Current Medical Imaging Reviews, 2007, 3, 45-59 45

1573-4056/07 $50.00+.00 2007 Bentham Science Publishers Ltd.

Small Animal Computed Tomography Imaging
Soenke H. Bartling
1,2,*
, Wolfram Stiller
2,*
, Wolfhard Semmler
2
and Fabian Kiessling
1,2
1
Junior Group Molecular Imaging,
2
Department of Medical Physics in Radiology, German Cancer Research Center
(DKFZ), Heidelberg, Germany
Abstract: Micro Computed Tomography (micro-CT) was suggested in biomedical research to investigate tissues and
small animals. Its use to characterize bone structures, vessels (e.g. tumor vascularization), tumors and soft tissues such as
lung parenchyma has been shown. When co-registered, micro-CT can add structural information to other small animal
imaging modalities. However, due to fundamental CT principles, high-resolution imaging with micro-CT demands for
high x-ray doses and long scan times to generate a sufficiently high signal-to-noise ratio. Long scan times in turn make the
use of extravascular contrast agents difficult. Recently introduced flat-panel based mini-CT systems offer a valuable trade-
off between resolution (~200 m), scan time (0.5 s), applied x-ray dose and scan field-of-view. This allows for
angiography scans and follow-up examinations using iodinated contrast agents having a similar performance compared to
patient scans. Furthermore, dynamic examinations such as perfusion studies as well as retrospective motion gating are
currently implemented using flat-panel CT.
This review summarizes applications of experimental CT in basic research and provides an overview of current hardware
developments making CT a powerful tool to study tissue morphology and function in small laboratory animals such as
rodents.
Keywords: Small animal imaging, Computed Tomography (CT), micro-CT, mini-CT, flat-panel detector, motion gating.
INTRODUCTION
Micro- and mini-CT are scaled down CT-imaging moda-
lities for small animals, which in principle provide the same
information about morphology and disease status or disease
progression for animals as clinical-scale CT does for
humans. Nonetheless, several major differences compared to
clinical CT scanning exist.
This review provides an overview of basic concepts and
techniques that apply to CT when used for in-vivo small
animal imaging. Technical and physical limitations of these
concepts as well as applications of micro- and mini-CT and
an outlook on future developments in the growing field of
research in small animal CT imaging are given and
discussed.
Definition
There is no unique definition of mini-CT and micro-CT.
Often, all CT scanners that provide higher spatial resolution
than current clinical scanners are named micro-CT. This
definition results in a very broad range of different CT
scanner design concepts ranging from CT scanners designed
and used for non-destructive material-testing with a
resolution in the order of a few m up to dedicated small
animal CT scanners with a resolution of a few hundred m.
Taking the different design constrains of these various types
of scanners into account, the term mini-CT should be used to
describe CT systems with a resolution ranging from 100 m
*Address correspondence to either author at the Molecular Imaging Group,
Department of Medical Physics in Radiology, German Cancer Research
Center (DKFZ), Im Neuenheimerfeld 280, 69120 Heidelberg, Germany;
Tel: +49 (0) 6221 422686; Fax: +49 (0) 6221 422557; E-mail:
s.bartling@dkfz.de; w.stiller@dkfz.de
*These authors contributed equally to this article.
to 500 m; micro-CTs are then scanners with a resolution of
below 100 m [Kalender W (2006), personal communica-
tions] [1] (Fig. (1), Table 1).
Small Animal Imaging
Due to the fact that more and more research is based on
animal models the interest in in-vivo small animal imaging
rises. With small animal imaging methods cross-sectional
study designs (e.g. where cohorts of animals are killed at
each time point and histology is examined) can be replaced
by an in-vivo study design permitting repeated measurements
of the same animals. Each animal acts as its own control,
therefore the variability that would normally be present
within a cohort of animals is accommodated and changes are
more readily detected by paired comparisons [2].
Additionally, CT imaging scaled down to small animals can
help to define, optimize and test future performance goals of
clinical-scale CT.
Scaling down CT imaging to the size of small animals is
challenging: The volumes of mammals morphological
structures and organs are proportional to their weight [3]. So,
to acquire CT data, e.g. of mice, that show their internal
organs with detail comparable to clinical CT scans of human
patients, a small animal imager needs a resolution of about
100 m [1].
For motion-compensated imaging (i.e. taking into
account cardiac and respiratory motion present during image
acquisition) stakes are similar. The heart rate of a mouse is
about 400-600 min
-1
and respiration frequency ranges from
30-60 per minute. To use the diastole as the phase of the
heart cycle that shows a minimum amount of motion, a basal
temporal resolution of the CT scanner in the order of 50 ms
in mice in comparison to 300 ms in humans is necessary [4].
46 Current Medical Imaging Reviews, 2007, Vol. 3, No. 1 Bartling et al.
BASIC DESIGN CONCEPTS
Several design factors of CT scanners such as geometry,
the employed x-ray source and the detector technology
influence the fundamental characteristics of CT scanners.
Inherent relationships between resolution (spatial, temporal
as well as soft-tissue contrast), noise, scan time, scan field-
of-view (FOV) and applied x-ray dose are imposed by the
fundamental laws of physics. Therefore there is not one
scanner design that is able to optimize all of these
fundamental scanner characteristics. Instead, several diffe-
rent scanner designs exist which are suited for different
kinds of diagnostic questions (Fig. (1), Table 1). Fundamen-
tal CT design concepts will be discussed in the next
paragraphs.
Fig. (1). Examples for three different design concepts of small animal CT scanners. A bench-top micro-CT (A, B) with rotating sample
holder, stationary area detector and micro-focus x-ray source offering variable magnification. Such a setup is mostly used for in-vitro
imaging. An optimized relationship between scan field-of-view and resolution as well as good animal handling due to a non-rotating sample
bed is provided by a rotating gantry based concept (C, D). Imposing fewer demands on spatial resolution, faster scanning and a bigger scan
field-of-view can be achieved with the displayed flat-panel detector, rotating gantry-based design with stationary sample bed (E, F).
Reprinted with permission from Willi Kalender, Computed Tomography, Publicis Corporate Publishing.
Small Animal Computed Tomography Imaging Current Medical Imaging Reviews, 2007, Vol. 3, No. 1 47
Rotating Sample vs. Rotating Gantry Systems
Small animal CT-scanners can be classified into two
major categories: rotating sample and rotating gantry
systems [5].
So-called rotating sample systems feature a stationary x-
ray source and a stationary x-ray detector usually mounted
on a mechanical bench. Both detector and x-ray tube are
facing each other and have to be precisely aligned with the
central axis of the CT system. A rotating sample holder is
placed on the mechanical bench in between source and
detector, also aligned with the central axis of the scanner
system. The sample holder can usually be precisely rotated
perpendicular to the central axis of the scanner system by a
computer-controlled motor-driven rotating stage (Fig. (1), A,
B). For in-vivo animal imaging this implies that the
anesthetized animal has to be immobilized and fixated in an
upright position (e.g. head up) on the rotating sample stage
which demands careful animal handling [6]. In addition,
administration of inhalation anaesthetics and injection of
contrast agents during a scan is difficult since the needed
apparatus may not inhibit or conflict with the sample
rotation.
Most of the rotating sample systems have the advantage
that the scanner geometry can easily be modified in between
scans. The source-to-isocenter (also called source-to-object)
distance (SOD) can be varied by shifting the x-ray tube
along the central axis of the system relative to the sample
holder defining the isocenter of the scanner. The source-to-
detector distance (SDD) can be varied by either shifting only
the detector relative to the scanners isocenter along its
central axis while leaving the source fixed or by moving both
source and detector relative to the isocenter. This flexible
design enables the user to choose the optimum scanner
geometry for the imaging application.
Systems with rotating gantry but stationary sample
accommodate x-ray source and detector mounted exactly
facing each other on the inside of a ring-shaped mechanical
support (gantry). Here, the gantry containing source and
detector is no longer stationary but rotates around the central
axis of the scanner. The sample which is to be imaged can be
placed in a prone or supine position on a patient bed or
patient table. (Fig. (1), C-F). The longitudinal axis of this
table is parallel to the central axis of the CT scanner and the
table is usually driven by a computer-controlled micrometer
stage, allowing exact and reproducible positioning of the
sample within the scanner system. Small animal CT systems
featuring a rotating gantry assembly can easily be identified
as scaled-down clinical CT scanners adapted to the specific
requirements of small animal imaging (Fig. (1), C-F).
Compared to scanners having a rotating sample holder,
rotating gantry based systems usually do not offer the
possibility to change the scanner geometry easily, normally
SDD and SOD are fixed numbers of the particular scanner
system. Animal handling during examination is much easier
in rotating gantry systems. In addition, these systems support
faster rotation times than rotating sample systems. However,
they usually are more complex mechanically and more
intricate from the point of view of systems control and
therefore more expensive.
