INSTITUTE OF PHYSICS PUBLISHING
PHYSIOLOGICAL MEASUREMENT
Physiol. Meas. 24 (2003) 165–178
PII: S0967-3334(03)52987-2
Quantitative analysis of bone mineral content by x-ray
microtomography
A A Postnov1,2, A V Vinogradov2, D Van Dyck1, S V Saveliev3
and N M De Clerck1
1
Department of Physics and Department of Biomedical Sciences, University of Antwerp
(RUCA), Groenenborgerlaan, 171, Antwerp B-2020, Belgium
2
Lebedev’s Physical Institute Moscow, Russian Academy of Sciences, Leninsky pr. 53,
Moscow 117333, Russia
3 Institute of Human Morphology, Russian Academy of Medical Sciences, Cjurupy st. 3,
Moscow 117418, Russia
E-mail: [email protected]
Received 3 September 2002
Published 17 January 2003
Online at stacks.iop.org/PM/24/165
Abstract
A new non-destructive method based on x-ray microtomography (micro-CT)
was developed to measure calcium density in bone. X-ray micro-CT was used
as a quantitative approach to acquire and reconstruct virtual cross-sections
through the sample. Accurate beam-hardening correction was implemented.
Grey values in the virtual cross-sections were calibrated as calcium mineral
density in bone. From these cross-sections, three-dimensional models were
created.
Calcium content was calculated directly from images and expressed as
percentage per volume and per weight. Calcium mineral density was studied
by this method in a unique set of bones isolated from newts (Pleurodeles waltlii
Michah) that had travelled into space. A demineralization of 10% was shown
as a consequence of sustained micro-gravity.
Keywords: quantitative micro-CT, bone mineral density, weightlessness
1. Introduction
X-ray microtomography (micro-CT) attracts more and more attention in biomedical research.
Micro-CT is a powerful technique that allows visualization of the internal structure of opaque
objects without destroying the sample. As with synchrotron illumination (Bonse 1997),
micro-CT with laboratory polychromatic x-ray sources already has several applications
in bone research (Elliott et al 1997, Stenstrom et al 2000, Ruegsegger et al 1976, 1996).
The overall bone quality is determined by its structural and material properties, such as
0967-3334/03/010165+14$30.00
© 2003 IOP Publishing Ltd Printed in the UK
165
166
A A Postnov et al
bone mass, geometry, architecture and composition of the bone (Einhorn 1992). Bone mineral
density (BMD) is commonly used as an indicator of bone status, both structural and metabolic
(Odgaard 1997).
In clinical (Cummings 1998) and experimental bone research (Ederveen and Kloosterboer
1999, Beamer et al 1996), it is a convenient practice to express mineral content as density
of the sample. Depending on the applied technique, linear (g cm−1), areal (g cm−2) or
volumetric densities (Archimedes principle, g cm−3) are used to compare different conditions
in bone (Augat et al 1998). However, to express mineral content as physical density, the
question remains how to measure the volume of the sample as accurately as possible (Ming
2000). A serious problem with these estimations arises from non-uniformity within the bone
and from the contribution of porosity, especially in cancellous bone (Odgaard 1997). In
addition to problems concerning the measurement of volume and the impact of porosity, some
discrepancies in calibration procedures have been reported (Martin and Reid 1999).
In the present study, a new method has been developed for measuring bone mineral content
using high-resolution desktop micro-CT. The major advantage is that all measurements are
completely non-destructive. Moreover, separate measurements for trabecular, cortical and
integral bone can be provided by micro-CT (Augat et al 1998). However, an important
and difficult challenge in the field of microtomography is to establish a reliable method for
calibration of x-ray attenuation in the reconstructed cross-sections. In contrast to synchrotron
illumination, the use of polychromatic x-rays creates a problem of beam hardening (i.e.,
spectra of the beam change while passing through the investigated object) (Herman 1980). To
overcome this problem, a special correction was included in the present method.
Consecutive virtual slices through a bone could be reconstructed resulting in the creation
of three-dimensional (3D) models. The bones were not affected by radiation and no extensive
sample preparation was required prior to scanning. After scanning, each bone could be used
for additional examination. Moreover, the method can be extended for application in living
animals during in vivo scanning (Postnov et al 2002a, 2002b).
To illustrate this new method, a unique pilot experiment was performed where mineral
content in bones isolated from newts that have travelled into space was analysed by micro-CT.
