See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228392483

Internal refractive index changes affect light transport in tissue

Article in Proceedings of SPIE - The International Society for Optical Engineering · July 2003
DOI: 10.1117/12.477365

CITATIONS READS
3 188

5 authors, including:

Hamid Dehghani Karthik Vishwanath

University of Birmingham Miami University
306 PUBLICATIONS 9,613 CITATIONS 111 PUBLICATIONS 1,304 CITATIONS

SEE PROFILE SEE PROFILE

Brian Pogue
Dartmouth College
1,022 PUBLICATIONS 28,534 CITATIONS

SEE PROFILE

All content following this page was uploaded by Karthik Vishwanath on 10 July 2014.

The user has requested enhancement of the downloaded file.

 Internal refractive index changes affect light transport in tissue
Ben Brooksby*1, Hamid Dehghani1, Karthik Vishwanath2, Brian W. Pogue1, Keith D. Paulsen1
1
Thayer School of Engineering, Dartmouth College, Hanover, New Hampshire 03755
2
Department of Physics and Astronomy, Dartmouth College, Hanover, New Hampshire 03755

ABSTRACT

This investigation explores the effect of index of refraction, as an optical property, on light transport through optically
turbid media. We describe a model of light propagation that incorporates the influence of refractive index mismatch at
boundaries within a domain. We measure light transmission through turbid cylindrical phantoms with different
distributions of refractive index. These distributions approximate the heterogeneous, layered nature of biological tissue.
Finite element method model calculations of diffuse transmittance through these phantoms show good agreement with
the trends observed experimentally. We see that phase measurements of light that propagates through approximately 90
(mm) of scatter-dominated media may vary by 10 degrees depending upon the values of refractive index of the medium.
Amplitude measurements are not as sensitive to this parameter as phase. Model calculations of diffuse reflectance from
a semi-infinite slab geometry with different layers also shows good agreement with Monte Carlo simulations. We
conclude that incorporating refractive index into light transport models may be worthwhile. Applying such a model in
tomographic image reconstruction may improve the estimation of optical properties of biological tissues.

Keywords: Numerical modeling, near infrared tomography, boundary conditions, refractive index

1. INTRODUCTION

The rapidly increasing use of light in both diagnostic and therapeutic medicine has created a need to understand how
radiation propagates in tissue. Mathematical models of light propagation in turbid media are required to estimate the
radiation field in tissue given its optical properties and the irradiation conditions. Such “forward” calculations are
necessary to predict the biological effect of therapeutic interventions such as laser surgery 1 or photodynamic therapy
(PDT) 2. The models are also used to infer the optical properties from measurements of light as it emerges from tissue.
These “inverse” calculations are characteristic of diagnostic techniques such as optical tomography 3 and near infrared
spectroscopy (NIRS) 4. Solution of the inverse problem usually involves iterative applications of a forward calculation
of the light field to improve estimates of the optical properties--absorption (µa) and transport scattering (µs/) coefficients.
Changes in optical properties can indicate physiological or pathological processes. While many homogeneous tissues
can be adequately represented by a unique set of µa and µs/ values, imaging heterogeneous tissues may also require
incorporation of a change in the index of refraction, n. While microscopic changes in n are generally interpreted as part
of the scattering coefficient, macroscopic changes in n are often overlooked in NIR imaging.
Imaging of tissue with near infrared (NIR) spectral tomography is emerging as a practicable method to map
hemoglobin concentration and oxygen saturation 5-10. Absorption due to tissue chromophores is relatively low in this
band. An optical fiber placed on the surface delivers an input optical signal (either continuous, amplitude modulated, or
ultra short pulses). Other optical fibers placed at different locations on the same surface detect the exiting photons that
have propagated through the volume under investigation. In the frequency domain, the amplitude and phase shift of
exiting photons provide information about the optical properties of the transilluminated tissue.
Monte Carlo simulations have been found to describe the propagation of light in tissue with high accuracy 11.
The drawbacks of such calculations is that they are very time consuming and application of the method to arbitrary
geometries with internal variations in properties is difficult to implement in an algorithm. Thus, the Monte Carlo
approach has historically not been widely used in a clinical environment. It is generally agreed that the propagation of
light in a strongly scattering medium such as tissue is accurately described by the equation of radiative transfer 12. The