No matter if the small animal CT scanner is rotating
gantry based or not, both system types can be either bench-
top systems as shown in Fig. (1) A-D or systems requiring
mechanical support structures too large to be scaled to fit a
bench top, e.g. if modified clinical x-ray tubes offering
higher x-ray photon fluxes than micro-focus tubes or very
large area detectors e.g. to cover large FOV sizes (Fig. (1),
Table 1. Comparison of Micro-, Mini- and Clinical-Scale CT
Micro-CT Mini-CT Clinical-scale CT scanner
Suited for Tissue samples, insects, mice, rats Mice, rats, rabbits, primates,
mini-pigs
Up to humans
Spatial resolution (isotropic) 5 m(single limbs) 100 m(whole animals) 100-450 m >450 m(z-axis >600 m)
Transaxial scan field-of-view
(FOV)
1-5 cm 5-20 cm >20 cm
Time to acquire a standard
volume (e.g. a whole animal)
Seconds to hours (CT scanners with single
slice acquisition within subsecond times exist)
Subsecond (0.5 s) to a few
seconds
A few seconds (with rotation
times down to 0.33 s)
Radiation dose >1 Gy can be reached ~10-500 mGy <50 mGy
Design Bench-top, rotating sample (with variable
geometry, resolution, scan field-of-view, etc.)
or rotating gantry
Rotating specimen or rotating
gantry (fixed geometry)
Rotating gantry (fixed geometry)
Cardiac- & respiratory motion
compensation
Prospective triggering Prospective triggering,
retrospective gating
Scan modulation, retrospective
gating
Example figures Fig. (1) A, B, C, D, (3), (4) Fig. (1) E, F, (2), (5), (6)
Overview of different in-vivo small animal CT scanner designs (values given areapproximations). Thetermmicro-CT is often used for all scanners that aresmaller than clinical
scalescanners, however, taking theperformancedifferences into account thetermmini-CT could beused for scanners with a resolution of around 100 m.
48 Current Medical Imaging Reviews, 2007, Vol. 3, No. 1 Bartling et al.
E, F) are to be used. An overview over commercial and
experimental laboratory micro-CT scanners of both types can
be found in [7], a compilation of the most recent small
animal CT scanner models, including those featuring flat-
panel detector systems can be found in [8].
Scanner Geometry
Independent of the x-ray source employed or the detector
technology chosen for a particular small animal CT-scanner
there are two possibilities for CT scanner geometry: the so-
called short scanner geometry where the SOD is small
compared to the object-to-detector distance (ODD) and the
so-called long scanner geometry where SOD and ODD
have about the same size [Kachelriess M (2006), personal
communication]. Both geometries have their advantages and
disadvantages on the imaging properties of the scanner but
share the requirement that the small animals imaged should
be entirely covered by the available FOV.
Short scanner geometries place the animal close to the x-
ray source, the ODD being larger than the SOD. The image
projected onto the detector can thereby be magnified by the
factor
M =
ODD
SOD
,
improving spatial resolution of the scanner system.
However, the resulting FOV is being decreased by the same
factor requiring larger detectors. With the advent of large-
area flat-panel detectors higher magnification factors will be
possible, since the active detector area will increase
significantly. Placing the animal close to the x-ray source,
however, requires the use of micro-focus x-ray tubes since
larger focal spot size will introduce a significant image blur
(so-called penumbral blurring), which degrades resolution.
The problem of penumbral blurring is aggravated by high
magnification of scanner systems since the penumbra of the
sources focal spot is also magnified by the factor M.
Long scanner geometries also have their advantages.
Image blur caused by the finite size of the x-ray source focal
spot is decreased if the object is placed closer to the detector
instead of being placed close to the source. However, the x-
ray photon flux will decrease exponentially with
1
SOD
2
,
which can be compensated by x-ray tubes with higher output
(but normally bigger focal spot because of anode heating).
Otherwise, the scan time per projection needs to be increased
or contrast resolution will be degraded since (statistical)
image noise (being related to the attenuated photon flux

att
reaching the detector by
1

att
) will increase.
The larger the distance between source and object (SOD)
is, the smaller the skin entrance dose to the animal imaged
will be. This implies that although scanners with short
geometry usually are more dose efficient and allow to
geometrically magnify the object, thereby improving spatial
resolution, long scanner geometries can significantly reduce
the skin entrance dose of the animal under investigation.
X-Ray Source
The scanner geometry and the desired spatial, low-
contrast spatial and temporal resolution govern the choice of
x-ray tubes for small animal CT imaging. An x-ray source
for micro- and mini-CT has to fulfill three demands: the
focal spot size has to be as small as possible, the source has
to emit a high photon flux , and x-ray energies should be
selectable over a suitable range.
As already mentioned above the focal spot size should be
as small as possible in order not to have a negative influence
on spatial resolution by image blurring. However, the size of
the focal spot limits the available x-ray photon flux. For
small focal spot sizes and stationary targets, heat dissipation
from the target area is proportional to the diameter of the
focal spot and radial to first order, the maximum power
emission being
P
max
1.4 x
f ,FWHM
( )
0.88
, with x
f ,FWHM
being the focal spot size in microns [9]. Since the emitted x-
ray photon flux is roughly proportional to the product of
the x-ray anode current I and the square of the tube voltage
U and the tube power is P = UI, the available x-ray flux is
limited by the size of the focal spot, which can only absorb a
certain amount of heat (anode heating). The maximum power
that an x-ray tube can emit thus also depends on the heat
capacity of the anode material and the tube technique used:
tubes with rotating anode support much higher output than
tubes with stationary anodes since heat is evenly distributed
along the focal spot trajectory.
High x-ray flux is desirable for small animal CT imaging
to achieve high temporal resolution and short scan times: if
the photon flux is high enough, sufficient x-ray photons
reach the detector and can be collected in short times for
each projection. A sufficient amount of x-ray photons is
required to limit image noise and allow good low-contrast
spatial resolution [6]. Through short scan times in-vivo
scanning of animal subjects is eased since the animals do not
have to be anesthetized for long times. Also, high photon
fluxes are needed to ensure high temporal resolution for
perfusion studies or motion compensated imaging (ECG or
respiratory gated imaging).
In order to achieve the best low-contrast spatial
resolution in small animal CT imaging, the x-ray tube should
allow choosing a range of tube voltages, thereby selecting
appropriate x-ray energies E. For higher x-ray energies the
energy-dependent absorption coefficient (E) is small and
low-contrast spatial resolution is limited due to the small
number of x-ray photons absorbed in the animal; if x-ray
energy is low and thus (E) large, most photons are absorbed
Small Animal Computed Tomography Imaging Current Medical Imaging Reviews, 2007, Vol. 3, No. 1 49
in the animal and the contrast resolution is limited by the
small amount of x-ray photons reaching the detector [9]. In
small animal CT imaging photon energies in the range of 30-
50 kV are commonly applied. Hereby, the advantage is that
x-ray attenuation is up to two orders of magnitude higher
than in clinical CT systems usually operated at 80-140 kV
which allows better discrimination of soft tissue types [1].
These low-energy x-ray photons are still able to penetrate the
animal since its examined volume is much thinner than that
of human patients. Additionally, the highest absorption
differences of iodine lie in the low energy range, leading to
an improvement in contrast-enhanced scans for the contrast
between iodinated contrast agents to the surrounding tissue.
For small animal CT scanning three different types of x-
ray sources have been used so far: micro-focus x-ray tubes,
diagnostic x-ray tubes as in clinical-scale human patient
radiological equipment and diagnostic x-ray tubes modified
to fulfill the special requirements of small animal CT.
Micro-focus x-ray tubes with focal spot sizes ranging for
5-50 m (~5-200 W power emission) have mostly been
employed in micro- and mini-CT and have the advantage of
allowing high isotropic spatial resolution down to a range of
10-50 m for FOV sizes of 30-50 mm [7]. Others have
employed diagnostic x-ray tubes with larger focal spot sizes
(0.3 mm at 9 kW and 1.0 mm at 11 kW) in order to profit
from the high x-ray flux at short exposure times [6, 10, 11]
to increase temporal resolution for cardiac and respiratory
gating in small animal CT while still achieving an isotropic
spatial resolution of 100 m. Experimental small animal CT
systems with modified diagnostics x-ray tube exist [12].
Here, the tube filament was shortened to decrease the focal
spot size of the rotating anode x-ray tube while still offering
considerable photon fluxes which make high scan speed
possible and are needed for the flat-panel area detector.
Detector Technology
Detector technology plays a crucial role for the
performance characteristics of all small animal CT systems.
Together with the focal spot size of the x-ray source the size
of the detector elements has the biggest influence on the
spatial resolution of the images, their size also influences the
low-contrast resolution together with x-ray flux.
No matter what detector technology is used (see below)
in a particular small animal CT system, all of its parts (e.g.
image intensifier screen, optical coupling medium, photo
detector, etc.) have to be optimized for the following factors:
high quantum efficiency and uniform response throughout its
whole surface at the chosen x-ray energies and good pixel
resolution through small detector pixelation. The detectors
dynamic range needs to be as large as possible to ensure
good low-contrast resolution and high dose efficiency,
whereas its noise and dark current should be as low as
possible. A high read-out rate is favorable for the systems to
achieve a maximum temporal resolution and to minimize
scanning times. Of course the area of the detector should be
as large as possible so large FOVs can be employed and
none of the detector elements should introduce geometrical
distortions [9].
From the first developments of small animal CT systems
onwards different detector types have been used. Apart from
an overview over the different types that have been used to
date we limit our focus to the emerging flat-panel x-ray
detectors, which will be elaborated on in more detail. For a
detailed look on digital x-ray detector technology please
refer to [13].
