PROCEEDINGS OF SPIE
SPIEDigitalLibrary.org/conference-proceedings-of-spie
In vivo tomographic imaging based
on bioluminescence
Wenxiang Cong, Durairaj Kumar, Yubin Kang, Patrick
Sinn, Earl Nixon, et al.
Wenxiang Cong, Durairaj Kumar, Yubin Kang, Patrick Sinn, Earl Nixon,
John Mienel, Melissa J. Suter, Lihong V. Wang, Geoffrey McLennan, Eric A.
Hoffman, Ge Wang, "In vivo tomographic imaging based on
bioluminescence," Proc. SPIE 5535, Developments in X-Ray Tomography IV,
(26 October 2004); doi: 10.1117/12.560522
Event: Optical Science and Technology, the SPIE 49th Annual Meeting, 2004,
Denver, Colorado, United States
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 9/25/2018 Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
In vivo
tomography imaging based on bioluminescence
Wenxiang Cong
a
, Kumar Durairaj
a
, Yubin Kang
c
, Patrick Sinn
c
, Earl Nixon
b
, John Mienel Jr
b
,
Melissa Suter
c
, Lihong V. Wang
d
, Geoffrey McLennan
c
, Eric. A. Hoffman
b
, and
∗
Ge Wang
a, b
a
Bioluminescence Tomography Laboratory, Department of Radiology
b
CT/Micro-CT Laboratory, Department of Radiology
c
Department of Internal Medicine,
University of Iowa,
200 Hawkins Drive, Iowa City, IA 52242, USA
d
Optical imaging laboratory, Department of biomedical engineering
Texas A&M University
ABSTRACT
The most important task for bioluminescence imaging is to identify the emission source from the captured
bioluminescent signal on the surface of a small tested animal. Quantitative information on the source location,
geometry and intensity serves for
in-vivo
monitoring of infectious diseases, tumor growth, metastases in the small
animal. In this paper, we present a point-spread function-based method for reconstructing the internal bioluminescent
source from the surface light output flux signal. The method is evaluated for sensing the internal emission sources in
nylon phantoms and within a live mouse. The surface bioluminescent signal is taken with a highly sensitive CCD
camera. The results show the feasibility and efficiency of the proposed point-spread function-based method.
Keywords:
Bioluminescence tomography, CT/micro-CT, molecular imaging, point-spread function (PSF).
1. INTRODUCTION
Biomedical applications of
in vivo
tomographic imaging have been instrumental for development of modern medicine.
With the advent of imaging agents, functional and molecular imaging attracts more and more attention. The recent
focus is to unfold molecular and cellular activities, promising to a
ccelerate progress in diagnostic methods and
therapeutic options. As a powerful way for molecular imaging, we utilize reporter virus vectors. As a result, the reporter
gene expression can be captured noninvasively. The advancement of molecular biology has led to various sorts of
reporter genes responsible for positron emission, florescent emission, bioluminescent emission,
etc
. These reporter
genes would facilitate rapid translation from
in vitro
results to pre-clinical and clinical trials with the aid of various
imaging methodologies such as single photon emission computed tomography, positron emission tomography,
magnetic resonance imaging, and optical imaging. These image modalities are extensively used in small animals for
tumor detection, drug monitoring,
etc
. The radionuclide-based techniques are sensitive and permit tomographic
reconstruction, but they request expensive facilities. A competing approach for analysis of gene expressions is optical
imaging. Even though the optical signal can be very blurry due to scattering in the tissue, bioluminescence imaging
enjoys several distinct advantages. Particularly, it allows a rather high sensitivity, detecting a small number of cells
instead of a mass of 3-5mm in size using PET
1-6
.