Bones from animals that were exposed to micro-gravity were chosen because micro-gravity
has been reported to alter bone mineral content (Saveliev and Besova 1993, Saveliev et al
1993). Micro-CT is the most suitable technique for this analysis as bones from the animals
aboard the spacecraft are unique and should not be destroyed but could be further analysed in
other studies.
2. Methods
2.1. Experimental set-up
All measurements were performed by a micro-CT desktop system, which was based on the
combination of x-ray projection microscopy with a tomographical reconstruction technique
(Van Dyck and Sasov 1998, Boyde et al 2000, Sasov and Van Dyck 1998, www.skyscan.be).
In this system (SkyScan-1072, Belgium), an air-cooled point x-ray source (focal spot size
∼8 µm in diameter, maximum voltage 80 kV) was used to illuminate the object with a
divergent beam. Magnified shadow pictures were detected by a two-dimensional CCD camera.
Cross-sections were reconstructed using the Feldkamp cone-beam algorithm (Feldkamp et al
1984) and were combined into 3D models if necessary.
All samples and phantoms were scanned under identical conditions. Scanning parameters
were as follows: anode voltage was 80 kV, 0.9◦ rotation step, exposure time was 7 s per view.
Quantitative analysis by x-ray microtomography
167
Table 1. Summary of different components of bone (Driessens and Verbeeck 1990) together with
their imaginary parts of form factors f2 (Henke et al 1993) and fraction of absorbed energy.
Element,
atomic
weight
Atomic
fraction by
weight (%)
Fraction by
number of
atoms (%)
Form factor
f2 at 20 keV
Form factor
f2 at 32 keV
Form factor
f2 at 42 keV
Estimated
fraction of
absorbed energy (%)
H, 1
C, 12
N, 14
O, 16
Na, 23
Mg, 24
P, 31
S, 32
Ca, 40
3.4
15.5
4.2
43.5
0.1
0.2
10.3
0.3
22.5
39.5
15.1
3.5
31.6
0.05
0.08
3.8
0.1
6.5
1.14 × 10−7
1.16 × 10−3
2.38 × 10−3
4.40 × 10−3
1.84 × 10−2
2.66 × 10−2
6.97 × 10−2
9.17 × 10−2
2.32 × 10−1
3.65 × 10−8
9.95 × 10−4
8.34 × 10−3
1.56 × 10−3
6.68 × 10−3
9.88 × 10−3
2.67 × 10−2
3.53 × 10−2
9.07 × 10−2
1.92 × 10−8
2.17 × 10−4
4.52 × 10−4
8.47 × 10−4
3.67 × 10−3
5.49 × 10−3
1.51 × 10−2
2.00 × 10−2
5.21 × 10−2
<0.1
0.7–0.9
0.3–0.4
6–7
<0.1
<0.1
13–14
<0.1
78–80
An aluminium filter was installed in the beam path to cut off the softest x-rays. This was
necessary to increase the accuracy of the beam-hardening correction (BHC). With this filter,
the detector response (i.e., the amount of counted photons versus object thickness) was close
to linear when bones were investigated.
After scanning, virtual cross-sections through the bone were reconstructed with 32-bit
dynamic range and converted into 8-bit bitmap (BMP) images according to the selected density
window. The advantage of the 8-bit BMP images is that they can be visualized on the computer
monitor. Every pixel in an 8-bit BMP image has a colour or grey value between 0 and 255.
Colour 255 was assumed to be white (void space), whereas 0 is black or the densest part of
the image. This colour code has been applied in all illustrations in this paper. It was the aim
in the present study to correlate different colours in the reconstructed cross-sections of the
micro-CT image to a given mineral content.
To express grey values as mineral content, appropriate phantoms were required. For this
calibration, hydroxyapatite (HA) Ca10(PO4)6(OH)2 was used, as the chemical composition
of this compound is very similar to the mineral part of bone. A demineralized bone was
studied as another phantom to mimic the non-crystalline, organic part of bone. Bones were
demineralized by a standard procedure where the bones were stored in 0.2 M EDTA for 7 days
(Bowman et al 1996).
2.2. Analysis of images and grey values
From a chemical point of view, bone is a very complicated compound (Driessens and Verbeeck
1990). Bones are composed of HA, collagen, yellow marrow (fat) and bloody marrow (water).