As in clinical-scale CT systems, detector systems in
micro- and mini-CT can consist of linear arrays of
photodiode detector elements illuminated by a so-called fan
beam. This approach to small animal CT was used in some
of the early systems but is still used by some commercial
small animal CT manufacturers [7]. After the development
of the so-called cone-beam reconstruction algorithm [14],
the first three-dimensional small animal CT systems were
described end of the 1980s and beginning of the 1990s until
the early 2000s. They employed two-dimensional detectors
consisting of phosphor screens which were optically coupled
to cooled charge-coupled detector (CCD) arrays via an
optical lens or a fiber-optic taper (e.g. [15, 16]). Though
being inefficient, the optical coupling via lenses offers the
possibility of variable image magnification, whereas fiber-
optic coupling is very efficient but uses fixed magnification
[1]. This detector setup is technically demanding but remains
one of the most sensitive x-ray detection methods [7].
Holdsworth and co-workers (1993) used x-ray image
intensifier (XRII) screens coupled to a video readout [17].
The rapid illumination response to x-ray radiation is the
main advantage of image intensifiers [1]. Since the 1990s
CCD-based detector arrays with scintillating plates (e.g.
made of CsI(Tl), GdO
2
SO
4
, etc.) have become common in
small animal CT systems [6]. Recently, flat-panel area
detectors have been introduced to small animal CT imaging
[10, 12, 18-20].
An overview of indirect (x-rays are converted to light in
the scintillating layer first, then the light is detected by
photodiodes and subsequently amplified) and direct
conversion (x-ray photons are directly detected) flat-panel
detector types and their technical differences can be found in
[21]. The flat-panel detectors used in recent flat-panel small
animal CT systems are of the indirect conversion type. They
usually consist of a scintillating crystal layer like thallium-
doped caesium iodide (CsI(Tl); or GdO
2
SO
4
, etc.) coupled to
a photodiode array in form of an amorphous silicone (a-Si)
wafer [10, 22, 23]. Others have employed photodiode arrays
fabricated by a complementary metal-oxide semiconductor
(CMOS) process [18, 19]. Compared to XRII detectors flat-
panel detectors have some advantages: their structure is thin
and they allow for large-area detection (see below) without
geometrical distortions [24]. Because of technological
advances they can be produced in good quality and with high
precision. The hydrogenated a-Si material used has the
advantage that it can be deposited on very large areas
(making large FOVs possible) at relatively low temperatures
(200-250 C, easing fabrication), has the properties of a
semiconductor (photoconductivity in the visible spectral
range) and does not age through x-ray exposure [25], which
is a drawback of CCD- and CMOS-based detectors.
However, a-Si flat-panel detectors have several draw-
backs: the pixel size of around 100-200 m
2
is relatively
large compared to pixel sizes of ~2.5 m
2
achievable with
CCD detectors [16], due to low fill-factors of ~45-70 %
50 Current Medical Imaging Reviews, 2007, Vol. 3, No. 1 Bartling et al.
([18, 26]) the geometric efficiency is not as high as in CCD
detectors, and the so-called image lag (ghost images of
previously acquired projections due to charge traps in the a-
Si layer, e.g. dangling bonds in the amorphous structure) has
a negative influence on spatial, low-contrast and temporal
resolution [27]. The effects of image lag can be quite serious
on image quality. They are described for flat-panel computed
tomography in general in [28]. Algorithms are applied to
correct for the effect of image lag; projection frames pre-
viously acquired are weighted and subsequently subtracted
from the currently acquired frame [23]. Compared to other
detector types, flat-panel detectors in small animal CT also
suffer from their smaller dynamic range which can
deteriorate image quality: insufficient resolution of pixel
gain and offset normalization causes ring artefacts and
truncation of low-density anatomical detail due to
insufficient dynamic range results in shadow artefacts and
incorrect Hounsfield numbers (HU) [29]. In addition, flat-
panel detectors need frequent and careful recalibration for
offset and pixel gain factors [23, 29].
Nonetheless, the advantages of flat-panel detectors used
in small animal CT imaging by far outweigh the disadvan-
tages stated and cited here. Further developments in flat-
panel detector technology ([21, 29]) will allow even more
studies with and new applications of small animal CT. The
use of other detector materials like CdTe or CdZnTe for flat-
panel detector fabrication might increase detector efficiency
for the x-ray energies employed and thus lead to improved
soft tissue contrast [30].
Spatial Resolution
In small animal CT spatial resolution is one of the key
parameters since the size of morphological structures is
approximately proportional to the weight of the (laboratory)
animal examined. This leads to the requirement that small
animal CT systems should be capable of providing a spatial
resolution in the order of 100 m. Spatial resolution of small
animal CT scanner systems is mainly determined by the
following factors [9]: the geometry of the scanner system,
which can be chosen to allow geometrical magnification of
the specimen imaged, the size of focal spot having an
influence on the magnitude of penumbral blurring (see
below) and the chosen detector technology, which
determines the minimum image pixel or voxel size through
its intrinsic resolution. Thus, the size of the single detector
elements sets the maximum spatial resolution a particular
micro- or mini-CT can theoretically achieve, even if image
reconstruction features image matrices with smaller pixel or
voxel sizes. Reconstruction voxel sizes chosen too large
compared to the intrinsic detector pixel size reduce the
spatial resolution of a CT scanner, while very small
reconstruction voxels size do not increase its intrinsic
resolution. Ideally, the reconstruction voxel size should be
chosen to be half the size of the intrinsic resolution of a CT
scanner complying with Nyquists sampling theorem.
Per definition spatial resolution is the smallest distance
between two objects still being distinguishable as two
separate entities. In order to describe the spatial resolution of
CT images obtained from a particular scanner, the
modulation transfer function (MTF) is employed. The MTF
plots the percentage of contrast that is transferred from the
imaged object (e.g. a line-pair phantom) to the CT image as a
function of spatial frequency (expressed as the number of
line pairs per unit length). It is common use to state the
spatial frequency which corresponds to 10 % contrast in the
CT image as the spatial resolution of the scanner. I.e.: the
finer the CT image detail still resolvable the higher the
spatial resolution of the scanner and thus the spatial
frequency at 10 % modulation transfer [12].
For small animal CT scanners the so-called slant-slit
method is often used to calculate the MTF of a scanner [18,
19]. Here the phantom consists of an acrylic plate having a
rectangular slit covered by a very thin metal foil (e.g. 18 m
of aluminium [18]). The cross-section of the foil is the slit
which can be slightly tilted with respect to the horizontal or
vertical axis of the image plane so that the line-profile of the
slit in the image domain can be sampled at sub-pixel level.
Then the MTF can be calculated as the Fourier
transformation of the line profile.
In practice, the spatial resolution of clinical scale CT
scanners and small animal CT flat-panel scanners featuring a
clinical-scale gantry is usually measured by scanning line-
pair phantoms, which provide sets of metal line-pairs (e.g.
tungsten bars) of various thicknesses at decreasing distances
cast into plastic material. Here the smallest feature size (in
millimeters) that is still visible can easily be calculated by
dividing 10 mm by the amount of distinguishable line-
pairs/cm times two.
The maximum theoretically achievable spatial resolution
according to the employed detectors intrinsic resolution and
the geometrical magnification of a particular small animal
CT system can seldom be achieved [5]. Apart from the
mechanical stability (e.g. vibrations, etc.) of the scanner and
the influence of the reconstruction algorithm, the image blur
introduced by the finite size of the focal spot ( x
f ,FWHM
) of
the x-ray tube limits the spatial resolution of the real system
(penumbral blurring is
b =
ODD
SDD ODD
x
f , FWHM
=
ODD
SOD
x
f ,FWHM
[6]). To reduce the influence of the focal spot shape and the
associated penumbral blurring Popescu and co-workers [23]
have modified and measured the focal spot shape and size of
the customized clinical x-ray source employed in the flat-
panel CT used for small animal imaging [31]. The
measurements of Popescu and co-workers [23] included an
investigation of the focal spot shapes throughout the whole
detector plane. These authors have shown that small
distortions of the focal spot introducing a negligible amount
of image blurring can still be observed even if the tube is
modified to match source and detector setup.
In order to achieve the spatial resolution of interest,
fluxes of 104-105 photons have to traverse a particular
region of interest (ROI) per resolution unit at minimum [32]
so a detector signal significantly larger than the statistical
noise due to the quantum nature of the x-ray photons can be
gained. Higher photon fluxes allowing shorter exposures
times (and thus better temporal resolution) require larger x-
Small Animal Computed Tomography Imaging Current Medical Imaging Reviews, 2007, Vol. 3, No. 1 51
ray tube focal spot sizes but dictate a relaxation of the
requirements of spatial resolution and geometrical
magnification of a small animal CT scanner system.
Low-Contrast Resolution
The highest spatial resolution achievable in small animal
CT is of little use unless the scanner is able to resolve small
differences between structures and neighbouring tissues
having a poor inherent contrast. In any CT system, contrast
in the images is a consequence of the difference in the
energy-dependent linear absorption coefficient (E) (also
called attenuation coefficient) of different tissues [7].
Thus, low-contrast detectability (LCD) can be defined as the
ability to detect fine variations in the (electron) density of an
object over the background [12].
