simultaneous 3d visualization and quantification of murine

Simultaneous 3D Visualization and Quantification of MurineBone and Bone Vasculature Using Micro-ComputedTomography and Vascular ReplicaPHILIPP SCHNEIDER,1 THOMAS KRUCKER,2 ERIC MEYER,3 ALEXANDRA ULMANN-SCHULER,3

BRUNO WEBER,4 MARCO STAMPANONI,5,6AND RALPH MULLER1*

1Institute for Biomechanics, ETH Zurich, Zurich, Switzerland2Global Imaging Group, Novartis Institutes for BioMedical Research, Cambridge, Massachusetts3Institute of Zoology, University of Zurich, Zurich, Switzerland4Institute of Pharmacology and Toxicology, University of Zurich, Zurich, Switzerland5Swiss Light Source (SLS), Paul Scherrer Institut (PSI), Villigen, Switzerland6Institute for Biomedical Engineering, ETH and University of Zurich, Zurich, Switzerland

KEY WORDS mouse; polyurethane; vascular corrosion cast (VCC); vascular contrast perfusion(VCP); barium sulfate; synchrotron radiation; morphometry

ABSTRACT Recent evidence suggests a close functional relationship between osteogenesis andangiogenesis as well as between bone remodeling and bone vascularization. Consequently, there isa need for visual inspection and quantitative analysis of the bone vasculature. We thereforeadapted and implemented two different vascular corrosion casting (VCC) protocols using a polyur-ethane-based casting resin in mice for a true three-dimensional (3D), direct, and simultaneousmeasurement of bone tissue and vascular morphology by micro-computed tomography (lCT). Forassessment of vascular replicas at the level of capillaries, a vascular contrast perfusion (VCP) pro-tocol was devised using a contrast modality based on a barium sulfate suspension in conjunctionwith synchrotron radiation (SR) lCT. The vascular morphology quantified using the VCP protocolwas compared quantitatively with the results of a previously established method, where thevascular network of cortical bone was derived indirectly from cortical porosity. The presented VCCand VCP protocols have the potential of serving as a valuable method for concomitant 3D quantita-tive morphometry of the bone tissue and its vasculature. Microsc. Res. Tech. 72:690–701,2009. VVC 2009 Wiley-Liss, Inc.

INTRODUCTION

Growing evidence suggests that vascular architec-ture within bone- and vascular-derived factors play animportant role regulating bone cell physiology and theconcerted action of the bone cell network (Collin-Osdoby, 1994). In general, bone cell survival dependson the proximity to vessels providing nutrients (Ham,1952). Moreover, interruption of the blood supply canlead to avascular necrosis (Cruess, 1986; Glimcher andKenzora, 1979). In particular, there are many indica-tions for a close relationship between osteogenesis andangiogenesis as well as between bone remodeling andbone vascularization (Colnot, 2005). Recently, the cou-pling observed between bone resorption and bone for-mation during bone remodeling was proposed to bemediated by bone vascularization (Parfitt, 2000). Fur-thermore, vascular endothelial growth factor (VEGF)and angiopoietins were located at modeling and remod-eling sites of growing bones of human neonatal ribs(Horner et al., 2001), and linear relationships betweenbone formation and vascular parameters in rat trabec-ular bone were identified (Barou et al., 2002). It istherefore essential to better understand the anatomicaland functional relationship between bone tissue andthe associated vascularization. However, simultaneous,direct and real three-dimensional (3D) methods for the

quantitative assessment of both bone and vasculatureare rare, especially in small rodents such as mice(Duvall et al., 2004). The lack of spatial resolution ofcurrent methods typically does not give access to thecapillary and cellular level of the vascularization andthe bone tissue. Consequently, there is a fundamentalneed for visual inspection and direct quantitative 3Danalysis of murine bone and bone vascularizationconcurrently.

In general, vascular corrosion casting (VCC) in con-junction with scanning electron microscopy (SEM) hasbeen widely used for imaging organ and tissue vascula-ture of many different species (Lametschwandtneret al., 1990). Casts are produced by filling an internalluminal system or space with a liquid medium that sol-idifies in situ. The surrounding tissue and bone is thenremoved (corroded) and the resulting replica is dried,

*Correspondence to: Ralph Muller, ETH Zurich, Institute for Biomechanics,Wolfgang-Pauli-Strasse 10, HCI E357.2, 8093 Zurich, Switzerland.E-mail: [email protected]

Received 26 October 2007; accepted in revised form 5 March 2009

Contract grant sponsor: Swiss National Science Foundation through the SNFProfessorship in Bioengineering; Contract grant numbers: FP 620-58097.99;PP-104317/1.

DOI 10.1002/jemt.20720

Published online 9 April 2009 in Wiley InterScience (www.interscience.wiley.com).

VVC 2009 WILEY-LISS, INC.

MICROSCOPY RESEARCH AND TECHNIQUE 72:690–701 (2009)

rendered conductive, and examined using SEM (Hoddeand Nowell, 1980; Northover et al., 1980). In particu-lar, VCC combined with SEM was applied for the studyof the vasculature and the microvasculature of bonesprimarily in rats (Aharinejad et al., 1995; Hirano et al.,1996; Morini et al., 1999, 2006; Okada et al., 2002; Pan-narale et al., 1997; Stanka et al., 1991). While vascularcorrosion casts represent the 3D architecture of thevascular network, SEM can only provide two-dimen-sional (2D) data. Because of these limitations, photo-grammetric methods have been devised, where stereo-scopic SEM images are digitally acquired (Komatsuet al., 1999; Malkusch et al., 1995; Manelli et al., 2007;Minnich et al., 1999;). Nevertheless, these methodsexclusively allow the reconstruction of the vascularsurfaces in terms of height maps and hence, are inher-ently pseudo-3D only (Manelli et al., 2007). Moreover,there is a wide range of error possibilities in the fieldof SEM photogrammetry that has to be taken intoaccount during image acquisition and through imagequantification (Minnich et al., 1999), which addition-ally complicate accurate morphometric measurements.