In the bioluminescent imaging experiment, a small animal, like a mouse, is infected by inhalation with luciferase (Ad-
luc), and then is anesthetized and injected with luciferin. The chemical reaction between luciferase and luciferin emits
photons. The bioluminescent light travels through the tissues, and is captured with a CCD camera when it goes out from
the surface of the animal. The central issue of bioluminescent tomography is to reconstruct a bioluminescent source
∗
Corresponding author: ge-wang@uiowa.edu
Developments in X-Ray Tomography IV, edited by Ulrich Bonse, Proc. of SPIE Vol. 5535
(SPIE, Bellingham, WA, 2004) · 0277-786X/04/$15 · doi: 10.1117/12.560522
212
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 9/25/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
distribution inside a mouse from collected bioluminescent data on the surface of mouse. The bioluminescent photon
propagation in the tissue can be described by the transport equation or the Monte Carlo model
7-9
. Generally, The inverse
problem to identify the bioluminescent source is ill-posted, and more difficult than the related forward problem. Some
priori knowledge is required. Wang,
et al
. discussed the uniqueness for this inverse source problems
10
, and pointed out
the need of some priori knowledge, including the optical parameters of the tissues, the possible region for the light
source and 3D anatomic structure of the region of interest, which can be obtained with medical imaging techniques like
micro-CT scanning. In this work, we report a point spread function (PSF)-based model to describe the bioluminescent
light transport, and reconstruct a bioluminescent source distribution in a small animal. Using this method, we
reconstruct bioluminescent sources in both physical phantoms and living mice from bioluminescent data measured on
the surfaces of these objects. Specifically, the bioluminescent data on the gene expression from the live mouse liver are
combined with the corresponding micro CT segmentation to demonstrate the feasibility of
in-vivo
imaging for
localization and quantification of chemi-luminescent activities within the organ. Finally, the relevant issues and further
research directions are discussed.
2. MATERIALS AND METHODS
2.1 Light source reconstruction method
In a bioluminescent imaging experiment with a small animal, some biological cells, transfected with luminescent
reportors such as luciferase, produce bioluminescence. The photons travel in the tissues, reach the external surface, and
the surface output flux is recorded with a highly sensitive CCD camera. This light transport can be described with a
linear system,
i.e.
the light flux density
()
fr
at location
r
is
()
( ) (
)
0
00 0
,
d
fsp
Ω
=
∫
rrrrr
()
00
,
∈Ω
∈Ω
rr
(1)
where
()
0
s
r
is the light source density at location
r
0
, and
()
,
p
0
rr
is a point spread function (PSF), which is generally
spatially variant. The region
Ω
is the part of the small animal where the bioluminescent process takes place. The light
source
()
0
s
r
is distributed in a sub-region
0
Ω
(
0
Ω⊂Ω
) where the reporter genes are tagged with the cells.
Bioluminescence tomography is to determine light source distribution
()
0
s
r
in
0
Ω
from measurement of
()
f
r
on the
surface of
Ω
,
∂Ω
.
There are several methods to compute PSF
()
0
,
p
rr
. Monte Carlo method
8
is a flexible and accurate to study the
photon transport in turbid tissues. In an infinite homogeneous medium, the PSF, denoted as
()
0
,
p
0
rr
, depends only on
the distance
()()
()
00
r
=−⋅−
rr rr
between the source and detection position. In this case the
PSF can be easily
obtained with Monte Carlo simulation. Another choice is based on the diffusion equation. In bioluminescent imaging,
photons propagate in a highly scattering tissue, and the diffusion approximation is sufficiently accurate to describe light
propagation
9
. In a homogeneous medium,
()
0
,
p
0
rr
can be derived from the diffusion equation, as
()
()
()()
()
00
0
0
exp
,
4
eff
r
pr
Dr
μ
π
−
==−⋅−
rr
rr rr
, (2)
where
D
and
eff
μ
are the diffusion and effective attenuation coefficients, respectively, defined in terms of the
absorption coefficient
a
μ
, scattering coefficient
s
μ
and anisotropy factor
g
,
()
()
()
131
sa
Dg
μμ
=−+
and
()
()
()
1/ 2
31-
eff
a
a
s
g
μμμμ
=+
.
Since a small animal contains various types of tissues, the light propagation medium must be considered heterogeneous.