For micro-CT analysis, bone can be considered as a crystalline fraction and an organic matrix.
The contribution to x-ray absorption of each component in bone was calculated and measured.
As shown in table 1, specific x-ray absorption occurs in all elements.
X-ray absorption (dispersion) is known to be proportional to the number of atoms of
each element. For the photons generated by an x-ray tube with 80 kV peak energy and
tungsten anode, and registered by x-ray CCD camera, incoherent scattering does not contribute
significantly. Thus interaction of x-rays with a physical substance can be described using
optical constants (http://physics.nist.gov/PhysRefData/Xcom/html/xcom1.html).
The index of refraction can be expressed as (Chantler 1995)
r0 2 nj fj
(1)
λ
nr = 1 −
2π
j
168
A A Postnov et al
where r0 = 2.818 × 10−15 m (classical electron radius), λ is the wavelength, nj is the atom
density and f is the form factor (complex value).
Atom density in formula (1) can be defined as
ρ
ξj
(2)
nj =
mp A j
where ρ is the physical density (weight/volume) of the material (bone, in the present study),
mp = 1.67 × 10−24 g (proton weight), Aj is the atomic weight number of the element and ξj
is the fraction of the jth element in material.
The imaginary part of the dispersion form factor (f2 ) plays a key role in the probability for
an atom to absorb or disperse a 20–40 keV x-ray photon. It defines which elements contribute
more to absorption contrast. Using these formulae, we calculated (table 1) that Na, Mg, S and
some other elements could be neglected because their fraction is not significant. In addition,
H can be excluded as it has an extremely low form factor f2 and, therefore, a negligible
absorption. Thus, as far as x-ray absorption is concerned, bone can be considered as a mixture
of four major elements: Ca, P, C, O. Although the x-ray radiation used is polychromatic,
the proportion in absorption between different elements remains relatively constant in the
whole energy range (20–40 keV), because these key elements have no absorption edges in this
energy interval.
According to table 1, P and Ca together absorb much more than the remaining organic
part of bone (approximately 93% versus 7% of x-ray photons) being only 10% by number of
atoms. This means that when an additional small fraction of P–Ca mixture is added to the
bone, this causes approximately 100 times more absorption than when the same additional
fraction of the remaining Ca-free components of the bone is added. However, carbon and
oxygen are major components of bone and the total absorption in these light elements cannot
be neglected (table 1).
Consequently, the composition of bone can be represented as a mixture of two compounds:
HA, and a mixture of carbon and oxygen (CO).
As the present method is much more sensitive to the calcium balance than to other
materials, we assumed for simplicity that x-ray absorption in the organic part remains constant
regardless of changes in mineral content. Demineralized bones were scanned to confirm this
hypothesis.
Thus, x-ray absorption Aexp (expressed in arbitrary units) measured in an experimental
cross-section in bone is given by the sum of the absorption caused by the HA fraction and the
organic CO part,
Aexp = nHA FHA + nCO FCO = nHA FHA + ACO
where n is the atom density and F is the factor that defines the average absorption by one
atom/molecule.
The unknown constants FHA and ACO were determined experimentally by measuring
absorption in two phantoms (HA and demineralized bone). After scanning a pure HA sample
with a known physical density we determined
AHA phantom = nHA phantom FHA .
AHA phantom is the absorption in the HA phantom in arbitrary units measured in the experiment.
Then FHA = AHA phantom/nHA phantom; nHA phantom is known or can be defined from
formula (2).
Demineralized bone has been chosen as the second phantom. We assumed that changes
in density in the organic part of the bone could not affect the distribution of calcium and that
Quantitative analysis by x-ray microtomography
169
ACO remained constant in every part of the bone. This fraction has been subtracted from the
initial signal, yielding the distribution of calcium in the HA fraction only.
Thus, the HA molecular density in a particular region in a bone can be obtained from the
formula
(Ai − ACO )
(3)
niHA =
nHA phantom .
AHA phantom
The calibration procedure was further extended to local calcium density expressed as atoms
µm−3. Total calcium amount can be expressed in number of atoms (niHA can be easily
converted into niCa knowing the formula of HA) or fraction of the bone weight (dried or wet).