The low-contrast resolution capabilities of a small animal
CT scanner largely depend on the type of detector used: the
more quantum efficient the detector is, the lower the amount
of incident photons to generate a detector signal can be; and
the larger the detectors dynamic range is, the smaller the
differences in photon flux still detectable and differentiable
can be. This implies that the low-contrast resolution
capability of a scanner also depends on the x-ray dose
applied per projection (view) of a scan (see below), since
resolving low-contrast structures actually means collecting
detector signals whose difference is significantly larger than
the noise level present. Micro-CTs that are not specifically
designed for small animal imaging usually have inherent
limitations in signal-to-noise ratio (SNR) performance
because of their small detector voxel size (to achieve high
spatial resolution) and their low x-ray exposure level (caused
by the employed micro-focus x-ray tubes). Therefore, their
low-contrast resolution is traded for high spatial resolution of
high-contrast structures such as bones [18]. Other
investigators trying to achieve high low-contrast detectability
have employed x-ray tubes featuring large foci [6, 33]. In
addition, a careful selection of the x-ray energy appropriate
to the imaging task is necessary if low-contrast detectability
should be optimal.
Experimental evaluations of the low-contrast resolution
of small animal CT systems usually employ custom made
contrast phantoms. These may consist of water-filled acrylic
hollow cylinders with small cylindrical low-contrast inserts
with densities close to the density of water [18]. They allow
to either determine the minimum dose level at which the
low-contrast inserts can be discriminated from the surroun-
ding water or used to quantify the low-contrast resolution
performance of a small animal CT at a fixed x-ray exposure
by measuring its contrast-to-noise ratio (CNR):
CNR
i
=
S
i
S
b

i
2
+
b
2
,
with S and being the mean CT number and the standard
deviation of the image pixel values in a specific ROI in HU,
the subscripts stand for the inserts (i) and the background (b)
[19]. Ideally, the CNR in CT images is proportional to the
square root of the applied dose [18, 34].
Contrast agents are used in small animal CT like in
clinical CT to increase low-contrast resolution. The increase
in CT number (CT Number) caused by the contrast agent
applied can be approximated by
( )
( )
C
E
E
Number CT
water
medium contrast

([9, 35]).
Here C is the concentration of the contrast agent in the
tissue (in mg/ml) and ( = cm
2
g [ ]
) is the
attenuation coefficient normalized for density.
Image Reconstruction
In small animal CT the same image reconstruction
techniques as in clinical-scale CT are applied. CT systems
using fan-beam geometries apply fan-beam reconstructions
based on filtered back-projection algorithms originating from
the early days of clinical CT. The recently introduced flat-
panel based micro- and mini-CTs employ variants of the
Feldkamp-David-Kress (FDK) algorithm [14] for cone-beam
reconstruction. Both reconstruction techniques have to be
adapted to the specific small animal scanner geometry. An
overview over the underlying principles of image
reconstruction can be found in [9, 35].
Contrast Media
Non-ionic extravascular water-soluble contrast agents, as
typically used for clinical CT examinations, rapidly clear
from the blood within minutes after intravenous injection.
The first-pass of the contrast media with the highest vascular
contrast is only retained for a few heart beats. This time
frame is too short for image acquisition with most micro-CT
scanners.
Contrast media that remain in the vasculature for longer
are called blood-pool contrast agents. In order to achieve
blood-pool effects the size of contrast media molecules can
be increased to larger sizes than that of capillary fenestra-
tions. Furthermore, phagocytosis by the reticuloendothelial
system should be avoided as long as possible through the
chemical design of the contrast media. Several different
agents have shown potential as blood-pool agent for CT
imaging [36].
A water-soluble macromolecular agent (dysprosium-
DTPA-dextran) was reported to provide blood-pool contrast
enhancement for up to 45 minutes [37]. Liposomal encapsu-
lation of iohexol resulted in a blood-pool effect for up to 3
hours after injection in rabbits [38]. In rats, iodine-containing
micelles caused enhancement in the blood, liver and spleen
for more than 3 hours [39].
An iodinated triglyceride emulsion (ITG-LE) packed into
the lipophilic core of a synthetic chylomicron remnant has
been described in [40]. It can be used to image liver
parenchyma because it is internalized by the hepatocytes,
whereas liver tumor cells do have less functional lipoprotein
receptors and therefore only show low enhancement. The
52 Current Medical Imaging Reviews, 2007, Vol. 3, No. 1 Bartling et al.
liver sequestration of ITG-LE from the blood-pool can be
delayed by adding polyethylene glycol-modified phospho-
lipids (ITG-PEG) to the polyiodinated triglyceride emulsion
resulting in a longer blood-pool time. The combined applica-
tion of both formulations of ITG enhances the healthy liver
parenchyma as well as the intrahepatical vessels at the same
time. This reduces the risk to misinterpret vessels as tumor
lesions. False-positive results may occur with hepatocyte-
specific contrast media alone, because liver vessels as well
as liver tumors are non-enhancing and can therefore be
confounded [41].
Dual-phase contrast enhancement can be achieved with
contrast media that first circulate in the blood and are
subsequently cleared via the hepatobiliary pathway hereby
enhancing the liver and the spleen for hours [36].
Also nanoparticles may deal as CT contrast agents. An
example is nanosized bismuth-sulfide with a polymer
coating. These nanoparticles were described to have x-ray
absorption characteristics that are five times better than that
of iodine-containing compounds, which reduces the risk of
viscosity problems [42]. The vascular half-life is longer than
that of iodine-based contrast media and their efficacy/safety
profile is comparable or better. Imaging of tracheobronchial
lymph nodes with nanoparticles in CT has already been
shown in dogs [43].
X-Ray Dose
Dose applied during image acquisition is of special
concern in small animal CT, especially if follow-up
examinations of the same animal are necessary. When all
other hardware and acquisition parameters are kept constant
a good low-contrast spatial resolution requires the
application of high x-ray doses: in order to be able to
differentiate neighboring tissues with similar (electron)
densities with a significant SNR, a particular voxel of the
imaged animal being the smallest unit of resolution needs to
interact with a certain minimum amount of photons. The
minimum amount of required photons is constant regardless
of the voxel size. However, dose is a function of the energy
deposited through photon absorption and energy loss via
Compton scattering interactions per volume, therefore dose
needs to increase with the power of four with the resolution
element if photon noise is supposed to be constant. In other
words: An increase in resolution from 1.0 mm to 0.5 mm
has to be paid for by a 16-fold increase in exposure if we
expect to see low-contrast lesions of 0.5 mm diameter with
the same clarity as lesions of 1.0 mm diameter in the
previous case [34].
Small animal CT needs to be performed with higher
resolution than human CT scanning because anatomical
structures within laboratory animals are smaller than in
human bodies. In order to achieve spatial and low-contrast
resolution (and thus noise levels) comparable to human CT
imaging higher x-ray doses are physically needed. For
example, mice imaging with an isotropic resolution of 135
mrequires a dose of 250 mGy supposing an ideal small
animal CT-scanner is used for the examination [44], which
amounts to a 10-30 fold increase in dose compared to a CT
examination of a human patient [45, 46]. An isotropic
resolution of 135 m results in a three-fold increase in
spatial resolution compared to clinical CT, whereas the
theoretically expected x-ray dose would increase by a factor
of 3
4
=27 (cf. [34]). This would mean that imaging mice with
the same image quality at an isotropic resolution of 35 m
would require a dose of 2500 Gy. From these studies one can
conclude that fundamental laws of physics impose the limits
of dose in small animal imaging and thereby the limits of in-
vivo small animal imaging [44]. Depending on the diagnostic
demand, however, x-ray dose can be decreased while
maintaining high spatial resolution if cut-backs in other
image quality parameters such as image noise and soft-tissue
contrast can be accepted. Much higher image noise is
acceptable e.g. when high-contrast structures such as bones,
lung or contrast media filled vessels are the focus of the
examination. So far only whole body doses have been
discussed, which means that depending on the experimental
focus higher doses could be applied to organs or parts of the
animal which are less dose-sensitive.
Dose limitations for imaging living animals are usually
stated in terms of the lethal dose (LD): LD
50/30
is the whole
body radiation dose which would kill 50% of an exposed
population within 30 days of exposure. Previous studies
indicated that the lethal dose of mice ranges between 5-7.6
Gy [44, 47, 48], depending on the strain [49], age and other
factors. This range certainly sets the upper limit for the
whole body x-ray exposure of a single examination of a
living animal. Fortunately, small animal CT examinations
usually do not even get close to this limit.
Since the main advantage of in-vivo small animal CT lies
in the fact that follow-up studies can be performed the
effects of sublethal dosages become relevant. The sum of the
biological effects of many sublethal doses is a complex
function that is influenced by many factors. While the lethal
dose might give an absolute limit, exposure to much smaller
x-ray dosages can already have biological effects, which
might interfere with experimental results. Several low-dose
radiation effects are described including effects on tumor
growth, hematopoiesis [45, 50] and bone growth [2]. If
follow-up studies are performed over a long timeframe a
possible contribution of radiation effects to the observed
results needs to be assessed: a group of animals that is only
imaged at the end of the time series could act as a control in
comparison to the scanned animals [51] or the contralateral
limb not having been exposed to radiation could be taken as
a non-irradiated control [2].