Hence, we decompose the diffusion and absorption coefficients as
()
()
0
DDD
δ
=+
rr
,
()
()
0
aaa
μμδμ
=+
rr
,
where
0
D
and
0
a
μ
are the
reference
values of the respective coefficients. The PSF in the heterogeneous medium can be
expressed in an integral form
9,11,12
:
() ()
()
()()
()
() ( )
()
000
00
,
,,
,
,d
a
pp
p
Dp
p
δδμ
Ω
=+
∇⋅∇−
∫
rr
rr
r
ζζζ
r
ζζ
r
ζ
. (3)
Proc. of SPIE Vol. 5535 213
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 9/25/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
where the Hamilton
∇
operates with
ζ
. According to the first-order perturbation theory, PSF
()
0
,
p
rr
can be
approximated with
11,12
:
() ()
()()()
()
()
()
()
()
000
000
00
0
,, ,,d , ,d
a
pp
pp
p
pD
δμ
δ
Ω
=−
−∇⋅∇
∫
rr
rr
r
ζζ
r
ζζ
r
ζζ
r
ζζ
. (4)
In practice, 3-D geometric structures of the tissues in the small animal can be obtained from an X-ray CT/micro-CT
scan, and the associated optical parameters (
s
μ
,
a
μ
,
g
) be taken from an optical database
20
. Therefore,
0
(
,
)
p
rr
in the
heterogeneous animal can be computed by (3) or (4). Thus, the bioluminescence tomography problem becomes to
reconstruct the light source distribution
()
S
x
inside the biological tissue region
0
Ω
, so that the surface flux calculated
via (1), with use of the accurate PSF (3) or the approximation (4), optimally matches the measured surface flux density
()
measured
f
x
. The reconstruction can be expressed as a minimization procedure
13
:
() ()
()
()
()
()
{
}
()
()( )
(
)
0
2
0
00 000
min
;
d
;,d,
measured
US
f
Sf
S
fS
s p
λη
Γ
≥≥
Ω
−+
=∈Ω∈∂Ω
∫
∫
rr
rrr
rrrrrrr
. (5)
in which
Γ
is a sub-domain of the external surface
∂Ω
,
η
is a stabilizing functional, and
λ
is a fixed regularization
factor. U represents that the unknown source density
S
must be constrained within the physically meaningful range. In
the experiment, the measured quantity is the output flux
()
Q
x
on
Γ
, which yields
()
measured
f
x
by
()
() ()
()
()
()
()
()
n
n
2
11
measured
fAQ
ARR
=−
=+
−
xxx
xxx
, (6)
in which
R
depends on the local refractive index
()
n
x
of the medium
14
.
2.2 Phantom experiments
Cylindrical tissue phantoms were made from a nylon material. Sizes of the specimens are 30mm in diameter and
30mm in height. The scattering coefficient, absorption coefficient, and anisotropy parameter for the material are
s
μ
=
34mm
-1
,
a
μ
=
0.003mm
-1
and
g
=0.90, respectively. These parameters were determined by light transmission
profile matching between experiment and numerical data
8, 9
.
Fig. 2. Bioluminescent experiment for phantom. (a-
c)
Luminescent views for one source phantom for rotation
angles of 0, 90 and 180 degrees respectively; (d-
f)
Luminescent views for two source phantom for rotation
angles of 0, 90 and 180 degrees respectively.
(a)
(c)
(b)
(e)
(f)
(d)
Fig 1. Biolumi
nescent phantoms. (a) One source
phantom (distance between cylindrical axis and source
mid-
point: 10mm), (b) Two sources phantom (distance
between cylindrical axis and source 1 & 2 mid-
points:
5mm and 10mm, respectively).
(a)
(b)
10mm
30mm
30mm
Source
5mm
10mm
30mm
30mm
Source
10mm
30mm
30mm
Source
10mm
30mm
30mm
Source
5mm
10mm
30mm
30mm
Source
5mm
10mm
30mm
30mm
Source
214 Proc. of SPIE Vol. 5535
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 9/25/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
The phantom specimens are schematically shown in Fig. 1(a) and (b). The emission sources are small polythene tubes
filled with red luminescent liquid, about 0.7mm in height and 0.5mm in radius. Two phantoms are prepared, one with a
single source placed 10mm off the axis of the specimen, the other with twin sources located at 10mm and 5mm off axis,
respectively. The experiments were performed in a dark environment. As schematically shown in Fig. 2 (a) to 2(f), a
specimen was placed on a sample holder in front of a nitrogen-cooled CCD camera, which captured the image of the
output flux Q from the cylindrical surface. The exposure time was 30 seconds. The sample holder was then rotated with
9
0
°
, and a second bioluminescent image. Finally, the third image was taken with the holder rotated with
1
80
°
. Due to
the symmetry, these profiles are sufficient to obtain complete light emission image on the cylindrical surface. After data
acquisition, the surface output flux Q was first calculated by transferring the pixel grey level with the CCD image into
light unit, incorporating the CCD reaction profile obtained with our meticulous calibration. Then, numerical simulations
were performed.
