 October 2008 WCNDT 2008 Proceedings Book.

[4] Van Bael, S., et all. Morphological and mechanical characterization of Ti6Al4V scaffolds produced with Selective Laser Melting. 1st International Conference on Tissue Engineering (ICTE). Leira - Portugal, 9-12 July 2009.

[5] Ponsonnet L. et al. Effect of surface topography and chemistry on adhesion, orientation and growth of fibroblasts on nickel–titanium substrates. Materials Science and Engineering C, 21 (2002) 157–165

Tested Sample Strain at max strength Strength Stiffness

[%] [MPa] [MPa]

As-produced 6.04 ± 0.32 13.00 ± 0.62 397.07 ± 29.95

Surface modified 7.02 ± 0.24 7.41 ± 0.88 226.15 ± 22.45

Anisotropy of Sandstones

W.-A. Kahl1, R. Hinkes1, V. Feeser1, A. Holzheid1

1 Institute of Geosciences, Christian-Albrechts-University Kiel, Ludewig-Meyn-Straße 10, D-24118 Kiel, Germany, wak@min.uni-kiel.de

Aims Sandstone aquifers are considered to be potential reservoirs for the geological sequestration of industrial carbon dioxide. Porosity and permeabilty of the sandstone as well as textural and compositional anisotropy of the grain phase are important boundary conditions during and after injection of the CO2. Petrophysical laboratory studies of mechanical and seismic bulk properties revealed their strong dependence upon orientation. To correlate those bulk properties with the microstructure, we attempt to locate the anisotropic directions in the reconstruction volume of the X-ray micro-computed tomography scans.

Method Aliquots of two Lower Cretaceous sandstones (localities: Bad Bentheim and Obernkirchen, both Germany), have been studied in a 3D multianvil pressure apparatus for cubic samples [1]. The samples were subjected to isotropic stresses up to 100 MPa at a temperature of 20 °C to map the basic relationships between seismic behaviour (P wave velocity) and mechanical behaviour (stress-strain relation). Subsequent to the geomechanical studies oriented 10 mm cores were drilled from the samples and scanned (at ambient pressure) using the SkyScan 1172 device of the experimental and theoretical petrology group at Kiel University with a beam energy of 100 kV, a flux of 100 µA and Al/CU foil with a resolution of 6.76 µm. Three dimensional stereology image analysis and volume rendering were done using the SkyScan software CT-Analyser.

Results Cretaceous sandstones that have been investigated in a high-pressure triaxial testing facility for cubic samples under controlled loading and unloading stress paths at a temperature of 20 °C and a maximum pressure of 100 MPa showed marked differences of the observed ultrasonic sound (frequency of Signal input 1MHz) velocities in three directions (Fig. 1 A and B). In both samples the lowest velocities (Z) are found in the direction perpendicular to the bedding. Obernkirchen sandstone shows higher velocities in all directions than Bentheim sandstone, with the highest (X) and intermediate (Y) values being almost similar, whereas the fastest (X) and the intermediate (Y) velocities in Bentheim sandstone exhibit a difference. In microstructural context, the mean intercept length (MIL) analysis is a measuring

tool for the isotropy of a structure. Mean intercept length is found by sending a grid of test lines over a large number of 3D angles through a spherical 3D image volume containing binarised objects, and dividing the length of the test line through the analysed volume by the number of times that the line passes through or intercepts part of the solid phase (see SkyScan-CTanalyser_parameters.pdf for more detailed information). This method will give an accurate result if analysing a volume containing a sufficiently large number of objects, which might be the case for

the VOI of the reconstructed sandstones: the sphere diameter in µm is the 59-fold (Obernkirchen) and the 38-fold (Bentheim), respectively, of the mean structural

thickness of the grain phase. An ellipsoid is fitted to the 3D distribution of MILs measured over the full range of 3D stereo-angles. This ellipsoid has 3 vectors which are orthogonal and describe the longest orientation (E1), and the length (E2) and width (E3) of the ellipse section at right-angles to the longest orientation.

To link the bulk seismic behaviour to microstructural aspects, we performed MIL analysis on cubic VOIs of 6 mm edge length (see Table 1) based on the grain phase as well as on the pore space. The results of the grain phase based MIL analysis are linked unexpectedly with the bulk seismic properties. The direction of the largest MIL in both oriented drill core samples are parallel to the slowest observed seismic P wave velocities in Z (Obernkirchen MIL || E1: 0.5763 mm, Bentheim MIL || E1: 0.5568). The high velocities in Obernkirchen X and Y are oriented parallel to the short MILs || E3 (0.4410 mm) and || E2 (0.4577 mm). The overall slower P wave velocities in Bentheim X (MIL || E3 0.4491 mm) and Y (MIL || E2 0.4711 mm) are at larger values than the Obernkirchen E3 and E2, respectively, whereas the slowest seismic velocity Bentheim Z has a smaller value than Obernkirchen Z. Table 1. Directions of seismic P wave velocity and MIL analysis results in sandstone.