Single organ doses for mice imaging were calculated
using a Monte Carlo simulation. Here the typical whole
body dose was 80 mGy (at 80 kV tube voltage) to 160 mGy
(at 50 kV tube voltage). Furthermore, a strong dependance
on the tube voltage as well as on the position within the
animal was found [50]. To allow the computation of x-ray
exposure doses from air kerma measurements dose coeffi-
cients for standardized mice, typical scanner geometries and
x-ray spectra were calculated using Monte Carlo simulations
[45].
NEW DEVELOPMENTS
Flat-Panel Based Small Animal CT
Flat-panel detector based micro- and mini-CT scanners
for small animal imaging have been described in bench-top
Small Animal Computed Tomography Imaging Current Medical Imaging Reviews, 2007, Vol. 3, No. 1 53
systems [18] as well as in gantry-based systems [10, 12, 20,
52]. As discussed in Detector Technology, flat-panel detec-
tors have several benefits over detectors that were used for
micro-CT earlier.
Large flat-panel detectors are ideally suited for gantry-
based small animal CT imaging systems with a geometry
that allows a reasonably high resolution with a large gantry
bore and a wide z-coverage so that bigger animals such as
rabbits and piglets [10, 12] can be scanned. The data of even
bigger animals can be acquired in one rotation.
A decrease in spatial resolution allows a relaxation in
dose performance and therefore a very fast acquisition of
single projections and therefore the whole volume in short
scan times. Fast, continuous scanning becomes possible.
Acquiring the same volume in short succession is a
prerequisite for perfusion imaging of fast contrast media
dynamics. Slow contrast media dynamics such as kidney or
liver uptake and bladder or gall bladder filling could already
be assessed with slower micro-CT scanners. However, fast
contrast media dynamics such as brain, tumor and scar
perfusion is of much higher interest. Using flat-panel
detector CT scanners, exemplary perfusion imaging of fast
dynamics for (whole) small animals has been shown [52].
This technique could provide valuable physiological and
functional information of many small animal models and
diseases. Several post-processing methods could be applied
to the contrast media time course data. First experiments
were promising: the time course of contrast media could be
traced, and parameters such as Mean Transit Time, Blood
Volume and Blood Flow of experimental tumors could be
calculated [52].
Fast and continuous scanning could also make retros-
pective cardiac and respiratory gating possible. For this goal,
the short exposure time of flat-panel CT that virtually freezes
the heart motion on the level of projections is especially
advantageous. Earlier micro-CT scanners were limited with
regard to motion gating by the long exposure times per
projection.
Compared to prospective triggering, retrospective gating
could decrease the scan times significantly. This could make
the use of standard, iodinated contrast media possible and
may open perspectives to lung or cardiac perfusion studies.
Future Perspectives
For a good overview of future developments of micro-CT
with focus on in-vivo small animal imaging please read the
review from Ritman [53]. If big technical and engineering
challenges could be solved, techniques such as x-ray fluore-
scence, x-ray diffraction or x-ray scatter imaging would be
big steps for CT imaging. Their potential for the improve-
ment of soft-tissue contrast and contrast media sensitivity are
enormous. However, a long time might still pass until these
technologies will reach operational readiness.
Another method possibly available in the near future is
K-edge subtraction imaging. It is based on synchrotron
generated narrow bandwidth rays that allow imaging of
certain chemical elements by imaging the volume twice at
different photon energies (once with a frequency just above
and once with a frequency just below the characteristic K-
edge of the selected element). When both data sets are
subtracted, density differences only remain for the selected
elements. However, the availability of synchrotron radiation
is currently limited to expensive synchrotron generators.
Furthermore, multi-source CT imaging could decrease the
scan time by radiating the object from several directions at
once.
APPLICATIONS
The following chapter gives an overview of the
applications of micro- and mini-CT in small animals. In the
compiled studies herein various scanner designs with a range
of scan parameters, scan times and spatial resolution are
used. Most studies are proof of principle studies. Never-
theless, the conclusions drawn from one study hold true for
other studies in case that similar mini-/micro-CT scanners
with comparable characteristics were used.
Lung Imaging
In mice, the total lung volume is 1.3 cm
3
. The thorax of
mice is about 2 cm in diameter. Proportionally, a mini-CT
scanner that provides an insight on the mice lung structures
as in a human lung should offer a resolution in the order of
75 m [54]. In vivo lung imaging is challenged by the
respiratory and cardiac motions and its compensated imaging
was described by Badea and co-workers [33].
Beside the lung parenchyma, the larger thoracic struc-
tures such as heart, oesophagus, trachea, bronchi and large
vessels can be imaged in non-enhanced scans (Fig. (2)).
Lung Tumors
The detection of experimentally induced lung tumors is
possible in mice using micro-CT. This was shown in a study,
in which urethane induced lung tumours were imaged in
living mice. The scans were performed without motion
compensation and with an isotropic resolution of 35 m [55].
Only few, small tumors that were found in histologic slices
were missed in the micro-CT scans.
Fig. (2). In-vivo thoracic imaging of a mouse (A) and a rat (B) from
flat-panel based CT. Data acquisition time was 5 s. However,
motion artefacts close to the ribs and diaphragm are pronounced.
Bronchi (white arrow) and vessels (black arrow) can be
distinguished from the lung parenchyma.
54 Current Medical Imaging Reviews, 2007, Vol. 3, No. 1 Bartling et al.
Follow-up studies of lung tumor size changes over time
could be performed [56-58]. Size distortions of the lung
tumors due to motion artefacts should be taken into account
when micro-CT without appropriate motion-compensation
algorithms is used for absolute measurement studies.
The minimal sizes of lung tumors that were found in
several studies were: 6.6 mm
3
in volume and 0.85 mm in
diameter in the free lung parenchyma and 1.4 mm in
diameter perihilar [58]. In other studies the size of lung
nodules regardless of their localization was reported to be
0.63 mm
3
minimum in volume [57] and 0.5 mm [56],
respectively, smaller than 0.2 mm in diameter [55]. In this
context, using a ventilator and prospective respiratory gating
the accuracy of tumor volume determination could be
improved [57].
In the reviewed studies almost all tumors that were
identified by micro-CT were confirmed histologically, but
vice versa not all tumors that were found in histology were
reliably identified in micro-CT even if they had the same
size as tumors that were found. J uxtahilar tumors were most
easily to detect, while the tumors next to the thoracic wall
and the big vessels were most difficult to visualize, which is
most likely due to motion artefacts that lead to a smearing of
the thoracic walls. Furthermore, it is difficult to differentiate
tumors from hilar structures and from vessels. Similarly,
because the lung tissue in the reviewed studies was normal
apart from the induced tumors, it remains to be proven by
other studies, how accurate micro-CT can differentiate
between scar tissue, microgranulomas, hyperplastic lympho-
cellular or other non-malignant lesions in lung tissue. The
use of macroscopic differential features to distinguish bet-
ween tissues as known from human thoracic imaging is
certainly limited by the resolution. Blood-pool contrast
media proved to be valuable to improve differentiation of
tumors and vessels [58].
Even pleural effusions occurring as the consequence of
tumor disease could be successfully imaged with micro-CT
[58].
General Lung Parenchyma Changes (Emphysema and
Fibrosis)
Using micro-CT the assessment of generalized lung
parenchyma changes is possible as shown in following
experiment: For example lung emphysema was induced in
C57BL/6J mice by intratrachael instillation of pancreatic
elastase. The mice were scanned using a micro-CT without
motion compensation and an isotropic resolution of 35 m.
On CT images mice treated with the highest dosage of
elastase showed the highest amount of pixels with low HU
values within their lungs. Furthermore, the lung volume was
larger in this group. In conclusion, lung emphysema can be
detected by micro-CT indirectly by the lower density of the
lung and by an increase in lung volume [59]. Unfortunately,
despite the sufficiently up-scaled spatial resolution a direct
assessment of the altered emphysematous lung structure as in
human high resolution (HR)CT-imaging is not possible [54].
This is mainly due to motion artefacts and resolution
limitations.
Vice versa, also the assessment of lung fibrosis in mice is
possible by analysis of CT numbers. Here the disease
progress was shown to be associated with an increase in CT
numbers on a clinical CT-scanner [60]. Using mini-CT it
was shown that not only the general density could be
assessed but also structural changes of bleomycin-induced
fibrosis such as ground glass opacity, septal thickening,
fibrotic strands and secondary dilation of the bronchial
system (bronchiectasis) [52].
Bone Imaging
Similar to its applications in very high resolution ex-vivo
bone imaging micro-CT can provide information about bone
Fig. (3). 3D reconstruction (A) and 2D slice (B) of a rat knee and its adjacent bones scanned with a bench-top micro-CT shown in Fig. (1) B.
The rat was rotated between the detector and the x-ray source while only the limb was radiated. High spatial resolution of 6 m
3
could be
achieved because a single limb fits into a very small scan field-of-view (image courtesy Prof. W. Kalender, Institute of Medical Physics,
Erlangen, Germany).
Small Animal Computed Tomography Imaging Current Medical Imaging Reviews, 2007, Vol. 3, No. 1 55
structure and its changes in vivo in small animals, however,
with stricter resolution constraints.
Microstructural Bone Imaging
Using micro-CT a morphological bone analysis of its
micro-architecture can be performed in vivo providing
several descriptors such as bone volume ratio, bone surface
ratio, trabecular thickness, trabecula separation, trabecular
number, connectivity density and structural model index [61-
63].