It should be mentioned that local density could not be measured in one pixel/voxel
because significant error may occur due to the partial volume effect. Local density should be
averaged over the statistically representative volume of approximately the same colour. Local
calcium definition is much less accurate than total calcium measurement due to many orders
less statistical information collected in one voxel relative to the whole sample.
2.3. Experimental animals
To validate the method, bones (n = 5) from adult, normal mice were isolated from various
parts of the skeleton, dried and scanned. After the determination of calcium density by the
method described above, the bones were treated and destroyed for an independent analysis of
calcium by atomic absorption (AAS).
As a pilot study, the distribution of the mineral content in bones from newts after a
condition of sustained micro-gravity was studied. Two groups of adult, normal male newts
(Pleurodelus waltlii Michah) were used. The newts were chosen as experimental animals
because of the presence of the ultimobranchial gland responsible for calcitonin production
(Saveliev and Besova 1993, Saveliev et al 1993). The first control group was composed of
animals staying on earth. The experimental group travelled through space for two weeks
aboard the biosputnik (BIOKOSMOS-1887, USSR, 1986).
Upon returning to earth, animals in both groups were sacrificed and preserved for further
investigations. The humeral bones of the newts were selected for analysis by micro-CT, as
bones from the legs have been reported to be more susceptible to conditions of micro-gravity
than others (Saveliev et al 1993).
All bones were dried up to a critical point by a standard technique suitable for preparing
samples for electron microscopy.
Five bones of each group (a total of 10) were chosen for the experiment. They were
approximately 10 mm long and about 1–2 mm in diameter. These geometrical parameters
allowed scanning with an isotropic voxel size of 10 × 10 × 10 µm.
3. Results
3.1. Stability of the x-ray scanner
The first and major prerequisite to use micro-CT as a precise instrument for a quantitative
analysis is the stability of the scanner.
A typical bone was scanned as a test object under the same orientation and in identical
experimental conditions to verify the stability of the measurements. The number of photons
used for the measurements has to be sufficiently high to reduce statistical noise. Several
measurements of the same bone were acquired under the same conditions to estimate the
statistical error.
170
A A Postnov et al
(a)
(b)
Figure 1. Illustration of the stability of the system. Two histograms show the result of scanning
the same bone under identical conditions of acquisition and reconstruction. The curves represent
all consecutive cross-sections through the whole bone. Data were acquired at the beginning (grey)
of the experiment and after one month (black). The histograms are identical except for deviations
due to unavoidable statistical photon noise. For the sake of clarity no error bars were added. (a)
Horizontal axis: grey values (0–255, 0: highest density, 255: lowest density) in the cross-sections,
vertical axis: number of voxels with a given grey value. (b) A histogram from a similar experiment
where grey values were converted into calcium density (calculated from experimental data using
formula (3)). The vertical axis represents the amount of calcium for each density.
The results of the stability test are summarized in figure 1. As expected, an identical
reconstruction was obtained within photon statistics error, when the same bone was scanned
under the same conditions at the beginning of the experiment and after one month (10 scans
Quantitative analysis by x-ray microtomography
(a)
171
(b)
Figure 2. Phantom made of pure HA before and after BHC. Notice that the artefact of denser
surface (a) was compensated by the BHC (b).
each time). From the serial scans, we estimated both statistical error and non-stability to be
less than 0.2%. In figure 1(a), the number of voxels was plotted versus the grey values present
in all the cross-sections. Integrating the surface under the curves in panel (a) results in the total
volume of the bone. In figure 1(b), the amount of calcium versus calcium density is shown;
the area under the curve in panel (b) now represents the amount of calcium in the bone.
3.2. Linearity of the CCD response: beam-hardening correction
The most difficult requirement for a quantitative analysis by micro-CT is the need for a linear
relationship between the recorded signal and real bone density. This effect of ‘nonlinearity’
is due to ‘beam hardening’ resulting from the illumination by a polychromatic x-ray source.
As a result of this effect, the outer surface of the sample usually seems denser than it really
is, whereas the central part of the sample looks lighter. This artefact can seriously affect
quantitative measurements. To avoid this, a correction for the recorded signal is needed before
reconstructing the image. The correction function was acquired from HA phantoms with
different known thicknesses and densities. To correct for beam hardening, a polynomial was
determined based on the relationship between the recorded signal and the object thickness
(figures 2 and 3).