*
Email: [email protected]; Tel: (603) 646-1094; Fax (603) 646-3856
radiation field at any point is characterized by the energy radiance, which is the energy transported by photons per unit
time in a given direction through a unit solid angle per unit area normal to that direction. If the radiance is not strongly
dependent on direction, it is possible to derive a much simpler equation for fluence rate (the number of photons passing
isotropically through the surface of a unit sphere per unit time). Previous results by a number of authors show that the
diffusion approximation to the radiative transfer equation is a suitable model to describe light propagation in scatter-
dominated materials where source and detection points are separated by at least a few scattering lengths 13. A finite
element method (FEM) algorithm can be implemented to provide accurate and reliable forward diffusion theory
calculations for arbitrary geometries and source and detector locations 14, 15. This approach is further studied in this
paper with the goal of adding the influence of internal changes in refractive index.
The refractive index of a material medium is the ratio of the phase speed of light in free space to that in the
medium. For optical frequencies, biological tissue is reasonably considered to be a dielectric, and accordingly n = ε r1 / 2
(where ε r is the relative permittivity). It is known that scattering characteristics of tissues are caused by microscopic
fluctuations in n. Refractive index is not listed in most of the literature as an optical property 16, 17 because it is a difficult
parameter to measure in turbid media. There have been some attempts to experimentally measure n of different tissues
18-20
, but in most numerical models index values from 1.33 to 1.6 are often assumed because the major constituent of
tissue is water. While n may be called a micro-index, the concept of a macro-index must be introduced to solve various
boundary problems
Optical fibers transporting laser light must be placed on the surface of the tissue. Diffusion theory must
account for this boundary if large errors are to be avoided in the determination of optical properties. There are three
common boundary conditions applied in diffusion theory formulas (type III or partial current, extrapolated, and Dirichlet
or zero) 15. Type III conditions can be modified to account for Fresnel reflections that arise from the refractive index
mismatch at the tissue-air interface 21. Models have been developed in order to describe light propagation across other
interfaces as well (i.e. tissue-tissue or tissue-fiber optic) 22-25. We establish a diffusion approximation model that
incorporates spatial variations in n inside turbid media. We apply a modified type III condition at interior boundaries
between two regions with different indices of refraction. We test this model in simulations, and perform two phantom
experiments. We compare experimental results and model calculations with Monte Carlo simulations, which we
consider to be accurate representations of the physical processes involved.

2. THEORY

Under the assumption that scattering dominates absorption in a region of interest, the Boltzmann transport equation can
be simplified to the diffusion approximation. In the frequency domain this is given by:
 iω 
− ∇ ⋅ D∇Φ (r, ω ) +  µ a + Φ(r, ω ) = S (r, ω ) (1)
 c 
where S (r, ω ) is an isotropic light source, Φ(r, ω ) is the photon density at position r, c is the speed of light in tissue, ω
1
is the frequency of light modulation, µa is the absorption coefficient, and D = is the diffusion coefficient.
3(µ a + µ s/ )
The best description of the air-tissue boundary is derived with an index-mismatched type III condition, in which
the fluence at the edge of the tissue exits and does not return. The flux leaving the boundary is equal to the fluence rate
at the boundary, times a factor that accounts for the internal reflection of light back into the tissue. This relationship is
described in the following equation 12, 14, 21, 26-29:
Φ(ξ , ω ) = 2 Anˆ ⋅ D∇Φ (ξ , ω ) (2)
where ξ is a point on the boundary, and A depends upon the relative refractive index mismatch between tissue and air.
A can be derived from Fresnel’s law 23:
2 / (1 − R 0 ) − 1 + cos θ c
3

A= 2
(3)
1 − cos θ c
where θ c = arcsin(n 2 n1 ) , the angle at which total internal reflection occurs for photons moving from a region with n1 to
(n n 2 − 1)
2

a region with n2, and R 0 = 1

. At external boundaries, n2=1, the index of refraction of free space.
(n 1
n + 1)
2

In finite element formalism, Φ(r, ω ) is discretized into a set of basis functions multiplied by weighting
coefficients, Φ = ∑ j =1 Φ jϕ j , which are determined as part of the solution process. In the Galerkin method of weighted
N