For a true 3D and direct assessment of the vascula-ture, X-ray micro-computed tomography (lCT) andsynchrotron (SR)-based lCT techniques have beenadapted (Heinzer et al., 2006; Jorgensen et al., 1998;Marxen et al., 2004; Plouraboue et al., 2004; Toyotaet al., 2002) and were used in combination with VCCand with a technique we would like to call vascularcontrast perfusion (VCP). For VCP, the vasculature isperfused with a highly X-ray absorbing contrast agent,and unlike VCC, the surrounding tissue is not alwayscorroded for CT imaging. Both vascular replica tech-niques (VCC and VCP) were primarily applied inrodents. Recently, lCT along with VCP was used with-out corrosion in rats to study weight-bearing inducedvessel formation during distraction osteogenesis(Moore et al., 2003) and in the mouse to analyze collat-eral vessel development after ischemic injury (Duvallet al., 2004). However, quantitative evaluation of bonevascularization in small rodents like mice remains achallenging task due to problems of casting material,specimen preparation as well as tomographic imaging.Recently, a new polyurethane (PU)-based casting mate-rial was described (Krucker et al., 2006; Meyer et al.,2007) with improved chemical and physical character-istics that allow less fragile reproduction of vascula-ture, increasing X-ray opacity, and high reproductionquality. Here, we propose an integrative approach forthe simultaneous assessment of murine bone and itsvasculature. This included the use of the PU-basedcasting material with three different adjusted vascularreplica protocols (two VCC, one VCP) to optimize X-rayopacity, partial tissue maceration to keep bone intact,and the necessary sample stability for conventionaland SR lCT. This allowed application of CT to imagevascular replicas and bone tissue down to the capillaryand cellular level respectively in a single measure-ment.

MATERIALS AND METHODSAnimals

Hind limbs of adult C57BL/6J mice (N 5 60, in housebreeding) were employed for the development of the

two VCC protocols. Additionally, C57BL/6J mice (N 55) (RCC Ltd, Fuellinsdorf, Switzerland) were dedicatedto the examination of the VCP protocol.

Materials

For the two VCC protocols, a polyurethane-basedcasting resin (PU4ii, vasQtec, Zurich, Switzerland)was used, which shows low viscosity, timely polymer-ization, and minimal shrinking and thus, provideshigh quality casts, including the finest capillaries(Krucker et al., 2006; Meyer et al., 2007). For the VCPprotocol, barium sulfate (BaSO4) particles (SachtlebenChemie GmbH, Duisburg, Germany) with a mean par-ticle size less than 1 lm were dispersed for 10 minusing a conical bead mill (Buhler AG, Uzwil, Switzer-land) in suspension. Before injection, 3% bovine gelatinwas added. The BaSO4 concentration of the injectedsuspension was 500 mg/mL. For concentrations higherthan 500 mg/mL, the decreased BaSO4 particle size ledto a significant increase in viscosity, up to a gel-likefluid, which made perfusion impossible. Therefore, theparticle size has been carefully optimized to reach aviscosity similar to that of blood. To test for homogene-ous contrast distribution, the intravascular BaSO4 per-fusion was assessed using light microscopy for a subse-quent D’Agostino-Pearson normality test (MATLABR2006b; The MathWorks, Inc., Natick, MA).

Animal Preparation

For the VCC protocols, the animal preparation wasdescribed in detail elsewhere (Krucker et al., 2004).Briefly, the mice were deeply anesthetized with 100mg/kg pentobarbital and intracardially perfusedthrough the left ventricle with heparinized artificialcerebro-spinal fluid (ACSF). ACSF was perfusedinstead of the more common phosphate buffered saline(PBS) since for these animals, brain vascular corrosioncasts were used for another study (Krucker et al.,2006). The perfusion (20–30 mL) was continued for 2–5min at a constant rate (5 mL/min) and immediately fol-lowed by injecting 4% paraformaldehyde (PFA) in PBS(20–30 mL) at the same rate for prefixation of the vas-cular system. Subsequently, the PU4ii resin (<10 mL)was infused at a lower rate (2.7 mL/min). For the VCPprotocol, deeply anesthetized mice were transcardiallyperfused with heparinized PBS followed by 4% PFA(same volume and rate as in the VCC protocol). Then,the dispersed BaSO4 suspension was injected (�5 mLin 1–2 min). No differences in vascular replica qualitywere observed due to the differences in the initial per-fusion media. The bodies were kept at 58C for a fewhours before one limb of each mouse was dissected,skinned, and stored in 4% PFA. All animal preparationwas performed by trained and licensed personnel andwas in accordance with federal and institutionalguidelines.

Vascular Replica Protocols

The two VCC protocols begin similarly with corro-sion (maceration) of the soft tissue, before they branchoff into two distinct protocols. The first VCC protocol(I) represents an adaptation of the classical VCCapproach, where the soft and bone tissues were

Microscopy Research and Technique

691SIMULTANEOUS ASSESSMENT OF MURINE BONE AND ITS VASCULATURE

resorbed and the remaining vascular replica wasstained and freeze-dried for CT imaging. The secondmore advanced VCC protocol (II) involved bone decalci-fication, delipidation of the marrow cavity, and freeze-drying of the sample. In the following, the two VCCprotocols are presented in detail. The VCP protocolrequired no further specimen treatment in addition tothe animal preparation steps described before.

Maceration. The absorption of the casting medium(PU4ii) and the soft tissue of the murine hind limbswere very similar for X-rays at the energy spectrumcovered by the lCT systems (around 30 keV). There-fore, for both VCC protocols, soft tissue was corroded(macerated) for X-ray absorption-based lCT assess-ment and segmentation of the vascular corrosion castby digital thresholding of the 3D tomographic dataset.Two different solutions for maceration of skinned hindlimbs were tested, 2% Alcalase1 (Novozymes A/S,Bagsvaerd, Denmark) and 7.5% potassium hydroxide(KOH) (Fluka Chemie GmbH, Buchs, Switzerland):immersing in Alcalase1 and KOH did not result in anychanges in PU quality over time as verified by visualinspection using a stereo microscope.

VCC Protocol I. For investigation of the vascularcorrosion cast only, bone needed to be separated beforelCT scanning (bone resorption). For this purpose,specimens were immersed in 5% formic acid (FlukaChemie GmbH, Buchs, Switzerland). Immersing in for-mic acid changed neither the morphology nor the qual-ity of the PU over time, as verified by visual inspectionunder a stereo microscope. In order to increase the sig-nal-to-noise ratio (SNR) and to reduce CT scanningtime, the PU corrosion casts were immersed in 2%osmium tetroxide (OsO4) solution (Riew and Smith,1971) and were mounted on a rocking shaker for 48 h(staining). The staining solution was prepared earlierby dissolving osmium crystals (Electron MicroscopySciences, Hatfield, PA, USA) in water during severaldays at 48C. OsO4 is a widely used staining agentapplied in transmission electron microscopy (TEM) toprovide contrast to the image. Prior to freeze-drying(lyophilization), macerated specimens were rinsedgently and frozen in water. The specimens were thenlyophilized (Christ, Alpha 1-2LD; Kuhner AG, Birsfel-den, Switzerland) for one day.