2.3 Mouse experiments
In the mouse liver studies, mice were injected via the tail vein with either 50μl of Ad5-CMV-Luciferase (titer: 1.8 x
1011pfu/ml) or 100μl of Ad5-CMVnuclear-targeted
β
-galactosidase (titer: 2 x 1010pfu/
ml). After three days the mice
were anesthetized with i.p. injection of 2.5% avertin with a volume of 100μl/10g body weight, Subsequently, the mice
were injected i.p. with 150mg/kg body weight of luciferin (Sigma). To detect the luciferase expression, the hair in the
abdominal area was removed using hair removal lotion (Nair), and gently restrained in the imaging chamber. After
luciferin administration, the mice were imaged using the CCD camera for 5 minutes. Three bioluminescent emission
profiles on the surface were taken with the mouse rotated by
0
°
,
9
0
°
and
1
80
°
, respectively, which were shown in
Fig. 3 (b). This three-image information reveals the transgene expression from transfected cells inside the mouse.
Pseudo-color images were generated for the bioluminescent views using the Adobe Photoshop software. Then, the
pseudo-color images were superimposed on the corresponding regions of the photographic images of the mouse, which
were shown in Fig. 3(a). After imaging, the mice were sacrificed and frozen in liquid nitrogen with the same body
posture as in the bioluminescence scanning. The frozen mouse was then scanned using a Siemens Sensation 16 CT
scanner and/or a SkyScan micro-CT scanner. The images of bioluminescent light emitted from the liver were
anatomically recognizable from all the three angles of view. The CT image volume were segmented into major
Anterior
-
posterior
Posterior
-
anter
ior
Left
-
lateral
0
W/Pixe
6.0E
-
15
Fig. 3.
I
n Viv
o
mouse liver bioluminescence tomography. In the mouse liver studies, we reconstructed
a bioluminescent tail vein with either 50 μl of Ad5-CMV-Lucif
erase (titer: 1.8 x 1011pfu/ml) or 100 μl
of Ad5-CMV-nuclear-targeted â-
galactosidase (titer: 2x 1010pfu/ml). Three days post injection, the
transgene expression was detected using the CCD camera or by X-
gal staining of liver tissue 5minutes
after luciferi
n administration. It was sequentially rotated 90 degrees to acquire photographic images
with lights on and bioluminescent views with lights off by exposing the camera for 0.1seconds and
10minutes, respectively. Pseudocolor images were generated for the bio
luminescent views to
superpose on the corresponding regions of the photographic images (a), and bioluminescent gray level
images (b). The hair in the abdominal area was removed using hair removal lotion (Nair) in the studies.
(b)
(a)
Proc. of SPIE Vol. 5535 215
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 9/25/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
anatomical components including the heart, lungs, liver, stomach, bones, and so on. Then, known optical parameters
(absorption coefficient
a
μ
, scattering coefficient
s
μ
and anisotropy factor
g
) were assigned to each of components, as
shown in Table 1
14
. A region of 250 by 400 pixels corresponds to liver part of the mouse was selected from the
photographic and bioluminescent views. The pixel gray levels of a bioluminescent view were transformed into
corresponding flux densities according to the intensity calibration relationship mentioned above. The transformed flux
densities contained some random noise and spurious effects of various types. These data were smoothened by low-pass
filtering.
Table 1. Optical parameters used in the source inversion.
Parameter
n
a
μ
(mm
-1
)
s
μ
(mm
-1
)
g
Muscle
1.37 0.01 4.0 0.90
Lung
1.37 0.35 23.0 0.94
Bone
1.37 0.002 20.0 0.90
Liver
1.37 0.18 20.0 0.90
Stomach
1.37 0.01 4.0 0.90
3. RESULTS
Fig. 2 shows the luminescent emission on the surface of the tissue phantom with a single cylindrical light source. It can
be observed that the brightest region is located at the middle portion of Fig. 2(a), moves to the right side of Fig. 2(b),
and becomes split to both sides of Fig. 3(c) with much less brightness. The surface output flux of these three
consecutive luminescent images is combined into a full surface profile as given in Fig. 4(a) and their corresponding
matching counterpart is shown in Fig. 4(b) that was numerically computed by the above described PSF method. Fig.