Figure 2 shows two reconstructed virtual slices through a phantom composed of pure
HA before and after applying the BHC. Notice that the artefact of a denser outer surface has
disappeared when the BHC was applied. Scanning of a phantom with a known density resulted
in the measurements of nCa phantom = 6.0 × 109 µm−3 or 2.7 mg cm−3 per single grey value
(density resolution of our method). These values were applied in all calibrations.
Nonlinearity of the detector itself, if present, was compensated by the same polynomial
while correcting for beam hardening.
Figure 3 illustrates the results of implementing the BHC. A test bone was scanned
under two different orientations to change the x-ray attenuation pathway through the sample.
BHC was applied during calculations of all virtual slices obtained through the bone in both
orientations. Panel (a) shows the histograms without any BHC, whereas the optimal correction
is applied to the data in panel (b). Notice that both traces nearly coincide. Measurements
became almost independent of the orientation of the sample.
Demineralized bones were scanned and reconstructed with the same density window as
for normal bones. Thus, the hypothesis that x-ray absorption in the organic (CO) part of the
bone remained relatively constant was confirmed.
To validate the procedure, a series of mice bones (n = 5) were isolated, dried and
scanned. Bones had different shapes and were isolated from different parts of the skeleton.
172
A A Postnov et al
(a)
(b)
Figure 3. Histograms of the same bone scanned under different orientations. The display format
is similar to (a) in figure 1. (a) Without BHC. (b) With precise BHC. The two histograms
almost coincide. Standard deviation from the mean value is indicated by the thickness of the
graph line.
Quantitative measurements of the calcium content in different bones have been summarized in
table 2. After scanning, the bones from the mice were prepared for two independent
measurements of the calcium concentration by atomic absorption spectrophotometry. The
results are also summarized in table 2. The correlation between data obtained by the noninvasive analysis by micro-CT and the established but destructive method of AAS proved to
be acceptable.
Quantitative analysis by x-ray microtomography
(a)
(b)
173
(c)
Figure 4. Three representative cross-sections; (a) the femur of a mouse (cortical bone),
(b) femur of a mouse (trabecular bone), (c) mouse vertebral column (trabecular bone). Note
the different grey values. Images are free of reconstruction artefacts and beam hardening.
Table 2. Summary of calcium content in different bones (every measurement was performed
twice).
Investigated
material
Mouse femur no 1
Mouse femur no 2
Mouse femur no 3
Mouse ribs (totally)
Mouse vertebral column
Newts (control)
Newts (space)
Ca per dry bone
(% by weight)
measured by AAS
Ca per dry bone
(% by weight)
measured by micro-CT
18.09
19.17
20.21
21.12
21.56
22.34
15.53
16.24
12.53
12.64
19.6
19.8
20.4
20.45
21.1
21.4
16.6
16.7
11.3
11.9
7.5 ± 0.5
6.0 ± 0.9 (−20%)
Ca per volume
(g cm−3)
0.25
0.24
0.25
0.19
0.19
0.154 ± 0.005
0.139 ± 0.011 (−10%)
Figure 4 shows three representative reconstructed slices from micro-CT through the shaft
(a) and the top (b) of a mouse femur and a vertebral column (c). Difference between cortical
and trabecular bones can be clearly distinguished.
In the second part of this study, the scanning procedure was applied to a series of bones
from newts exposed to conditions of micro-gravity. Detection of mineral redistribution as a
consequence of space flight was the aim of this investigation. Of each group, five humeral
bones were scanned.
Figure 5 is a classical histological slice of a humeral bone of a newt that remained
under conditions of weightlessness. The osteoclasts can be seen to invade mineralized bone,
causing demineralization. However, with this histological slicing it is not possible to obtain
quantitative data of the real loss of calcium, because during sample preparation calcium has
to be removed from the bone. We could use x-ray micro-CT for this study as a non-invasive
method requiring no treatment of the sample that might influence mineral content.
From a number of consecutive virtual slices, 3D models were created (figure 6). The
spatial distribution of calcium with given densities is illustrated in a representative bone from
174
A A Postnov et al
Figure 5. Bone histology (standard technique) microscopic section of Pleurodeles waltlii Michah
humeral bone, stained by Mallory’s connective tissue stain, no interference. Notice calcium
resorption by osteoclasts. The dark grey colour corresponds to the part of the bone that contained
calcium.