VCC Protocol II. Macerated specimens were decal-cified using 14% ethylenediamine tetra-acetic acid(EDTA) at pH 5 7.3 for 1 day (Kiviranta et al., 1980).Because the goal was to map the complete bone vascu-larization, the vascular corrosion cast needed to beassessed in the shaft of long bones (medullary cavity),where the vasculature is imbedded in the bone marrow.To provide optimal conditions for vascular corrosioncast segmentation within the medullary cavity, delipi-dation was necessary. In this process, marrow fat isremoved and the differential contrast between thereconstructed vascular corrosion cast and its surround-ings in the reconstructed tomographic data set isincreased. Therefore, chemical bone delipidation usingdifferent organic solvents (Beaupied et al., 2006; Chap-pard et al., 2004, 2006; Johnson et al., 2000; Moreauet al., 2000; Pruss et al., 1999; Worth et al., 2005) wastested. To test for possible physical damage, blue PUwas submerged for several days in acetone, butanone,chloroform, dichloromethane, ethanol, methanol, and

methyl salicylate before measuring with lCT. Second,bone delipidation was employed after decalcificationand prior to lyophilization to visualize properly thebone tissue and the vasculature within the medullarycavity at the same time. Finally, the specimens wererinsed gently in water and air-dried to extract residualsolvents, before they were rehydrated for later lyophili-zation as described in the VCC protocol I.

CT Imaging and SEM

Specimens of the VCC protocol I were measuredusing a lCT40 desktop scanner (Scanco Medical AG,Bruttisellen, Switzerland) at high resolution (10 lmvoxel size, 1000 projections over a range of 1808) andat 50 kV peak voltage. Specimens were imaged a firsttime by taking the average gray values of 10 projec-tions at every angular position (frame averagingmode) to increase the SNR prior to staining, and asecond time after staining and without frame averag-ing (FA). With this setup, one measurement with andwithout FA took eight hours or less than one hour,respectively. Reconstructions were conducted usingfiltered backprojection. After segmentation by meansof global thresholding, component labeling eliminatedsmall artifacts or fragments separated from thesample.

To verify that the corrosion cast penetrated the corti-cal bone and reached the medullary cavity in the caseof the VCC protocol II, where the murine bone and itsvasculature was assessed concurrently, six specimenswere prepared for SEM (Hitachi S-4000; Hitachi High-Technologies, Krefeld, Germany) by gold sputtering. Inaddition, the cortical shell of three bones was brokenintentionally before sputtering to provide direct accessto the medullary cavity using SEM. The other speci-mens of the VCC II and the VCP protocol were assessedby SR-based lCT. In comparison to conventional lCT,SR lCT provides higher SNR at increased resolution,which was important in this study for the examinationof the devised VCC II and VCP protocol. SR lCT meas-urements were performed in air at the X-ray Tomo-graphic Microscopy (XTM) station of the MaterialsScience (MS) beamline and at the beamline for tomo-graphic microscopy and coherent radiology experi-ments (TOMCAT) at the Swiss Light Source (SLS)(Patterson et al., 2005; Stampanoni et al., 2002, 2007).The coherence of the synchrotron light was reduced bydesign using a rotating paper filter to flatten the illu-mination of the field of view (FOV). For both setups,1001 projections were acquired over a range of 1808 ata photon energy of 17.5 keV. The data was recon-structed using filtered backprojection. Measurementswere performed in two different modes, in a global andlocal CT setup with corresponding nominal resolutionsof 3.7 and 1.4 lm, respectively. In the conventional(i.e., global) CT setup, the projections of the entire spec-imen were recorded. This is in contrast to the local CTsetup, where the specimen was bigger than therecorded FOV perpendicular to the rotation axis andtherefore, only a portion of the whole sample wasassessed. This local setup was required for high-resolu-tion lCT (nominal resolution of 1.4 lm) withoutdestruction of the specimen.


692 P. SCHNEIDER ET AL.

Image Preprocessing and Morphometry

To partially suppress noise within the reconstructeddata sets in both the global and local CT setup a con-strained Gaussian filter (r 5 1.2; filter support 5 1.0)was applied (Muller et al., 1994). After segmentation ofthe bone tissue and the vascular corrosion cast bymeans of global thresholding, component labeling elim-inated small artifacts or fragments separated from thesample. The canal network and the osteocyte lacunaewere segmented by means of negative imaging. In thiscontext, negative imaging denotes the technique tofirst measure the bone matrix using CT, and subse-quently, to extract the enclosed porosity as a negativeimprint of the surrounding matrix. The intracorticalporosity was then subdivided into the canal networkand the osteocyte lacunar system (Schneider et al.,2007). For quantitative comparison between the canalnetwork and the vascular replica of the VCP protocolassessed by global SR lCT, we quantified them bystandard and element-based morphometry asdescribed elsewhere (Schneider et al., 2007). In partic-ular, we calculated canal volume density (Ca.V/Ct.TV),mean canal volume (hCa.Vi), mean canal diameter(hCa.Dmi), and mean canal length (hCa.Lei) for thecanal network and the corresponding morphometricindices V.V/Ct.TV, hV.Vi, hV.Dmi, and hV.Lei for the vas-cular replica, where element-based indices (markedwith brackets h i) are given as mean values over thetotal number of elements. The vascular replica ofthe VCP protocol were evaluated first exclusivelyin the cortical bone (region 1) for direct comparisonwith the canal network and secondly, as a whole in themedullary cavity and the cortical bone together (region2). Since by design, the intracortical canal network isformed by the cortical bone (region 1), it was only eval-uated there. Finally, differences between the vascularindices in regions 1 and 2 and between the cannularand vascular morphometric measures in region 1 werecompared using unpaired t-tests.

RESULTSVascular Replica Protocols

Maceration. A time series (N 5 10) demonstratedthat seven to eight hours were sufficient for completemaceration of the soft tissue of mouse hind limbs (Fig.1A) using KOH, whereas full maceration was notachieved before 36 h in the case where Alcalase1 wasemployed as maceration solution. Figure 1B showsmouse hind limbs after successful maceration. No dif-ference in X-ray absorption was observed compared tountreated bone; neither for specimens corroded usingKOH, nor for samples treated with Alcalase1 (data notshown here).