residuals, Eq. (1) is multiplied by an identical set of weighting functions, ϕ i , and it is integrated over the entire problem
domain. After further manipulation through Green’s theorem we reach Galerkin’s weak form 14, 30:
 iω 
D∇Φ ⋅ ∇ϕ i − ∫ n̂ ⋅ D∇Φϕ i ds +  µ a + Φϕ i = Sϕ i . (4)
 c 
The integrand of the surface integral is the normal component of the flux through the boundary surface. This term does
not contribute to matrix assembly at node points internal to the finite element mesh boundary because of flux continuity.
At interior nodes which lie on an interface between two media with different indices of refraction, we apply the
conditions used by Schmitt et al and Takatani and Graham 22, 25.
nˆ ⋅ D1∇Φ 1 (ξ , ω ) = nˆ ⋅ D2 ∇Φ 2 (ξ , ω ) (5a)
2
Φ 1 (ξ , ω )  n1 
=  (5b)
Φ 2 (ξ , ω )  n 2 
Equation (5a) ensures that the flux across a boundary remains continuous (so the flux integral in (4) still does not
contribute at internal nodes), while equation (5b) establishes a discontinuity in the fluence based upon the two refractive
indices defining the interface.
Implementation of (5b) in (4) occurs by separating the contributions from the domain integrations in the
Galerkin weak form during matrix assembly according to the side of the interface a node belongs at an index of
refraction mismatch. A simple vehicle for accomplishing this task is to duplicate nodes on the interfaced such that they
possess coincident coordinate positions but unique node numbers (and therefore unique columns in the global system of
algebraic equations when serving as basis functions that define the numerical photon density solutions). This allows Φ 1
and Φ 2 to simultaneously exist in the list of unknowns at the index-mismatched interface. On matrix assembly, one of
the node numbers is preselected to act as the weighting function, and therefore, provide the row location for the
discretized version of (4) on the interface. The row number associated with the partnered node is initially empty and to
close the algebraic system, it is ready to accept explicit enforcement of (5b) rewritten as the equation
2
n 
Φ 1 −  1  Φ 2 = 0 (6)
 n2 

3. METHODS AND RESULTS

Two experiments were performed, which consisted of collecting data using a near infrared tomography imaging system
31
. This system is designed to image breast tissue using light with wavelengths of 660-860 (nm). It is well characterized
and currently being used in a clinical trial for the diagnosis of breast disease. Sixteen evenly spaced fibers contact the
surface of the pendant breast. One fiber acts as the source and the other fifteen as detectors. Here, 785 (nm) light
transmission data was collected using two cylindrical phantoms with known homogeneous optical coefficients µa and
µs/. The phantoms are made using a mixture of resin, ink (for absorption), and TiO2 (for scattering).
In the first experiment we began by measuring transmission through a solid cylinder (µa=0.0055 (mm-1) and
µs =0.871 (mm-1)) with a diameter of 85 (mm). We then drilled a large hole through its center (d=66 (mm)), filled it
/

with a mixture of water and intralipid with the same µa and µs/ as the plastic, and measured again. We used the imaging
system to confirm differences in µa and µs/ for the two materials of less than 1%. The resin is transparent, and we
measured its refractive index by observing the angle by which it deflects a collimated beam. We assume this value
(n=1.58) is approximately equal to n of the solid phantom material, and we take n of the intralipid mixture as that of
water (n=1.33). Assuming no change in absorption or scattering of the medium, we look to relate the changes in the
measurements to the change in n within the phantom.
In Figure 1, the solid line connecting stars indicates amplitude and phase measurements (for one source and
fifteen detectors, where detector #8 is located on the opposite side of the cylinder) acquired when cylinder #1 existed as
a solid homogeneous object. The dashed line connecting stars represents measurements recorded after the phantom was
drilled into a hollow cup and filled with intralipid. The data suggests that decreasing n in the center of the medium
causes amplitude at detector #8 to decrease slightly. Decreasing n in the center has a greater effect on the phase, causing
a decrease of nearly ten degrees at detector #8.

Fig. 1. Measured and simulated data from Cylinder #1. n1 and n2 indicate the values in the exterior and interior of the medium,
respectively. Solid lines mark data recorded from phantom with a single n, and dashed lines mark data acquired from the phantom
after its interior was replaced with a lower value.

NIRFAST 14 (Near Infrared Frequency-domain Absorption and Scatter Tomography), an algorithm for
numerical modeling of light transport and image reconstruction, was used to solve the diffusion equation (Eq. 1), subject
to the boundary conditions (Eq. 2 and Eq. 5). The solid and dashed lines connecting circles represent the two
dimensional (2D) model calculations while the solid and dashed lines connecting triangles represent the three
dimensional (3D) model computations. Both models also show slight decreases in amplitude and a decrease in phase of
approximately 17 degrees when the n in the center of the phantom is reduced. In Figure 1, the calculated data are
normalized to the values measured at detector number one.
The same procedures were followed with a second phantom to investigate a different change in n. In this case
the homogeneous reference was a cylinder filled with intralipid solution (µa =0.0164 (mm-1), µs/=0.33 (mm-1), n=1.33,
d=92 (mm)), and the mismatch was created with a solid inclusion in the interior (n=1.58, d=73 (mm)) having the same
absorption and scattering properties. The results from this experiment are shown in Figure 2. With the inclusion of a
higher n, we measured a 4 degree increase in phase, and calculated a 7 degree increase in phase with both the 2D and 3D
models. We found virtually no change in the measurements, or in the model calculations of amplitude before and after
the change in n.