Figure 6. Semitransparent 3D models of two representative bones of the newts; A: control group,
B: ‘space’ group. The spatial distribution of calcium is visualized over the bone. The white colour
represents the spatial distribution of one specific range of calcium density (0.27–0.31 g cm−3)
superimposed on the semitransparent background of the shape of the bone. Notice the decrease in
calcium in the bone that was subjected to micro-gravity.
the control group and the ‘space’ group. Demineralization that took place in the space group
could be localized.
In figure 7, all the data from the newts have been summarized. Histograms allow
estimating changes in volume as well as changes in mineral distribution. The results clearly
show that the bones that had been exposed to micro-gravity suffered from demineralization.
An average demineralization rate of 10% by volume density and 20% by mass fraction was
observed, as shown in table 2.
Quantitative analysis by x-ray microtomography
175
(a)
(b)
Figure 7. Histogram summarizing the calcium distribution in all the newt bones of the control
(black) and ‘space’ groups (grey). (a) Number of voxels versus calcium density, (b) amount of
calcium of every density versus calcium density.
4. Discussion
In the present report, polychromatic x-ray micro-CT was used as an analytical method to
measure mineral density and calcium content in bone. The developed quantitative approach
was based on image analysis of the local grey values in the reconstructed cross-sections. Once
the stability and linearity of the system were verified, a procedure to calibrate the grey values
present in the 8-bit BMP images in the virtual cross-sections of the bone was developed.
As compared to medical CT scanners (Cummings1998), a higher spatial and 3D resolution
was achieved. Thus, very tiny calcified structures could be investigated (Postnov et al 2002b).
176
A A Postnov et al
The resolution of any scan is at least 500–700 times higher than the dimensions of the sample
with a maximum of 8 µm in our present device.
Experimentally, the accuracy of the method has been estimated to be within 2% (mainly
because of beam hardening). This is acceptable and comparable to other methods where
micro-CT or other techniques are applied (Cummings1998, Batchelar et al 2000, Gleason et al
1999). An experimental error remains, due to unavoidable photon statistical noise and some
small discrepancies in linearity that were not fully corrected by the BHC. It is a challenging
problem for future studies to increase the accuracy of this method by increasing acquisition
time and by a more accurate BHC. Special attention should be paid to the creation of HA
phantoms without any structural non-uniformity.
Results of mineral content measurements can be expressed as grams (or atoms) per
volume, or as grams (atoms) per weight (dry weight in the present study). Both definitions
were used. They differ because a bone of the same volume and mineral content can have a
different density and therefore a different weight (table 2).
In this table, independent measurements by AAS were summarized. It is obvious that the
correlation between both the techniques is acceptable. Thus the measurements by AAS
represent a validation of the reliability of the calcium measurements by micro-CT. Our
observations also correlate with reported data when calcium was measured in the dry mineral
bone fraction (Aerssens et al 1997). The major advantage of micro-CT is its non-destructive
nature.
The assumption that x-ray absorption in the organic part of the bone was relatively constant
was confirmed by scanning demineralized bone.
In the literature, differences in the expression of density (Cummings 1998, Ederveen and
Kloosterboer 1999, Beamer et al 1996, Augat et al 1998), together with some discrepancies
in the calibration of the data (Martin and Reid 1999), have been reported.
An important advantage of the present approach is that mineral content or the exact
amount of calcium present in a sample was measured directly. Samples that cannot be
weighed or isolated, such as bones in a living animal, can be investigated. The size of the
samples can vary in a wide range, from 15 mm to 0.1 mm and calcium content down to
10 µg can be detected without losing accuracy. Measurements with micro-CT are fast. The
total time required for acquisition and reconstruction of a complete dataset is in the range
2–4 h.
A number of bones isolated from different anatomical sites in mice were studied. As
reported in humans (Aerssens et al 1997) and animals (Boivin and Meunier 2002), bones
from different anatomical sites have various calcium densities. This may be determined by
the distribution of trabecular and cortical bone.
In addition to the determination of total calcium amount in a bone, the present study opens
perspectives to estimate local calcium density variations in an arbitrary cross-section. In the
virtual slices, a clear distinction can be made between cortical and trabecular bones. However,
the accuracy of the definition of local calcium density still remains a challenge for future
theoretical and experimental investigations. Theoretical error calculations seem difficult due
to the very complicated spectra that are registered by the x-ray detector. While analysing one
cross-section, the noise in every part of the image is determined by photon statistics. This
noise is higher in bone than in a void space (such as air) as part of the signal is absorbed and
the total amount of registered photons decreased.