VCC Protocol I. Depending on their size, bones ofmurine hind limbs were resorbed completely after notmore than 18 h when using 5% formic acid. Figure 1Crepresents mouse hind limbs after successful soft tis-sue corrosion and bone resorption. In Figure 1D, thecorresponding reconstructed tomographic data of thestained sample is given. The specimen was measured afirst time before staining without and with 10 timesFA. Component labeling of the 3D data sets revealedthat number of disjoint vascular components wasreduced by more than a factor of 10 and that the big-

gest vascular component accounted for more than 90%of all voxels when the scan was operated in the FAmode. In a second step, the same sample was stainedand measured again without FA. Component labelingin the same region of interest (Fig. 1D) showed that thenumber of disjoint vascular components was reducedby more than a factor of 20 compared to the unstainedspecimen.

VCC Protocol II. Figures 2A–2C display one speci-men in the region above the knee that has been imagedusing SEM at different magnification levels. The corro-sion cast penetrated the cortical bone, as it wasobserved for all six specimens analyzed by SEM. Fur-thermore, the corrosion cast was shown to reach themedullary cavity for all bones whose cortical shell wasbroken intentionally to provide direct insight. Anexample is given in Figures 2D–2F, where the castedvasculature within the broken neck of femur can bestudied. Relating to the actual VCC protocol II, over-decalcification was not an issue for decalcification withEDTA. We also tested a commercial decalcifier (Decal1;quartett Immunodiagnostika Biotechnologie Produk-tion und Vertriebs GmbH, Berlin, Germany), which ismore aggressive than EDTA. A time series analysis (N5 10) performed at the murine femoral mid-diaphysisusing lCT indicated that the cortical bone was entirelydecalcified using Decal1 within 15 min and that subse-quently, over-decalcification damaged the bone tissuesubstantially by partial resorption. For that reason,mild decalcifiers should be employed to prevent poten-tial over-decalcification. In Figure 3, one specimen isshown after corrosion, decalcification, and freeze-dry-ing; particularly in the region of the femoral mid-dia-physis (Fig. 3A) and the femoral head (Fig. 3B). Al-ready at low resolution (10 lm voxel size) using lCT(Figs. 3A and 3B) it became clear that marrow fatwithin the medullary cavity impeded proper vessel seg-mentation due to comparable X-ray absorption of themarrow fat and the casting material (PU4ii). Duringdelipidation, PU became slightly less elastic and occa-sionally very brittle when immersed in acetone or chlo-roform, respectively. Moreover, it began dissolving andtherefore lost resistance to a certain extent in ethanol,and more pronounced in methanol. For butanone,dichloromethane, and methyl salicylate, the PUbecame brittle—in particular for dichloromethane—and shrank at the same time. In view of the fact thatPU4ii is a polyurethane elastomer (Meyer et al., 2007)and given that solvents can attack plastic, their detri-mental influence on PU4ii becomes clear. Ultrasonica-tion, which can help removing fat traces mechanically,was not employed as it can introduce bone damage(Johnson et al., 2000). Finally, care needed to be takenwhen freeze-drying the macerated specimens as bonetissue became brittle when treated with KOH and,splintered during freeze-drying (Fig. 4). In Figure 5,one specimen is displayed after preparation accordingto the VCC protocol II as assessed with the highestachievable nominal resolution for global SR lCT in thisstudy (3.7 lm). The marrow fat was effectivelyextracted and did not interfere with proper segmenta-tion of the vascular corrosion cast, as it was previouslyobserved for nondelipidated specimens (Figs. 3Aand 3B).



VCP Protocol. The intravascular distribution of theBaSO4 particles is shown in Figure 6, which wasassessed using light microscopy. A D’Agostino-Pearsonnormality test indicated that the BaSO4 particles werenormally distributed (P < 0.05), which led to a homoge-nous contrast distribution, as required for quantitativeassessment of the vasculature using X-ray-based CT.One specimen filled with the BaSO4 suspension isdepicted in Figure 7A, which was measured by meansof local SR lCT. The vascular corrosion cast within themarrow cavity and the cortical bone could be success-fully assessed. For the purpose of comparison, thecanal network within the cortical bone (Fig. 7B) wasderived by negative imaging.

Morphometry

The quantitative comparison of the morphometricmeasures for the canal network and the vascular rep-lica is given in Tables 1 and 2. All element-based indi-ces of the vascular replica (hV.Vi, hV.Dmi, and hV.Lei)

were larger in region 2 (cortical bone 1 medullary cav-ity) in comparison with region 1 (cortical bone), whichturned out to be statistically significant for hV.Vi (P <0.002) and with a trend for hV.Dmi (P 5 0.06). In addi-tion, the vascular volume density (V.V/Ct.TV) wasfound to add up to 10% of the canal volume density(Ca.V/Ct.TV) (cf. Fig. 7) within the cortical bone(region 1). In the same region, no significant differen-ces between the canal and the vascular indices werefound (P > 0.05). However, all element-based morpho-metric indices, describing the mean extension of sin-gle cannular and vascular elements, were systemati-cally larger for the vascular replica compared to thecanal network.

DISCUSSION

Because there is growing evidence for a close func-tional relationship between osteogenesis and angiogen-esis as well as between bone remodeling and bone vas-cularization, there is a need for visual inspection and

Fig. 1. Mouse hind limbs before (A) and after (B) soft tissue corro-sion (maceration) in 7.5% KOH at 588C. A, B: The intact vascular cor-rosion cast (PU4ii) and the bone tissue are clearly visible. Mouse hindlimbs after soft tissue corrosion (maceration) and bone resorptionusing 5% formic acid (C), where the display detail shows the same

specimen after staining with 2% osmium tetroxide in the region of thefemur (VCC protocol I). D: Reconstructed lCT data of stained vascu-lar corrosion cast. C, D: Data were assessed by lCT at 10 lm nominalresolution.



Fig. 2. A–C: Murine specimen in the region above the knee aftersoft tissue corrosion, decalcification, delipidation, and freeze-drying(VCC protocol II). The corrosion cast penetrated the cortical bone asexpected. Data were assessed by SEM at (A) 322, (B) 390, and (C)3330 magnification. D–F: Murine specimen in the region of the neckof femur after soft tissue corrosion, decalcification, delipidation, and

freeze-drying (VCC protocol II). The cortical shell of the bone was bro-ken intentionally to provide direct insight into the medullary cavity,which proved that the corrosion cast reached also the inside of thebone. Data were assessed by SEM at (D) 319, (E) 365, and (F) 3120magnification.

quantitative analysis of bone vasculature. We thereforeimplemented three different protocols for vascular rep-licas and devised CT methods for a true 3D, direct, andsimultaneous measurement of vascular morphology

and bone tissue down to the capillary and cellularlevel. Moreover, vascular morphology was examined inthe light of the results based on a previously estab-lished method, where the canal network of corticalbone was derived only indirectly from cortical porosity.