Sandstone sample locality characteristic P wave velocity

Mean Intercept Length (mm) grain phase ; pore space based MIL analysis pore space MIL ellipsoid vector

Obernkirchen Z, slowest

0.5763 | 0.0740 E-vector 1, largest

Obernkirchen Y, intermediate

0.4577 | 0.0576 E-vector 2, intermediate

Obernkirchen X, fastest

0.4410 | 0.0554 E-vector 3, shortest

Bad Bentheim Z, slowest

0.5568 | 0.1197 E-vector 1, largest

Bad Bentheim Y, intermediate

0.4711 | 0.1008 E-vector 2, intermediate

Bad Bentheim X, fastest

0.4491 | 0.0959 E-vector 3, shortest

The results of the pore space based MIL analysis correlate well (inversely) with the bulk seismic properties. The direction of the largest MIL in both oriented drill core samples are parallel to the slowest observed seismic P wave velocities in Z (Obernkirchen MIL || E1: 0.0740 mm, Bentheim MIL || E1: 0.1197). All three MILs of the Obernkirchen sandstone (0.0554..0.0740), which has higher velocities in all directions than the Bentheim sample, are smaller than the MILs observed for the latter (0.0959..0.1197). The almost identical high velocities in Obernkirchen X and Y correlate with the small and similar short MILs || E3 (0.0554 mm) and || E2 (0.0576 mm) for Obernkirchen. The overall slower but also more different P wave velocities in Bentheim X and Y are met by the MIL || E3 (0.0959 mm) and MIL || E2 (0.1008) values, which are more diverse and at larger values than the Obernkirchen E3 and E2.

E

1 cm

D

1 cm

C

Ok X

(E3)

Ok Z

(E1 up)

Ok Y

(E2)

Be X

(E3)

Be Z

(E1 up)

Be Y

(E2)

E1

E3

E2

Ok

Be

0 30 60 90

3.6

3.8

4.0

4.2

Vp

(km

/s)

stress (MPa)

Loading

Ok2.2z

Ok2.2y

Ok2.2x

Unloading

Ok2.2z

Ok2.2y

Ok2.2x

Obernkirchen

(Ok)

0 30 60 903.2

3.4

3.6

3.8

Vp (km

/s)

stress (MPa)

Loading

Be2.2z

Be2.2y

Be2.2x

Unloading

Be2.2z

Be2.2y

Be2.2x

Bentheim

(Be)

E

1 cm

D

1 cm

C

Ok X

(E3)

Ok Z

(E1 up)

Ok Y

(E2)

Be X

(E3)

Be Z

(E1 up)

Be Y

(E2)

E1

E3

E2

Ok

Be

0 30 60 90

3.6

3.8

4.0

4.2

Vp

(km

/s)

stress (MPa)

Loading

Ok2.2z

Ok2.2y

Ok2.2x

Unloading

Ok2.2z

Ok2.2y

Ok2.2x

Obernkirchen

(Ok)

0 30 60 90

3.6

3.8

4.0

4.2

Vp

(km

/s)

stress (MPa)

Loading

Ok2.2z

Ok2.2y

Ok2.2x

Unloading

Ok2.2z

Ok2.2y

Ok2.2x

Obernkirchen

(Ok)

0 30 60 903.2

3.4

3.6

3.8

Vp (km

/s)

stress (MPa)

Loading

Be2.2z

Be2.2y

Be2.2x

Unloading

Be2.2z

Be2.2y

Be2.2x

Bentheim

(Be)

0 30 60 903.2

3.4

3.6

3.8

Vp (km

/s)

stress (MPa)

Loading

Be2.2z

Be2.2y

Be2.2x

Unloading

Be2.2z

Be2.2y

Be2.2x

Bentheim

(Be)

Figure 1: Bulk seismic anisotropy and microstructural anisotropy (directions of stereological MIL analysis) in two Cretaceous sandstones. Anisotropy of P wave

velocities (A) in Obernkirchen sandstone and (B) in Bad Bentheim sandstone with velocity in Y direction being the fastest and in Z the slowest. (C) The Eigen vectors of the pore space MIL ellipsoids plotted in a stereographic projection (using Stereo32 1.0.1 by K. Röller and C. A. Trepmann). Eigenvector E3 denotes the shortest mean intercept length, E1 the largest. Orientations of the Eigen vectors shown in the reconstruction volume of (D) Obernkirchen sandstone and (E) Bad Bentheim sandstone. The orientation of the slowest seismic veloctity, the Z axis, in the oriented drill cores is upward, which is the orientation of the direction with the largest pore space MIL. The directions of the E-vectors of the pore space based MIL analysis that describe the 3D variation of the mean intercept length of the pore space within the VOI are plotted in a stereographic projection (Fig. 1C). The orientation of the main directions of MIL variation is also visualized in two reconstruction images of Obernkirchen and Bentheim sandstone (Fig. 1 D and E). Both images are taken from the level that contained the origin of the spherical 3D volume, which was used for the mean intercept length analysis.