4(c) shows the real source location whereas Fig. 4(d) depicts the source position obtained by the numerical inversion
technique. The density peak of the recovered source locates exactly at the center of the real source. The total power of
the emission source is found to be 3.3x10
-12
watts, which is a bit lower than the real value of 3.51x10
-12
watts by a
relative error
of 6%.
(b)
(a)
(c)
(d)
Fig. 4. Experiment with the one source phantom. (a) the experimental surface flux and (b) estimated
counterpart of (a), as well as the corresponding source distributions (c) in the experiment and (d) from
the reconstruction.
Q
C
Q
C
Z
Z
216 Proc. of SPIE Vol. 5535
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 9/25/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Figure 5 presents the experimental and reconstruction results due to the presence of two light sources. The power is
3.5x10
-12
watts for both sources. The appearance of the bright region in Fig. 2(d) to 2(f) is similar to the corresponding
parts in Fig. 2(a) to 2(c), however, is broader in size and is brighter. The data processing is similar to that for single-
source case. The complete surface output flux profile, Fig. 5(a), is obtained by combining Fig. 2(d) to 2(f). The real and
reconstructed light source distributions are given in Fig. 5(c) and (d), respectively. The reconstruction accurately caught
the center locations of the real sources. The estimated sources powers are 3.17x10
-12
watts 3.12x10
-12
watts, which are
9.4% and 10.8% lower than the real values.
In vivo
study results on a mouse are presented in Figures 6-7.
Three consecutive bioluminescent images obtained are shown in Fig 3(a-b), superposed on the corresponding
photographs of the mouse body. The full output flux profile
Q
around the mouse surface is obtained as Fig 6(a). The
light source distribution
S
is then reconstructed with the proposed method and is presented in Fig. 6(b). The light source
is approximated as spherical sources distributed in the liver region of the mouse. The output flux corresponds to the
reconstructed light source is shown in Fig. 6(c). For better comparisons, representative slices are taken from Fig. 6(a)
and (c) with Z=3, 8, 13, 18, 23 and 28, respectively, and are shown as Fig. 7(a-f), with continuous lines for the
experimental data and dotted lines for the simulation results. Again, an excellent agreement was obtained, which
validates the proposed reconstruction method for in vivo bioluminescent study.
Fig. 5. Experiment with the two source phantom. (a) The experimental surface flux and (b) estimated
counterpart of (a), as well as the corresponding source distributions (c) in the experiment and (d) from
the
reconstruction.
Z
Q
Z
Q
C
C
(b)
(a)
(c)
(d)
Proc. of SPIE Vol. 5535 217
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 9/25/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
4. DISCUSSIONS AND CONCLUSION
The tissue phantoms have been fabricated with nylon material whose optical characteristics fall within the range of the
biological tissues. In the luminescent imaging experiment using phantoms with one or two sources, the brightness
variation in the images from different orientations is due to the proximity of the luminescent sources to the periphery of
the cylinder. The diffuse light flux pattern on the surface reveals that the photons have undergone absorption and
(a)
(b)
(c)
(d) (e) (f)
Fig. 7. Representative slice
s are taken from Fig. 6(a) and (c) with difference Z: (a) Z=3; (b)
Z=8; (c) Z=13; (d) Z=18; (e) Z=23 and (f) Z=28. Continuous lines correspond with Fig. 6(a),
dotted lines with Fig. 6(c).
Z
Q
C
Z
Q
C
Z
Q
C
Z
Q
C
Fig. 6. (a)
I
n vi
vo
bioluminescent output flux
Q
on the
surface of a mouse; (b) R
econstructed light source; (c)
Numerical results for
Q
with the reconstructed light
source. Z is with the longitudinal direction and C the
circumference direction.
(c)
(b)
(a)
218 Proc. of SPIE Vol. 5535
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 9/25/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
multiple scattering inside the medium. The experimentally measured surface output flux has been utilized to localize
and quantify the sources. To solve the inverse source problem, a PSF method has been developed and applied
successfully. The estimated source location information using the proposed method has well matched the true location
inside the phantom. The estimated source strength variation (up to 10.8%) can be explained by the inaccuracy in
determining the optical parameters and measuring the source flux. These phantoms experimental results have indicated
that the PSF method can be indeed employed for bioluminescence tomography.