Vascular Replica Protocols

VCC Protocol I. Soft tissue corrosion of the speci-mens should be performed using mild agents such asAlcalase1 at moderate alkalinity if the bone tissue hasto be preserved. Otherwise, there is the risk that thebone tissue will be irreversibly damaged during lyophi-lization. Further, staining with OsO4 largely increasesthe SNR in the reconstructed tomograms. In contrast,measurements with comparable SNR of unstainedspecimens would result in very high FA numbers andwith that, in long scanning times on the order of sev-eral days. For that reason, treatment of the vascularcorrosion cast with a highly X-ray absorbing contrastagent such as OsO4 can help reducing scanning timeon the one hand, and increasing SNR on the other, alsoenabling proper segmentation of the vascular corrosioncast (Figs. 1C and 1D), which is a prerequisite for mor-phometrical quantification. Unfortunately, OsO4 istoxic as well as expensive and thus, should be replacedin the future by a harmless and cheaper staining agentif possible. Moreover, it became difficult to orient thevascular corrosion cast after bone resorption, andalmost impossible to decide whether a vessel ranwithin the medullary cavity, within cortical bone, oractually was outside of the bone tissue. lCT measure-ments of macerated specimens revealed that simulta-neous assessment of the bone tissue and the vascularcorrosion cast was not feasible without any furtherspecimen treatment as the highly absorbing bone tis-

Fig. 3. Murine specimen in the region of the femoral mid-diaphysis after soft tissue corrosion, decal-cification, and freeze-drying (A) and in the region of the femoral head (B), respectively. The marrowfat within the medullary cavity impedes proper vessel segmentation due to comparable X-ray absorptionof the marrow fat and the casting material (PU4ii). Data were assessed by lCT at 10 lm nominalresolution.

Fig. 4. Murine knee following soft tissue corrosion with KOH andafter freeze-drying. Bone tissue splintered during freeze-drying andthus, mild agents such as Alcalase

1

at moderate alkalinity should beused for corrosion if the bone tissue has to be preserved (VCC protocolII).



sue masked the low absorbing vascular corrosion castduring tomographic reconstruction. This effect wasprobably due to the beam hardening correction, whichis included in the filtered backprojection algorithmsused by conventional lCT desktop scanners and whichcan only be adjusted for one tissue or material at thesame time. Decalcification before lyophilization cansolve this problem (VCC protocol II) by dissolvingcalcium stored in the bones and thereby, lowering theirX-ray absorption.

VCC Protocol II. The X-ray absorption of bone cansuccessfully be lowered, using preferably a mild decal-cifier such as 14% EDTA. Lowering the X-ray absorp-tion of the bone tissue by means of decalcificationrepresents an effective method for the simultaneousassessment of the vascular corrosion cast and the bonetissue using CT (Figs. 3A and 3B). Finally, soft solventssuch as acetone, which impaired neither the composi-tion and morphology of bone, nor the integrity of thecasting material, can help delipidating specimens (Fig.5), particularly in the medullary cavity of long bones,where marrow fat can interfere with proper segmenta-tion of the vascular corrosion cast.

VCP Protocol. Although the vascular corrosion castand the bone tissue were available at the same time forthe VCC protocol II, they could not be separatedthrough segmentation based on global thresholding,given that the lowered X-ray absorption of the bone tis-sue approaches those values for the vascular corrosioncast material (PU4ii). In Figure 5, the interweavementof vascular corrosion cast, cortical, and trabecular bone

can largely be discerned by an operator familiar withthe murine bone microarchitecture and ultrastructure.Yet, many of these components are continuously con-nected with each other and thus, cannot be separatedautomatically by component labeling. Therefore,human operator interaction for segmentation by handis necessary, which is time-consuming and beyond,leads to nonreproducible morphometric results. An al-ternative approach - which may solve this problem - isto inject a suitable contrast agent to the vascular sys-tem in vivo (Kiessling et al., 2004) as it was done inthis study by using a customized BaSO4 suspension. Inthis manner, the vascular replica and the bone tissuecould be isolated by simple global thresholding (Fig.7A). In addition, it must be pointed out here that com-pared to the VCC protocols, no additional sample prep-aration step was required after dissection of the mu-rine limb. In such a way, not only time is saved, butalso experimental uncertainties with respect to themorphology and integrity of the vascular corrosion castcan be avoided, which are usually involved in everyVCC protocol.

Morphometry

Compared to the assessment of the canal network byusing negative imaging (Fig. 7B), all vascular replicaprotocols presented in this study (VCC I and II, andVCP) have the advantage that they provide directmethods for the 3D assessment of the vascular replica.All morphometric measures for the canal networkreported in this study (Tables 1 and 2) correspond topreviously reported values for the murine cannularmorphology in the same region, i.e., the femoralmid-diaphysis (Schneider et al., 2007). Moreover, thesmaller dimensions of the vascular elements in region1 (cortical bone) compared to region 2 (cortical bone 1medullary cavity) was due to the fact that the vascula-

Fig. 5. Murine specimen after soft tissue corrosion, decalcification,delipidation, and freeze-drying (VCC protocol II). The vascular corro-sion cast and the bone tissue are accessible at the same time, whereasthey cannot be disentangled through segmentation based on globalthresholding. Data were assessed by global SR lCT at 3.7 lm nominalresolution.

Fig. 6. Intravascular distribution of the BaSO4 particles used forthe VCP protocol. Quantitative evaluation showed that BaSO4 par-ticles were normally distributed, leading to a homogenous contrastdistribution within the vasculature. Data were analyzed by lightmicroscopy.



ture is confined by the cortical bone in region 1,whereas large branches of the vasculature can be foundwithin the medullary cavity (cf. Fig. 7A). Interestingly,all element-based morphometric indices within thecortical bone were found to be larger for the vascularreplica in contrast to the canal network, yet not statis-

tically significant. This is counterintuitive because thespace between the cannular and the vascular surface isoccupied by the endothelium, an �1-lm-thick cell layerat the interior surface of the vascular system (van denBerg et al., 2006), and consequently, the extension ofthe cannular elements was expected to exceed the vas-cular dimensions. There are two possible explanationsfor this finding. First, only a fraction of the canal net-work was originally occupied by functional vessels inthe animal, specifically by larger branches of the vascu-lature. The remaining volume of the canal networkwould then be empty and/or inhabited by bone remod-eling units (including osteoclast lacunae). Secondly, thevascular contrast perfusion was not complete. Forinstance, round or tapered dead ends and the lack ofendothelial imprints on the cast surface are wellknown indicators of incomplete intravascular filling

Fig. 7. A: Murine specimen in the region of the femoral mid-dia-physis, whose vasculature was filled with a barium sulfate suspensionin vivo (VCP protocol). Vascular replica (vessels in red) within themarrow cavity and the cortical bone (semitransparent shell in white),which includes osteocyte lacunae (prolate ellipsoids in yellow). Thevascular replica and the bone tissue were isolated by global threshold-

ing. B: Canal network (tubes in green) and osteocyte lacunae (prolateellipsoids in yellow) within cortical bone (semitransparent shell inwhite) were extracted from reconstructed lCT data using negativeimaging. Data were assessed by local SR lCT at 1.4 lm nominalresolution.