Besides a direct determination of calcium content, imaging bone density using
3D reconstructions becomes possible. This opens promising perspectives as suggested
previously (Elmoutaouakkil et al 2002). Besides local changes, volume transfer and regional
mineralization can be studied. Other techniques are unable to provide this information.
Quantitative analysis by x-ray microtomography
177
As a pilot study for applying our method,a unique experiment in space was chosen. Microgravity is a remarkable extreme condition inducing changes in mineral distribution in bones
and in the calcium balance of the entire organism (Snetkova et al 1995, Saveliev and Besova
1993, Saveliev et al 1993). The calcium regulating hormone calcitonin is believed to play a
major role in the demineralization process (Saveliev and Besova 1993). In man, calcitonin
is secreted by the C-cells in the thyroid gland. In the newt (Pleurodeles waltlii Michah),
calcitonin is released from an anatomically separated gland, i.e., the ultimobranchial gland.
For this reason the newt was selected as an experimental animal for space flight. The legs of
the animal have been shown to be more affected than other parts of the skeleton (Saveliev et al
1993).
Calcium content in newts that have been aboard the Biosputnik decreased by 10% in
calcium density per volume. This result is in good agreement with previous reports (11% in
Saveliev et al (1993)).
Our study also demonstrated that this demineralization is not uniformly distributed in the
bone. It is more pronounced in the shaft of the humeral bone. This may lead to the conclusion
that the osteoclasts responsible for bone resorption penetrate the bone in the most dense zones
with the highest calcium concentration, when shortage of calcium in blood plasma occurs
during space flight.
As we used unique, experimental material subjected to weightlessness, we were unable to
refer to standard statistical procedures due to the limited amount of bones available. However,
we consider the reported information important because this experiment cannot be repeated
in the near future. Additional investigations will be needed to fully explain the underlying
mechanisms of demineralization under conditions of micro-gravity.
5. Conclusions
In the present study, high resolution x-ray micro-CT was used for a quantitative analysis
of the mineral content in bone. A technique to correlate grey values of the reconstructed
cross-sections to real calcium density in bone was developed. Overall calcium density was
determined in bones from different parts of the skeleton of mice. Measurements of calcium
density were validated by an independent determination of calcium.
As an application, this non-destructive method was applied to bones from newts that had
suffered a period of sustained micro-gravity. Measured demineralization in the bones caused
by a space flight was 10%.
Acknowledgments
The authors wish to thank Dr A Sasov for fruitful and stimulating discussions, Dr H Weinans
for providing demineralized bones and Dr W Van Hul for preparing mice bones. We are
thankful to Professor Luyten and Professor D’Haese for atomic absorption measurements.
This work was supported by European Union grant QLRT-1999-02024 (MIAB).
References
Aerssens J, Boonen S, Joly J and Dequeker J 1997 Variations in trabecular bone composition with anatomical site
and age: potential implications for bone quality assessment J. Endocrinol. 155 411–21
Augat P, Fuerst T and Genant H K 1998 Quantitative bone mineral assessment of the forearm: a review Osteoporos.