CT density values as a questionable surrogate marker for
bone status could also be acquired using micro-CT. Dual-
energy scans as used in clinical CT setups to assess the bone
density have not been reported using micro-CT yet.
Small animal bone scans can be performed on single
limbs. Customized jigs can assure immobilization. Due to
the low diameter of those limbs resolution optimized scanner
geometries (e.g. 15 m isotropic [2]) with only a small scan
FOV can be applied (Fig. (3)).
Macroscopic Skeletal Imaging
Micro-CT can be used to image the whole macroscopic
animal skeleton. Applications are studies on the skeletal
development [64] or studies in which structural changes in
the skeleton anatomy occur as a consequence of disease. For
these applications, the demand for large scan volumes as
provided by clinical-scale CT or flat-panel based mini-CT
scanners is higher than the need for high resolution. Clinical
scale scanners as well as flat-panel based mini-CT systems
can provide an FOV that is big enough to image a whole
mouse or rat during one acquisition. Using appropriate post
processing the whole skeleton can be displayed in high
spatial resolution three dimensionally [20, 52].
Bone Metastasis Screening
Micro-CT can be used to screen with a high sensitivity
and specificity for osteolytic bone metastases in whole
mouse skeleton [58]. Visualization of the soft tissue of the
metastases that extend beyond the bone surfaces, however, is
a challenge for the soft-tissue contrast resolution capability
of the micro-CT. Its detection has not yet been described
using micro-CT.
Angiographic Imaging
In-vitro imaging using micro-CT resulted in astonishing
results of the vasculature and microvasculature of small
animals. Often, these images were generated after casting the
vasculature with contrast media such as silicon-based
compound (Microfil MV-122, Flow Tech, Carver, MA,
USA) [65].
For in-vivo imaging various contrast media have been
discussed in Section. Due to the long scan times of micro-CT
mainly blood-pool contrast agents have been used to image
the vasculature (Fig. (4)) [37-39]. The detection of the aorta,
pulmonary trunk, inferior vena cava, renal artery and vein in
mice and rats [35, 66] has been described.
Standard iodinated contrast media typically used for
clinical examinations clear too fast from the vessels to be
used in micro-CT scanners. However, mini-CT-scanning in
such short times became recently feasible. In-vivo imaging of
all central vessels of mice and even that of dilated
subcutaneous vessels that drain implanted tumors as well as
smaller vessels inside the tumor were demonstrated using
flat-panel mini-CT in combination with standard iodinated
contrast media (Fig. (5)). Characteristics of tumor vessel
architecture could be assessed [52] and their changes traced
Fig. (4). Blood-pool contrast media enhanced scan of a rat in an early (A) and later (B) phase as scanned by a micro-CT shown in Fig. (1) D.
The scanners spatial resolution is optimized for the dimensions of rats and mice. Due to the scan time of 180 s blood-pool contrast media
were used (image courtesy Prof. W. Kalender, Institute of Medical Physics, Erlangen, Germany).
56 Current Medical Imaging Reviews, 2007, Vol. 3, No. 1 Bartling et al.
in follow-up studies [20].
Assessment of Slow Contrast Agent Kinetics
To assess the slow time-resolved contrast media
enhancement in micro-CT long circulating contrast media
were used [39, 66]. Other authors assumed that even with
iodinated contrast agents quantification of several aspects of
renal function as shown in clinical CT imaging may be
successful [1, 35]. The renal cortex, medulla, pelvicalyceal
system and ureter could be differentiated in contrast
enhanced studies. A hydronephrosis model has been charac-
terized using this method successfully [35].
Through the use of faster flat-panel CT scanners, also
fast dynamic contrast-enhanced applications may become
broadly available [52] (Fig. (6)).
Cardiac and Thoracic Motion-Compensated Imaging
Thoracic imaging of living animals is challenged by heart
and lung movements. These movements lead to imaging
artefacts, which reduce the accuracy of thoracic imaging of
small animals and make reliable measurements e.g. of lung
tumors more problematic [57, 67] (Fig. (2)).
To compensate for such movements prospective as well
as retrospective gating and triggering methods can be
applied. In prospective methods, the imaging process is
modulated according to the movements of the small animal.
Therefore only one animal can be imaged during one
acquisition. Prospective methods take longer scan times than
retrospective methods but are usually more dose efficient
than retrospective methods.
Regardless of whether prospective or retrospective
methods are used, both methods add up projections acquired
during defined phases of cardiac or respiratory cycles to
form one complete dataset for CT image reconstruction.
Therefore, position reproducibility between several breathing
as well as cardiac cycles is crucial. However, this reprodu-
cibility is not better than 100 m in small animals [68],
limiting the maximum spatial resolution of thoracic imaging
of free-breathing small animals to a resolution in this order.
So far predominantly prospective methods have been
implemented in small animal motion compensated CT
imaging: To physically stop breathing excursions small
animals can be intubated and ventilated. Breathing can be
arrested for a short time period. Due to the long scan times
of micro-CT complete datasets can not be acquired during
one breath hold.
However, the short breathing movement arrests that
occur during the physiological ventilation plateau phase can
Fig. (6). Perfusion CT using standard iodinated contrast media of a
bone metastasis model in rats [76] as scanned with flat-panel CT.
The contralateral (black arrow) as well as the partially destroyed
ipsilateral tibia (white arrow) can be seen. Rotation time was 6 s,
half-scan reconstruction was performed every 3 s using 180 data.
Injection of contrast media was initiated together with scanning. At
0 s (A) no contrast media was detected in the selected ROI, which
covered the bone metastasis, while at 24 s (B) pronounced rim and
vessel enhancement can be found. The time-course of averaged CT
numbers within the ROI is given in (C) (unpublished data by Dr.
Tobias Baeuerle, German Cancer Research Center (DKFZ),
Heidelberg, Germany).
Fig. (5). Mini-CT angiography using standard, iodinated contrast
media of a mouse scanned using a flat-panel detector CT as in Fig.
(1) F [12]. 100 l contrast media were injected three seconds before
scan; data acquisition took three seconds, resulting in a late
arterial/venous phase. A volume rendering, in which the slice
positions of the axial slices (A-D) are indicated, is given in (A).
Furthermore, a coronal slice is given in (E), showing the kidneys
(long black arrow) and suprarenal glands (long white arrow). The
contrast enhancement in the caval vein (short black arrow) is more
pronounced than in the aorta (short white arrow). Small structures
such as the splenal vein (white dashed arrow), the hepatic venous
confluence (white arrow head) and the suprarenal vessels (black
dashed arrow) can be detected. Regardless of cardiac motion the
right and left ventricle (black asterisk) can be distinguished clearly.
Small Animal Computed Tomography Imaging Current Medical Imaging Reviews, 2007, Vol. 3, No. 1 57
be used to acquire at least a part (e.g. one projection) of the
imaging dataset [4, 33, 57]. The trigger point can be
optimized so that data acquisition is performed in phases of
the respiratory cycle that show least motion [69].
In another study, the animal could breathe freely. Its
breathing excursions were tracked by a pneumatic cushion
that was rigidly attached to the animal. In a definable
breathing excursion position projections were acquired.
After each acquisition the animal was rotated to the next
projection view [70]. Acquiring various datasets at various
breathing excursion points allowed four-dimensional
imaging of the thorax and lung.
With a variation of the study design multiple projections
at one angle were acquired and retrospectively sorted accor-
ding to various breathing cycles. Intrinsic image information
was used for the projection sorting, thereby not requiring an
additional device to collect gating information [71].
Cardiac motion compensation can be implemented using
a prospective ECG triggering [4] combined with breathing
triggering in intubated and ventilated mice. Repeating at
various trigger points resulted in 4D imaging of a rodent
heart [67].
The benefits of lung motion compensation are obvious.
Lung parenchyma, diaphragm and vessels were displayed
sharper than in non-gated scans [4, 70]. Respiratory gating
made the measurement of lung tumor sizes more precise
because smearing caused by lung motion artefacts could be
reduced or avoided. Smearing results in inaccurate size
measurements [57]. ECG-gating further improved the
display of lung and cardiac anatomy [4].
Soft Tissue Imaging
While micro- and mini-CT provide high resolution of
high contrast structures such as bone and contrast media
filled vessels, the soft-tissue contrast is relatively low. As
mentioned, the soft-tissue contrast is strongly disturbed by
image noise and is therefore strongly depending on the
applied radiation dose.
To this point it is uncertain, how suited micro- and mini-
CT are to differentiate soft-tissue structures because the
differentiation of soft tissue in mini- and micro-CT is usually
not as good as in clinical scale CT scanners.
However, differentiation of fat from other soft tissue
structures is possible. Measurement of fat volume as well as
fat distribution in a mouse disease model has been demons-
trated [35]. Fat measurements open promising perspectives
in imaging diabetic animal models or in studying kachexia
(e.g. during neoplastic disease). Most likely, if fat can be
differentiated from surrounding soft tissue, determination of
muscle volume also becomes feasible and can be used to
characterize musculoskeletal disorders, e.g. in transgenic
mice with muscle dystrophia.
Organs in the peritoneal cavity can already be differen-
tiated in non-enhanced scans. Liver, kidney, spleen, heart,
lung, stomach, adrenal gland, gut and bladder can be
identified and their borders be delineated.