TABLE 1. Morphology of the canal network

Region

Canal network

Ca.V/Ct.TV(%)

hCa.Vi(1023 � mm3)

hCa.Dmi(lm)

hCa.Lei(lm)

Cortical bone(region 1)

1.22 6 0.26 19 6 2 10.3 6 0.4 89 6 6

Values are mean 6 standard deviation (N5 5).Ca.V/Ct.TV, canal volume density; hCa.Vi, mean canal volume, hCa.Dmi, meancanal diameter; hCa.Lei, mean canal length.



(Lametschwandtner et al., 1990; Weber et al., 2008).Both assumptions are supported by the observationthat the vascular volume density (V.V/Ct.TV)amounted to no more than 10% of the canal volumedensity (Ca.V/Ct.TV) (Tables 1 and 2). Eventually, asboth explanations do not exclude each other, a combi-nation of both factors may explain morphometric differ-ences between the canal network and the vascularreplica. Future investigations should unambiguouslyidentify the actual factors that cause the observeddifference between the extension of the cannular andvascular elements.

Limitations

There were several sources of error within the pre-sented protocols, which can affect the morphometricalquantification of the bone tissue and the vascular rep-lica. They are summarized in the following. First, scan-ning and reconstruction of the vascular corrosion castaccording to VCC protocol I without prior stainingleaves many disjoint vascular components. As a result,subsequent quantification provides wrong morphomet-ric measures of the vascular corrosion cast. On thisaccount, staining with a highly X-ray absorbing con-trast agent such as OsO4 is a prerequisite for propersegmentation of the vascular corrosion cast for theVCC protocol I. Although the procedure is simple andstraightforward, OsO4 is unfortunately toxic and ex-pensive and should be replaced at some point by aharmless and cheaper staining agent. Second, over-decalcification can happen in the VCC protocol II whenusing aggressive decalcifiers. Consequently, bone tis-sue integrity is impaired by partial resorption and themorphometric outcomes are distorted. For this reason,only mild decalcifiers should be used for decalcification,such as EDTA. Third, inappropriate selection of thesolvent for delipidation in line with the VCC protocol IIcan particularly cause dissolving and shrinking of thepolyurethane-based casting material, which introducesa bias of the derived vascular morphometric indices.Therefore, soft solvents must be used for delipidation,which do not alter the morphology of the vascular cor-rosion cast material. In this study, acetone turned outto be the optimal solvent since it did not change themorphology of the polyurethane-based casting materialand only slightly reduced its elasticity. Finally, for bothVCC protocols (I and II), human operator interactionwas required for the orientation or segmentation of thevascular corrosion cast and the bone tissue. This leadsto nonreproducible morphometric results. The VCPprotocol overcomes this limitation, where the vascular

replica and the bone tissue were separated automati-cally by simple global thresholding.

CONCLUSIONS

There is one published preclinical work (Duvallet al., 2004), where lCT at nominal resolutionsbetween 16 and 36 lm along with VCC was used to an-alyze collateral vessel development after ischemicinjury. In addition, barium sulfate as contrast agent forlCT at 70 lm nominal resolution has been used beforein a preclinical study for in vivo imaging of tumorangiogenesis (Kiessling et al., 2004). In our basicresearch study, the goal was the development of severalprotocols for the concomitant quantification of the vas-culature and the bone tissue at very high 1.4 and 3.7lm nominal resolution, which allows for reliable 3Dquantification and morphometric analysis of murinebone ultrastructure as recently introduced (Schneideret al., 2007).

In particular, we devised two different VCC proto-cols, identifying and handling a number of challengesduring sample preparation. The first VCC protocol (I)represents an adaptation of the classical VCCapproach, where the soft and bone tissues wereresorbed and the remaining vascular replica wasstained and freeze-dried for CT imaging. It is straight-forward and thus, easy to implement. However, it wasdifficult to locate the vascular corrosion cast in relationto the resorbed bone, which makes it particularly prob-lematical to compare quantitative measures of the vas-cular corrosion cast from different animals. The VCCprotocol II solved this problem, where the bone tissuewas decalcified for the concomitant lCT measurementof the bone tissue and the vascular corrosion cast.Unfortunately, the vascular corrosion cast and thedecalcified bone tissue must then still be separated bya human operator since the two phases showed similarX-ray absorption and thus, comparable contrast inthe reconstructed CT data. For this reason, we alsoadapted SR lCT methods to combine them with a novelcontrast modality (VCP protocol), where a customizedcontrast suspension was directly injected to the vascu-lar system in vivo. In this manner, vascular replicaswere imaged and quantified successfully down to thecapillary level with simultaneous measurement of thebone tissue down to a cellular regime. In addition, wecould now separate the vascular replica and the bonetissue automatically by simple global thresholdingdue to the significant difference in X-ray absorptionbetween the BaSO4 suspension and the bone tissue.Another advantage of the VCP protocol was that com-pared to the VCC protocols, no additional sample prep-aration step was required after dissection of the mu-

TABLE 2. Morphology of the vascular replica

Region

Vascular replica

V.V/Ct.TV (%) hV.Vi (1023 � mm3) hV.Dmi (lm) hV.Lei (lm)

Cortical bone (region 1) 0.13 6 0.10 28 6 10 11.9 6 3.1 102 6 46Cortical bone 1 medullary cavity (region 2) — 56 6 10 15.0 6 0.6 150 6 56Difference (%) 1100a 126b 148c

Values are mean 6 standard deviation (N5 5). V.V/Ct.TV is only defined in region 1, i.e., within the cortical total volume (Ct.TV).V.V/Ct.TV, vascular volume density; hV.Vi, mean vascular volume; hV.Dmi, mean vascular diameter; hV.Lei, mean vascular length.aUnpaired t-test: P (region) < 0.002.bUnpaired t-test: P (region) 5 0.06.cUnpaired t-test: P (region) > 0.1.



rine limb. By this means, time was saved and severalsources of error within the VCC protocols (see Limita-tions) were eliminated, which can affect the morpho-metrical quantification of the bone tissue and the vas-cular replica. One disadvantage of the VCP protocol isthough that the vascular replica is not readily avail-able for other imaging modalities such as optical orelectron microscopy. Since this is the case for the twoVCC protocols, there is still a need to keep these proto-cols as well.