Conclusion Two oriented cores of sandstone samples that previously have been used in petrophysical laboratory experiments to study ultrasonic sound velocities were investigated using X-ray micro-computed tomography. It was possible to correlate the main orientations as well as the relative amounts of bulk anisotropic sound velocities (frequency of Signal input 1MHz) and mechanical behaviour to the main directions of microstructural anisotropy of the pore space in the reconstruction volumes. On this basis we will start to investigate the effect of the connectivity of the grain frame work on mechanical strength. Furthermore, the linkage between bulk seismic behaviour and the rock microstructure may allow the assessment of seismic tortuosity.

References: 1. Kern, H.; Ivankina, T. I.; Nikitin, A. N.; Lokajíček, T.; Pros, Z. "The effect of

oriented microcracks and crystallographic and shape preferred orientation on bulk elastic anisotropy of a foliated biotite gneiss from Outokumpu." - Tectonophysics 457, 143-149 (2008)

The High Porosity of Strontium-containing Cement May Lead to Early Migration of Total

Hip Arthroplasty: A micro-CT study

William Weijia Lu 1, Ting Wang 1, Chun-yi Wen 1, Fu-yuen Ng 2, Man-kit Fong 1, Wing-

moon Lam 1, Kwong-yuen Chiu 1, 2

1 Department of Orthopaedics and Traumatology, Li Ka Shing Faculty of Medicine, The University of Hong Kong. & 2. Department of Orthopaedics and Traumatology,

Queen Mary Hospital, Hong Kong SAR, China, wwlu@hkusua.hku.hk

Aims Strontium-containing hydroxapatite (Sr-HA) cement is a newly developed bioactive bone cement for biological fixation of cemented prosthesis in total hip arthroplasty (THA). The advantages of such bioactive bone cement for strengthening the bonding at the bone-cement interface have been well established. This study aimed to observe the porosity of Sr-HA Cement and its relationship to the migration of prosthesis in THA using a micro-CT system.

Method The prostheses were implanted in swine bones and fixed by Sr-HA or PMMA respectively with six for each group for comparison. The biomechanical evaluation, including pull-out and fatigue loading were conducted with MTS system. The continuous micro-CT scans were taken with the pixel size at 35μm (Skyscan 1076, Belgium). The analysis was conducted on the porosity of the bone cements and its contact surface with the prostheses. Statistical analysis The comparisons were performed using student t tests with p value set at 0.05.

Results There was no significant difference between two groups in mechanical properties under both pull-out test and fatigue loading test. It was revealed under micro-CT that the porosity of Sr-HA cement was significantly higher than PMMA (Sr-HA: 7.5±0.3% vs. PMMA: 4.1±0.2%, p<0.01, see Fig. 1). The closed porosity of Sr-HA cement were much more than PMMA (Sr-HA: 5.2±0.2% vs. PMMA: 1.5±0.1%, p<0.001). There was no significant difference between them in the open porosity as well as the contact surface (Sr-HA: 2.3±0.3% vs. PMMA: 2.6±0.2%, p>0.05). The pore size of Sr-HA cement significantly decreased after incubation in Ringer’s solution (dry: 138±9 vs. wet 59±10, p<0.001). However, the porosity of wet Sr-HA cement remained higher than wet PMMA (Sr-HA: 5.5±0.4% vs. PMMA: 3.3±0.4%, p<0.001).

Conclusion There was no significant difference from biomechanical pointview between Sr-HA and PMMA. The findings generated from this study suggested that the high porosity of Sr-HA cement, particularly the closed pores, might explain the prosthetic subsidence in THA and Micro-CT is one of the useful tool for similar studies.

Figure 1 3D images of pore structure of Sr-HA and PMMA bone cements.

References:

1. Li, Y.W. et al. J Biomed Mater Res. 2000 2. Lu, W.W. et al. Spine. 2001 3. Kuehn KD et al. Orthop Clin North Am. 2005

Acknowledgements: Hong Kong ITF GHP/009/06 & RGC HKU7147/07E