Int. 8 299–310
178
A A Postnov et al
Batchelar D L, Dabrowski W and Cunningham I A 2000 Tomographic imaging of bone composition using coherently
scattered x-rays Proc. SPIE (Medical Imaging 2000: Physics of Medical Imaging) 3977 353–61
Beamer W, Donahue W G, Rosen L and Baylink C 1996 Genetic variability in adult bone density among inbred
strains of mice Bone 18 397–403
Boivin G and Meunier P J 2002 The degree of mineralization of bone tissue measured by computerized quantitative
contact microradiography Calcif. Tissue Int. 70 503–11
Bonse U 1997 Rontgen-mikrotomographie Phys. Bl. 53 211–4
Bowman S M, Zeind J, Gibson L J, Hayes W C and McMahon T A 1996 The tensile behavior of demineralised bovine
cortical bone J. Biomech. 29 1497–501
Boyde A, De Clerck N and Sasov A 2000 Micro-CT of bones and soft tissues Microsc. Anal. (European edn) 3 41
Chantler C T 1995 Theoretical form factor, attenuation, and scattering tabulation for Z = 1–92 from E = (1–10 eV)
to (0.4–1.0 MeV) J. Phys. Chem. Ref. Data 24 71
Cummings S R 1998 Review of the evidence for prevention, diagnosis and treatment and cost-effective analysis
Osteoporos. Int. 8 (suppl. 4) S1–S88
Driessens FC and Verbeeck R H 1990 Biomaterials (Boca Raton, FL: CRC Press) ch 10–14
Ederveen A G H and Kloosterboer H J 1999 Tibolone, asteroid with a tissue-specific hormonal profile, completely
prevents ovariectomy-induced bone loss in sexually mature rats J. Bone Miner. Res. 14 1963–70
Einhorn T A 1992 Bone strength: the bottom line Calcif. Tissue Int. 51 333–9
Elliott J, Davis G R, Anderson P, Wong F, Dowker S and Mercier C 1997 Application of laboratory microtomography
to the study of mineralised tissues An. Quim. Int. Edn. 93 S77-S82
Elmoutaouakkil A, Peyrin F, Elkafi J and Laval-Jeantet A M 2002 Segmentation of cancellous bone from highresolution computed tomography images: influence on trabecular bone measurements IEEE Trans. Med. Imaging
21 354–62
Feldkamp L A, Davis L C and Kress J W 1984 Practical cone-beam algorithm J. Opt. Soc. Am. A 1 612–9
Gleason S S, Sari-Sarraf H, Paulus M, Johnson D, Norton S and Abidi M 1999 Reconstruction of multi-energy x-ray
computed tomography images of laboratory mice IEEE Trans. Nucl. Sci. 46 1081–6
Henke B L, Gullikson E M and Davis J C 1993 X-ray interactions: photoabsorption, scattering, transmission, and
reflection at E = 50–30 000 eV, Z = 1–92 At. Data Nucl. Data Tables 54 181–342
Herman G T 1980 Image Reconstruction from Projections (New York: Academic) p 317
Martin J C and Reid D M 1999 Radial bone mineral density and estimated rates of change in normal Scottish women:
assessment by peripheral quantitative computed tomography Calcif. Tissue Int. 64 126–32
Ming Ding 2000 Age variations in the properties of human tibial trabecular bone and cartilage Acta Orthop. Scand.
71 45
Odgaard A 1997 Three-dimensional methods for quantification of cancellous bone architecture Bone 20 315–28
Postnov A, De Clerck N, Sasov A and Van Dyck D 2002b 3D in vivo x-ray microtomography of living snails
J. Microsc. 205 201–5
Postnov A, Van Dyck D, Saveliev S, Sasov A and De Clerck N M 2002a Definition of local density in biological
calcified tissues using x-ray microtomography Proc. SPIE Meeting: Medical Physics (San Diego, 23–28
February 2002) vol 4681 pp 749–56
Ruegsegger P, Elsasser U, Anliker M, Gnehm H, Kind H and Prader A 1976 Quantification of bone mineralization
using computed tomography Radiology 12 93–7
Ruegsegger P, Koller B and Muller R 1996 A microtomographic system for the nondestructive evaluation of bone
architecture Calcif. Tissue Int. 58 24–9
Sasov A and Van Dyck D 1998 Desktop x-ray microscopy and microtomography J. Microsc. 191 151–8
Saveliev S V and Besova N V 1993 The influence of weightlessness on amphibians: I. Ultimobranchial body Bull.
Exper. Biol. Med. 7 87–90
Saveliev S V, Besova N V and Chernicov V P 1993 The influence of weightlessness on amphibians skeleton and
mineral metabolism Bull. Exper. Biol. Med. 7 90–2
Snetkova E, Chelnaya N, Serova L, Saveliev S, Cherdanzova E, Pronych S and Wasserzug R 1995 Effects of space
flight on Xenopus laevis larval development J. Exp. Zool. 273 21–32
Stenstrom M, Olander B, Lehto-Axtelius D, Madsen J E, Nordsletten L and Carlsson G A 2000 Bone mineral density
and bone structure parameters as predictors of bone strength: an analysis using computerized microtomography
and gastrectomy-induced osteopenia in the rat J. Biomech. 33 289–97
VanDyck D and Sasov A 1998 Desk-top microtomograph: gateway to the 3D-world Microsc. Anal. (European edn)
3 21–3