Eventually, cannular and vascular morphometricindices, as they were derived in this study, can now berelated directly to measures of bone biomechanics, pro-viding important insights in the structure functionrelationships between bone tissue and bone vasculari-zation. Generally speaking, our presented protocolscan be applied to study quantitatively the vascular pat-tern of many different normal and pathological organsand tissues in a truly 3D fashion (Lametschwandtneret al., 1990), such as in the brain, heart, kidney, lung,and the skin, to name only a few. In the long run, webelieve that the presented framework of an extendedVCC protocol and a novel VCP modality together withconventional as well as adapted SR-based CT methodscan serve as a potent and complementary combinationfor simultaneous 3D quantitative morphometry of thebone tissue and its vascularization at the same time,providing new insights in the close relationshipbetween bone tissue and the vascular network.

ACKNOWLEDGMENTS

We thank Dr. Johannes Vogel for the animal care,Dr. Bernhard Stalder from Buhler AG (Uzwil, Switzer-land) for expert help in the BaSO4 dispersion proce-dure, and Dr. Amela Groso for her support during themeasurements at the Swiss Light Source.

REFERENCES

Aharinejad S, Marks SC Jr, Bock P, MacKay CA, Larson EK, Taham-tani A, Mason-Savas A, Firbas W. 1995. Microvascular pattern inthe metaphysis during bone growth. Anat Rec 242:111–122.

Barou O, Mekraldi S, Vico L, Boivin G, Alexandre C, Lafage-ProustMH. 2002. Relationships between trabecular bone remodeling andbone vascularization: A quantitative study. Bone 30:604–612.

Beaupied H, Chappard C, Basillais A, Lespessailles E, Benhamou CL.2006. Effect of specimen conditioning on the microarchitecturalparameters of trabecular bone assessed by micro-computed tomo-graphy. Phys Med Biol 51:4621–4634.

Chappard D, Moquereau M, Mercier P, Gallois Y, Legrand E, BasleMF, Audran M. 2004. Ex vivo bone mineral density of the wrist:Influence of medullar fat. Bone 34:1023–1028.

Chappard D, Pascaretti-Grizon F, Gallois Y, Mercier P, Basle MF,Audran M. 2006. Medullar fat influences texture analysis of trabec-ular microarchitecture on X-ray radiographs. Eur J Radiol 58:404–410.

Collin-Osdoby P. 1994. Role of vascular endothelial cells in bonebiology. J Cell Biochem 55:304–309.

Colnot C. 2005. Cellular and molecular interactions regulating skele-togenesis. J Cell Biochem 95:688–697.

Cruess RL. 1986. Osteonecrosis of bone. Current concepts as to etiol-ogy and pathogenesis. Clin Orthop Relat Res 208:30–39.

Duvall CL, Taylor WR, Weiss D, Guldberg RE. 2004. Quantitativemicrocomputed tomography analysis of collateral vessel develop-ment following ischemic injury. Am J Physiol Heart Circ Physiol287:H302–H310.

Glimcher MJ, Kenzora JE. 1979. Nicolas Andry award. The biology ofosteonecrosis of the human femoral head and its clinical implica-tions, Part 1: Tissue biology. Clin Orthop Relat Res 138:284–309.

Ham AW. 1952. Some histophysiological problems peculiar to calcifiedtissues. J Bone Joint Surg Am 34:701–728.

Heinzer S, Krucker T, Stampanoni M, Abela R, Meyer EP, Schuler A,Schneider P, Muller R. 2006. Hierarchical microimaging for multi-scale analysis of large vascular networks. Neuroimage 32:626–636.

Hirano T, Majima R, Yoshida G, Iwasaki K. 1996. Characteristics ofblood vessels feeding the femoral head liable to osteonecrosis inspontaneously hypertensive rats. Calcif Tissue Int 58:201–205.

Hodde KC, Nowell JA. 1980. SEM of micro-corrosion casts. ScanElectron Microsc 2:89–106.

Horner A, Bord S, Kelsall AW, Coleman N, Compston JE. 2001. Tie2ligands angiopoietin-1 and angiopoietin-2 are coexpressed with vas-cular endothelial cell growth factor in growing human bone. Bone28:65–71.

Johnson GS, Mucalo MR, Lorier MA. 2000. The processing and char-acterization of animal-derived bone to yield materials with biomedi-cal applications, Part 1: Modifiable porous implants from bovinecondyle cancellous bone and characterization of bone materials as afunction of processing. J Mater Sci Mater M 11:427–441.

Jorgensen SM, Demirkaya O, Ritman EL. 1998. Three-dimensionalimaging of vasculature and parenchyma in intact rodent organswith X-ray micro-CT. Am J Physiol Heart C 275:H1103–H1114.

Kiessling F, Greschus S, Lichy MP, Bock M, Fink C, Vosseler S, MollJ, Mueller MM, Fusenig NE, Traupe H, Semmler W. 2004. Volumet-ric computed tomography (VCT): A new technology for noninvasive,high-resolution monitoring of tumor angiogenesis. Nat Med10:1133–1138.

Kiviranta I, Tammi M, Lappalainen R, Kuusela T, Helminen HJ.1980. The rate of calcium extraction during EDTA decalcificationfrom thin bone slices as assessed with atomic-absorption spectro-photometry. Histochemistry 68:119–127.

Komatsu F, Tsubusaki K, Miyoshi M. 1999. Measurement of geometri-cal dimensions using scanning electron microscopy. J ElectronMicrosc (Tokyo) 48:407–415.

Krucker T, Schuler A, Meyer EP, Staufenbiel M, Beckmann N. 2004.Magnetic resonance angiography and vascular corrosion casting astools in biomedical research: Application to transgenic mice model-ing Alzheimer’s disease. Neurol Res 26:507–516.