Appropriate prior knowledge obtained by CT/micro-CT has been incorporated into the animal model to overcome the
ill-poseness of the inverse problem. In the bioluminescent imaging experiments, the mouse liver has been considered. It
is larger than the other organs and subject to transfect with adenovirus, which leads to enzymatic catalysis in the whole
organ. Hence, the light emitting sources has been spread in the entire liver. The signal from the source undergoes
multiple scattering and absorption before reaching the surface of the animal. The experiment data have shown that the
bioluminescent data have provided sufficient information for quantification of the underlying light source. Further work
is in progress to perform the
in-vivo
measurement of the optical parameters of various organs of the mouse, and to
improve the reconstruction algorithm for higher accuracy and faster speed.
In conclusion, tissue phantoms have been made and used in the experiments to evaluate the PSF method for localization
and quantification of an underlying bioluminescent source distribution. The preliminary results with both the phantoms
and living mice have demonstrated that bioluminescence tomography is indeed doable by incorporating the prior
knowledge obtained by CT/micro-CT. Further possibilities are open for bioluminescent imaging of the small animals.
We are actively pursuing along this direction, and will report more results in the future.
Acknowledgment
This work is partially supported by NIH/NIBIB grants (EB001685).
REFERENCES
1. M. Rudin and R. Weissleder, “Molecular imaging in drug discovery and development”,
Nat. Rev. Drug Discovery
,
2
, 123-131, 2003.
2. S. Wang, J. Petravicz, X. O. Breakefield, “Bioluminescence Imaging of regulated gene expression,using HSV-1
Amplicon vectors in rodent brain”,
Molecular Therapy
,
9
, S27,2004.
3. Christopher H. Contag and Michael H. Bachmann, “Advances in in vivo bioluminescence imaging of gene
expression”,
Annu. Rev. Biomed. Eng.,
4
, 235–60, 2002.
4. G. Vassaux and T. Groot-Wassink, “In Vivo Noninvasive Imaging for Gene Therapy”,
J. Biomed. Biotechnol
.,
2
,
92–101, 2003.
5. Pritha Ray, Anna M. Wu and Sanjiv S. Gambhir, “Optical Bioluminescence and Positron Emission Tomography
Imaging of Novel Fusion reporter Gene in Tumor Xenografts of Living Mice”,
Cancer Research
,
63
, 1160-1165,
2003.
6. B. W. Rice, M. D. Cable and M. B. Nelson, In vivo imaging of light-emitting probes, Jl. of Biomedical Optics, 6,
pp. 432-440, 2001.
7. S. R. Arridge, M. Schweiger, M. Hiraoka and d. T. Delpy, “A finite element approach for modeling photon
transport in tissue”,
Med. Phys
.,
20
, 299-309, 1993.
8. L-H. Wang, S. L. Jacques and L-Q Zheng, “MCML---Monte Carlo modeling of light transport in multi-layered
tissues”,
Comput Methods and Programs in Biomed.,
47
, 131-146, 1995.
9. S. R. Arridge, “Optical tomography in medical imaging”,
Inverse problems,
15
, R41-R93, 1999.
10. Ge Wang, Yi Li and Ming Jiang, “Uniqueness Theorems in Bioluminescence Tomography”,
Medical Physics
(In
press).
11. S. Economou,
Green’s Functions in Quantum Physics
, Springer-Verlag, New York, 1983.
12. V. A. Markel and J. C. Schotland, “Inverse problems in optical diffusion tomography. I. Fourier-Laplace inversion
formulas”,
J. Opt. Soc. Am. A.,
18
, 1336-1347.
13. P. P. B. Eggermont, “Maximum entropy regularization for fredholm integral equation of the first kind”,
SIAM J.
Math. Anal.,
24
, 1557-1576, 1993.
14. A. J. Welch and M. J. C. van Gemert,
Optical and Thermal response of laser-irradiated tissue
, Plenum Press, New
York, 1995.
Proc. of SPIE Vol. 5535 219
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 9/25/2018
Terms of Use: https://www.spiedigitallibrary.org/terms-of-use