To enhance the contrast between organs or to different-
tiate tumors and organs in the peritoneal cavity, standard,
iodinated contrast media have been injected in to the perito-
neal cavity, facilitating visualization of their boundaries.
In contrast-enhanced flat-panel micro-CT scans also the
longitudinal assessment of urethral tumors in the bladder
became feasible [72], where tumors were identified as
negative shapes in the contrast media filled bladder.
Image Registration with Other Modalities
The diagnostic value of studies can be improved by
fusing and matching the data from different imaging
modalities thus directly combining morphological and
functional features: registration is the method to align the
modalities in space, while fusion is a function that combines
all or parts of the information of the modalities into a new
dataset.
Potential complementary imaging data for registration
with micro-CT can derive from radionuclide, optical and MR
imaging as well as histology.
In general, registration methods can be characterized as
follows: intrinsic (registration based on the image infor-
mation itself) or extrinsic (registration based on additional
markers or frames), retrospective (after scanning) or pros-
pective (before scanning, e.g. through reference frames or
scan beds), manual (user based) or automatic (software
algorithm), rigid (no spatial distortion is allowed to register
both datasets) or non-rigid (deformation is allowed).
Small animal multimodality registration can be done by
combining both imaging modalities in one scanner and
perform simultaneous or near simultaneous data acquisitions
[16, 73]. The advantage of combining two modalities in one
scanner system is that there is no need to move the animal
between acquisitions, meaning both scanners can use the
same coordinate system, resulting in very good registration
accuracies. The main disadvantage of simultaneous imaging
devices is a certain loss of the flexible choice of scanner
geometry.
Another method is the use of a scan bed or a framework,
to which the animal can be fixated. The scan bed can then be
used to transfer the animal from one modality to the next
insuring accurate animal positioning. This method has been
suggested as the most feasible method in small animal
imaging that provides a very good accuracy [74].
If multi-modality imaging acquisition is not conducted in
a combined scanner or using fixed reference frames, retros-
pective methods such as software-based or manual methods
need to be applied. Multiple software-based approaches for
retrospective, automatic methods are also described [75].
Registration accuracy was in the order of 1 mm. The used
software approaches still needed significant user interaction.
Alternatively, marker based approaches could be chosen,
here registration markers that are visible in both modalities
were attached rigidly to the animal. So far radionuclide
image registration with micro-CT has been published mostly.
Registration between
18
F-fluoride PET bone images and
micro-CT of the rat skull [75],
18
F-FDG PET of the rat and
micro-CT (Fig. (7)) [74, 75] as well as SPECT and micro-CT
[16, 74] was described.
58 Current Medical Imaging Reviews, 2007, Vol. 3, No. 1 Bartling et al.
ACKNOWLEDGEMENTS
We thank Willi Kalender and Michael Grasruck for their
support by providing image material.
REFERENCES
[1] Ritman EL. Micro-computed tomography-current status and
developments. Ann Rev Biomed Eng 2004; 6: 185-208.
[2] Boyd SK, Davison P, Muller R, Gasser J A. Monitoring individual
morphological changes over time in ovariectomized rats by in vivo
micro-computed tomography. Bone 2006; 39: 854-62.
[3] Calder WA. Size, function and life history. 2 ed: Dover
Publications 1996.
[4] Badea C, Hedlund LW, J ohnson GA. Micro-CT with respiratory
and cardiac gating. Med Phys 2004; 31: 3324-9.
[5] Holdsworth DW, Thornton MM. Micro-CT in small animal and
specimen imaging. Trends Biotechnol 2002; 20: 34-9.
[6] Badea C, Hedlund LW, Wheeler CT, Mai W, J ohnson GA.
Volumetric micro-CT system for in vivo microscopy. IEEE Int
Symp Biomed Imag: Macro to Nano 2004: 1377-80.
[7] Wang G, Vannier M. Micro-CT scanners for biomedical
applications: an overview. Adv Imag 2001; 16: 18-27.
[8] Knollmann F, Valencia R, Buhk J H, Obenauer S. Characteristics
and applications of a flat panel computer tomography system. RoFo
2006; 178: 862-71.
[9] Paulus MJ , Gleason SS, Kennel SJ , Hunsicker PR, J ohnson DK.
High resolution X-ray computed tomography: an emerging tool for
small animal cancer research. Neoplasia 2000; 2: 62-70.
[10] Ross W, Cody DD, Hazle J D. Design and performance
characteristics of a digital flat-panel computed tomography system.
Med Phys 2006; 33: 1888-901.
[11] Siewerdsen J H, Moseley DJ , Bakhtiar B, Richard S, J affray DA.
The influence of antiscatter grids on soft-tissue detectability in
cone-beamcomputed tomography with flat-panel detectors. Med
Phys 2004; 31: 3506-20.
[12] Gupta R, Grasruck M, Suess C, et al. Ultra-high resolution flat-
panel volume CT: fundamental principles, design architecture, and
systemcharacterization. Eur Radiol 2006; 16: 1191-205.
[13] Yaffe MJ , Rowlands J A. X-ray detectors for digital radiography.
Phys Med Biol 1997; 42: 1-39.
[14] Feldkamp LA, Davis LC, Kress J W. Practical cone-beam
algorithm. J Opt Soc Am1984; 1: 612-9.
[15] Kohlbrenner A, Koller B, Hammerle S, Ruegsegger P. In vivo
micro tomography. Adv Exp Med Biol 2001; 496: 213-24.
[16] Kastis GA, Furenlid LR, Wilson DW, Peterson TE, Barber HB,
Barrett HH. Compact CT/SPECT small-animal imaging system.
IEEE Trans Nucl Sci 2004; 51: 63-7.
[17] Holdsworth DW, Drangova M, Fenster A. A high-resolution XRII-
based quantitative volume CT scanner. Med Phys 1993; 20: 449-
62.
[18] Lee SC, KimHK, Chun IK, Cho MH, Lee SY, Cho MH. A flat-
panel detector based micro-CT system: performance evaluation for
small-animal imaging. Phys Med Biol 2003; 48: 4173-85.
[19] KimHK, Lee SC, Chun IK, et al. Performance evaluation of a flat-
panel detector-based microtomography system for small-animal
imaging. IEEE Nucl Sci Symp 2003; 3: 2108-113.
[20] Kiessling F, Greschus S, Lichy MP, et al. Volumetric computed
tomography (VCT): a new technology for noninvasive, high-
resolution monitoring of tumor angiogenesis. Nat Med 2004; 10:
1133-8.
[21] Asahina, H. Selenium-based flat panel x-ray detector for digital
fluoroscopy and radiography. Visions 2001; 1: 31-40.
[22] Kalender WA. The use of flat-panel detectors for CT imaging.
Radiologe 2003; 43: 379-87.
[23] Popescu S, Stierstorfer K, Flohr T, Suess C, Grasruck M. Design
and evaluation of a prototype volume CT scanner. Proc SPIE 2005;
5745: 600-608.
[24] J oon KimH, Kyung KimH, Cho G, Choi J . Construction and
characterization of an amorphous silicon flat-panel detector based
on ion-shower doping process. NIMA 2003; 505: 155-8.
[25] Voelk M, Hamer OW, Feuerbach S, Strotzer M. Dose reduction in
skeletal and chest radiography using a large-area flat-panel detector
based on amorphous silicon and thallium-doped cesium iodide:
technical background, basic image quality parameters, and review
of the literature. Eur Radiol 2004; 14: 827-34.
[26] Colbeth RE, Boyce S, Fong R, et al. 40 x 30 cmflat-panel imager
for angiography, R&F, and cone-beamCT applications. Proc SPIE
2001; 4320: 94-102.
[27] Siewerdsen J H, J affray DA. A ghost story: spatio-temporal
response characteristics of an indirect-detection flat-panel imager.
Med Phys 1999; 26: 1624-41.
[28] Siewerdsen J H, J affray DA. Cone-beam computed tomography
with a flat-panel imager: effects of image lag. Med Phys 1999; 26:
2635-47.
[29] Roos PG, Colbeth RE, Mollov I, et al. Multiple-gain-ranging
readout method to extend the dynamic range of amorphous silicon
flat-panel imagers. Proc SPIE 2004; 5368: 139.
[30] Khodaverdi M, Pauly F, Weber S, et al. Preliminary studies of a
micro-CT for a combined small animal PET/CT scanner. IEEE
Nucl Sci Symp 2001; 3.
[31] Grasruck M, Gupta R, Reichardt B, et al. Combination of CT
scanning and fluoroscopy imaging on a flat-panel CT scanner. Proc
SPIE 2006; 875-882
[32] Motz J W, Danos M. Image information content and patient
exposure. Med Phys 1978; 5: 8-22.
[33] Badea CT, Fubara B, Hedlund LW, J ohnson GA. 4-D micro-CT of
the mouse heart. Mol Imaging 2005; 4: 110-6.
Fig. (7). Micro-PET and micro-CT co-registration. Structural information of a mouse from micro-CT (A) is registered to
18
F-FDG micro-PET
data (B), resulting in a combination of both modalities (C). The combination image provides anatomic information on the tumor location
(white arrows) and its metabolism (from Meei-Ling J an et al., Institute of Nuclear Energy Research, Longtan, Taiwan, 2005 IEEE).