Krucker T, Lang A, Meyer EP. 2006. New polyurethane-based mate-rial for vascular corrosion casting with improved physical andimaging characteristics. Microsc Res Tech 69:138–147.

Lametschwandtner A, Lametschwandtner U, Weiger T. 1990. Scan-ning electron microscopy of vascular corrosion casts—techniqueand applications: Updated review. Scanning Microsc 4:889–940, dis-cussion 941.

Malkusch W, Konerding MA, Klapthor B, Bruch J. 1995. A simpleand accurate method for 3-D measurements in microcorrosion castsillustrated with tumour vascularization. Anal Cell Pathol 9:69–81.

Manelli A, Sangiorgi S, Binaghi E, Raspanti M. 2007. 3D analysis ofSEM images of corrosion casting using adaptive stereo matching.Microsc Res Tech 70:350–354.

Marxen M, Thornton MM, Chiarot CB, Klement G, Koprivnikar J,Sled JG, Henkelman RM. 2004. MicroCT scanner performance andconsiderations for vascular specimen imaging. Med Phys 31:305–313.

Meyer EP, Beer GM, Lang A, Manestar M, Krucker T, Meier S, Mihic-Probst D, Groscurth P. 2007. Polyurethane elastomer: A new mate-rial for the visualization of cadaveric blood vessels. Clin Anat20:448–454.

Minnich B, Leeb H, Bernroider EW, Lametschwandtner A. 1999.Three-dimensional morphometry in scanning electron microscopy:A technique for accurate dimensional and angular measurementsof microstructures using stereopaired digitized images and digitalimage analysis. J Microsc Oxford 195:23–33.

Moore DC, Leblanc CW, Muller R, Crisco JJ III, Ehrlich MG. 2003.Physiologic weight-bearing increases new vessel formation duringdistraction osteogenesis: A micro-tomographic imaging study.J Orthop Res 21:489–496.

Moreau MF, Gallois Y, Basle MF, Chappard D. 2000. Gamma irradia-tion of human bone allografts alters medullary lipids and releasestoxic compounds for osteoblast-like cells. Biomaterials 21:369–376.

Morini S, Pannarale L, Franchitto A, Donati S, Gaudio E. 1999. Mi-crovascular features and ossification process in the femoral head ofgrowing rats. J Anat 195:225–233.

Morini S, Pannarale L, Conti D, Gaudio E. 2006. Microvascular adap-tation to growth in rat humeral head. Anat Embryol (Berl)211:403–411.

Muller R, Hildebrand T, Ruegsegger P. 1994. Non-invasive bone bi-opsy: A new method to analyse and display the three-dimensionalstructure of trabecular bone. Phys Med Biol 39:145–164.

Northover JM, Williams ED, Terblanche J. 1980. The investigation ofsmall vessel anatomy by scanning electron microscopy of resincasts. A description of the technique and examples of its use in the



study of the microvasculature of the peritoneum and bile duct wall.J Anat 130:43–54.

Okada S, Yoshida S, Ashrafi SH, Schraufnagel DE. 2002. The canalic-ular structure of compact bone in the rat at different ages. MicroscMicroanal 8:104–115.

Pannarale L, Morini S, D’Ubaldo E, Gaudio E, Marinozzi G. 1997.SEM corrosion-casts study of the microcirculation of the flat bonesin the rat. Anat Rec 247:462–471.

Parfitt AM. 2000. The mechanism of coupling: A role for the vascula-ture. Bone 26:319–323.

Patterson BD, Abela R, Auderset H, Chen Q, Fauth F, Gozzo F, IngoldG, Kuhne H, Lange M, Maden D, Meister D, Pattison P, Schmidt T,Schmitt B, Schulze-Briese C, Shi M, Stampanoni M, Willmott PR.2005. The materials science beamline at the Swiss Light Source:Design and realization. Nucl Instrum Methods Phys Res A 540:42–67.

Plouraboue F, Cloetens P, Fonta C, Steyer A, Lauwers F, Marc-Vergnes JP. 2004. X-ray high-resolution vascular network imaging.J Microsc Oxford 215:139–148.

Pruss A, Kao M, Kiesewetter H, von Versen R, Pauli G. 1999. Virussafety of avital bone tissue transplants: Evaluation of sterilizationsteps of spongiosa cuboids using a peracetic acid-methanol mixture.Biologicals 27:195–201.

Riew CK, Smith RW. 1971. Modified osmium tetroxide stain formicroscopy of rubber-toughened resins. J Polym Sci A1 9:2739–2744.

Schneider P, Stauber M, Voide R, Stampanoni M, Donahue LR, Mul-ler R. 2007. Ultrastructural properties in cortical bone vary greatly

in two inbred strains of mice as assessed by synchrotron light basedmicro- and nano-CT. J Bone Miner Res 22:1557–1570.

Stampanoni M, Borchert G, Wyss P, Abela R, Patterson B, Hunt S,Vermeulen D, Ruegsegger P. 2002. High resolution X-ray detectorfor synchrotron-based microtomography. Nucl Instrum MethodsPhys Res A 491:291–301.

Stampanoni M, Groso A, Isenegger A, Mikuljan G, Chen Q, Meister D,Lange M, Betemps R, Henein S, Abela R. 2007. TOMCAT: A beamlinefor TOmographic Microscopy and Coherent rAdiology experimenTs.Synchrotron Radiat Instrum 879 (Parts 1 and 2):848–851.

Stanka P, Bellack U, Lindner A. 1991. On the morphology of the ter-minal microvasculature during endochondral ossification in rats.Bone Miner 13:93–101.

Toyota E, Fujimoto K, Ogasawara Y, Kajita T, Shigeto F, MatsumotoT, Goto M, Kajiya F. 2002. Dynamic changes in three-dimensionalarchitecture and vascular volume of transmural coronary microvas-culature between diastolic- and systolic-arrested rat hearts. Circu-lation 105:621–626.

van den Berg BM, Spaan JA, Rolf TM, Vink H. 2006. Atherogenicregion and diet diminish glycocalyx dimension and increase intima-to-media ratios at murine carotid artery bifurcation. Am J Physiol-Heart C 290:H915–H920.

Weber B, Keller AL, Reichold J, Logothetis NK. 2008. The microvas-cular system of the striate and extrastriate visual cortex of the mac-aque. Cereb Cortex 18:2318–2330.

Worth A, Mucalo M, Horne G, Bruce W, Burbidge H. 2005. The evalu-ation of processed cancellous bovine bone as a bone graft substitute.Clin Oral Implants Res 16:379–386.



simultaneous 3d visualization and quantification of murine

Documents