Small Animal Computed Tomography Imaging Current Medical Imaging Reviews, 2007, Vol. 3, No. 1 59
[34] Kalender WA. Computed tomography: fundamentals, system
technology, image quality, applications. Munich, VCH
Verlagsgesellschaft 2005.
[35] Paulus MJ , Gleason SS, Easterly ME, Foltz CJ . A review of high-
resolution X-ray computed tomography and other imaging
modalities for small animal research. Lab Anim2001; 30: 36-45.
[36] Ford NL, GrahamKC, GroomAC, Macdonald IC, Chambers AF,
Holdsworth DW. Time-course characterization of the computed
tomography contrast enhancement of an iodinated blood-pool
contrast agent in mice using a volumetric flat-panel equipped
computed tomography scanner. Invest Radiol 2006; 41: 384-90.
[37] Vera DR, Mattrey RF. A molecular CT blood pool contrast agent.
Acad Radiol 2002; 9: 784-92.
[38] Kao CY, Hoffman EA, Beck KC, Bellamkonda RV, Annapragada
AV. Long-residence-time nano-scale liposomal iohexol for X-ray-
based blood pool imaging. Acad Radiol 2003; 10: 475-83.
[39] Torchilin VP, Frank-Kamenetsky MD, Wolf GL. CT visualization
of blood pool in rats by using long-circulating, iodine-containing
micelles. Acad Radiol 1999; 6: 61-5.
[40] Bakan DA, Doerr-Stevens J K, Weichert J P, Longino MA, Lee FT,
Counsell RE. Imaging efficacy of a hepatocyte-selective
polyiodinated triglyceride for contrast-enhanced computed
tomography. AmJ Ther 2001; 8: 359-65.
[41] Weichert J P, Lee FT, Chosy SG, et al. Combined hepatocyte-
selective and blood-pool contrast agents for the CT detection of
experimental liver tumors in rabbits. Radiology 2000; 216: 865-71.
[42] Rabin O, Manuel Perez J , GrimmJ , Wojtkiewicz G, Weissleder R.
An X-ray computed tomography imaging agent based on long-
circulating bismuth sulphide nanoparticles. Nat Mater 2006; 5:
118-22.
[43] Ketai LH, Muggenberg BA, McIntire GL, et al. CT imaging of
intrathoracic lymph nodes in dogs with bronchoscopically
administered iodinated nanoparticles. Acad Radiol 1999; 6: 49-54.
[44] Ford NL, Thornton MM, Holdsworth DW. Fundamental image
quality limits for microcomputed tomography in small animals.
Med Phys 2003; 30: 2869-77.
[45] Boone J M, Velazquez O, Cherry SR. Small-animal x-ray dose from
micro-CT. Mol Imaging 2004; 3: 149-58.
[46] Goertzen AL, Meadors AK, Silverman RW, Cherry SR.
Simultaneous molecular and anatomical imaging of the mouse in
vivo. Phys Med Biol 2002; 47: 4315-28.
[47] Sato F, Sasaki N, Kawashima N, Chino F. Late effects of whole or
partial body x-irradiation on mice: life shortening. Int J Radiat Biol
Relat Stud Phys ChemMed 1981; 39: 607-15.
[48] Mole RH. Quantitative observations on recovery fromwhole body
irradiation in mice II. Recovery during and after daily irradiation.
Br J Radiol 1957; 30: 40-6.
[49] Kohn HI, Kallmann RF. The influence of strain on acute x-ray
lethality in the mouse I. LD50 and death rate studies. Radiat Res
1956; 5: 309-17.
[50] Taschereau R, Chow PL, Chatziioannou AF. Monte Carlo
simulations of dose frommicroCT imaging procedures in a realistic
mouse phantom. Med Phys 2006; 33: 216-224.
[51] Waarsing J H, Day J S, Verhaar J A, Ederveen AG, Weinans H.
Bone loss dynamics result in trabecular alignment in aging and
ovariectomized rats. J Orthop Res 2006; 24: 926-35.
[52] Greschus S, Kiessling F, Lichy MP, et al. Potential applications of
flat-panel volumetric CT in morphologic and functional small
animal imaging. Neoplasia 2005; 7: 730-40.
[53] Ritman EL. Molecular imaging in small animals--roles for micro-
CT. J Cell BiochemSuppl 2002; 39: 116-24.
[54] Ritman EL. Micro-computed tomography of the lungs and
pulmonary-vascular system. Proc AmThorac Soc 2005; 2: 477-80.
[55] De Clerck NM, Meurrens K, Weiler H, et al. High-resolution x-ray
microtomography for the detection of lung tumors in living mice.
Neoplasia 2004; 6: 374-9.
[56] Kennel SJ , Davis IA, Branning J , Pan H, Kabalka GW, Paulus MJ .
High resolution computed tomography and MRI for monitoring
lung tumor growth in mice undergoing radioimmunotherapy:
correlation with histology. Med Phys 2000; 27: 1101-7.
[57] Cody DD, Nelson CL, Bradley WM, et al. Murine lung tumor
measurement using respiratory-gated micro-computed tomography.
Invest Radiol 2005; 40: 263-9.
[58] Li XF, Zanzonico P, Ling CC, O'Donoghue J . Visualization of
experimental lung and bone metastases in live nude mice by x-ray
micro-computed tomography. Technol Cancer Res Treat 2006; 5:
147-55.
[59] Postnov AA, Meurrens K, Weiler H, et al. In vivo assessment of
emphysema in mice by high resolution x-ray microtomography. J
Microsc 2005; 220: 70-5.
[60] Plathow C, Li M, Gong P, et al. Computed tomography monitoring
of radiation-induced lung fibrosis in mice. Invest Radiol 2004; 39:
600-9.
[61] Tamada T, Sone T, J o Y, Imai S, Kajihara Y, Fukunaga M. Three-
dimensional trabecular bone architecture of the lumbar spine in
bone metastasis from prostate cancer: comparison with
degenerative sclerosis. Skeletal Radiol 2005; 34: 149-55.
[62] David V, Laroche N, Boudignon B, et al. Noninvasive in vivo
monitoring of bone architecture alterations in hindlimb-unloaded
female rats using novel three-dimensional microcomputed
tomography. J Bone Miner Res 2003; 18: 1622-31.
[63] Waarsing J H, Day J S, Weinans H. Longitudinal micro-CT scans to
evaluate bone architecture. J Musculoskelet Neuronal Interact
2005; 5: 310-2.
[64] Guldberg RE, Lin AS, Coleman R, Robertson G, Duvall C.
Microcomputed tomography imaging of skeletal development and
growth. Birth Defects Res C Embryo Today 2004; 72: 250-9.
[65] J orgensen SM, Demirkaya O, Ritman EL. Three-dimensional
imaging of vasculature and parenchyma in intact rodent organs
with x-ray micro-CT. AmJ Physiol 1998; 275: 1103-14.
[66] Mukundan S, Ghaghada KB, Badea CT, et al. A liposomal
nanoscale contrast agent for preclinical CT in mice. Am J
Roentgenol 2006; 186: 300-7.
[67] Badea CT, Bucholz E, Hedlund LW, Rockman HA, J ohnson GA.
Imaging methods for morphological and functional phenotyping of
the rodent heart. Toxicol Pathol 2006; 34: 111-7.
[68] Mai W, Badea CT, Wheeler CT, Hedlund LW, J ohnson GA.
Effects of breathing and cardiac motion on spatial resolution in the
microscopic imaging of rodents. Magn Reson Med 2005; 53: 858-
65.
[69] Walters EB, Panda K, Bankson J A, Brown E, Cody DD. Improved
method of in vivo respiratory-gated micro-CT imaging. Phys Med
Biol 2004; 49: 4163-72.
[70] Ford NL, Nikolov HN, Norley CJ , et al. Prospective respiratory-
gated micro-CT of free breathing rodents. Med Phys 2005; 32:
2888-98.
[71] Hu J , Haworth ST, Molthen RC, Dawson CA. Dynamic small
animal lung imaging via a postacquisition respiratory gating
technique using micro-cone beam computed tomography. Acad
Radiol 2004; 11: 961-70.
[72] J ohnson AM, Conover DL, Huang J , et al. Early detection and
measurement of urothelial tumors in mice. Urology 2006; 67:
1309-14.
[73] Meei-Ling J , Yu-Ching N, Kou-Wei C, et al . A combined micro-
PET/CT scanner for small animal imaging. NIM A 2006; in press.
[74] J an ML, Chuang KS, Chen GW, et al. A three-dimensional
registration method for automated fusion of micro PET-CT-SPECT
whole-body images. IEEE Trans Med Imaging 2005; 24: 886-93.
[75] Vaquero J , Desco M, Pascau J , et al. PET, CT and MR image
registration of the rat brain and skull. IEEE Trans Nucl Sci 2001;
48: 1440-5.
[76] Bauerle T, Adwan H, Kiessling F, Hilbig H, Armbruster FP, Berger
MR. Characterization of a rat model with site-specific bone
metastasis induced by MDA-MB-231 breast cancer cells and its
application to the effects of an antibody against bone sialoprotein.
Int J Cancer 2005; 115: 177-86.
Received: October 27, 2006 Revised: December 13, 2006 Accepted: December 14, 2006