Fig. 2. Measured and simulated data from Cylinder #2. Solid lines mark data taken from the phantom with a single n, and dashed
lines mark data taken from the phantom after its interior was replaced with a higher n. With the change, there is very little effect on
amplitude, and a more substantial increase in phase.

To further test the validity of our model, we compared it with Monte Carlo simulations of diffuse reflectance
from a semi-infinite slab geometry. The Monte Carlo code had been validated for fluorescence lifetime spectroscopy 32.
We performed calculations using a slab with a thin top layer (5 (mm)) and a much larger (semi-infinite) bottom layer.
We allow for different combinations of n in the different layers. A point source of photons was located in the center of
the top surface and detectors measure photons exiting the slab at even distance intervals away from the source. Figure 3
shows the amplitude and phase data from simulations when the refractive index in both layers is equal (n1=n2=1.33), and
when it increases in the bottom layer (n1=1.33; n2=1.58). FEM and Monte Carlo calculations are in good agreement, and
show comparable phase increases at distances greater than 10 (mm) away from the source when the change is
introduced. There is also a minimal decrease in amplitude at distances of 5-30 (mm). Figure 4 shows the case where
both layers have the same refractive index (n1=n2=1.58), and when it decreases in the bottom layer (n1=1.58; n2=1.33).
Both models show decreases in phase at distances greater than 15 (mm), and slight increases in amplitude 10-30 (mm)
from the source.
Fig. 3. Finite element method (FEM) (top two graphs) and Monte Carlo (bottom two graphs) calculations of diffuse reflectance from
a semi-infinite slab geometry with layers are shown. Amplitude (left) and phase shift (right) are plotted vs. distance from the source
in both cases. The two models show qualitative agreement when refractive index in the lower region of the slab increases.

Fig. 4. Graphs of calculated amplitude (left) and phase shift (right) are shown for a thick flat slab with homogeneous
refractive index, and then when the refractive index decreases in the lower region. Again, Finite element method (FEM)
calculations are shown in the top two graphs, and Monte Carlo data are shown in the bottom two plots.
 We hypothesize that a model which accurately accounts for heterogeneities in index of refraction will improve
biological tissue absorption (µa) and scattering (µs/) image reconstructions. Figure 5 (a) shows a possible distribution of
optical properties µa, µs/, and n. We calculate light transmission through this circular geometry, and then attempt to
reconstruct 2D images of µa, µs/. Figure 5 (b) shows the reconstructions when a model was used that did not account for
internal boundary reflections due to differences in n. When we use the model that does allow for these reflections, we
obtain the images in Figure 5 (c). These images recover both the absorbing and scattering anomalies with better
quantitative accuracy, and contain fewer artifacts. The improvements due to the n-sensitive model are evident in the
cross sections of these reconstructions (Figure 5 (d)).

(a)

absorption coefficient scattering coefficient index of refraction

(b) (c)

absorption coefficient scattering coefficient absorption coefficient scattering coefficient

(d)
Fig. 5. (a) The target optical properties used in forward simulations are shown, including absorption coefficient, scattering coefficient
and index of refraction. In this case, the effect of the outer ring of different index of refraction upon the image reconstruction
algorithm is tested. Reconstructions of property distributions shown in (a) using a model which is insensitive to the refractive index
change is presented in (b). Reconstruction using an algorithm that incorporated the change in refractive index a priori is shown in
(c). Cross sections of the properties reconstructed in (b) and (c) are shown in (d).
 4. DISCUSSIONS AND CONCLUSION

The FEM model calculations do not show exact quantitative agreement with the experiments reported here (Figures 1
and 2), or the Monte Carlo simulations (Figures 3 and 4), but do represent good qualitative agreement. This work is a
first attempt at incorporating multiple refractive indices into our model of light propagation through turbid media.
Experimental measurements of diffuse light transmission appear sensitive to heterogeneities in refractive index. For
example, in the circular array of sources and detectors that we use in tomographic imaging we see that an increase in n in
the center of the phantom volume causes an increase in phase and minimal change in amplitude measured at distant
detectors. This result is logical considering that wavelength is reduced when n is higher, and more reflection occurs
when photons pass into regions with lower n (based upon Fresnel’s law). The increase in n in the center should not
directly affect how many photons are absorbed, but the photon pathlength should increase. This phantom could
represent the heterogeneous refractive index distribution of breast tissue. Adipose and glandular tissues have different
compositions.
More experiments are necessary to confirm the consistency of this model and experimental observations. Our
model in both 2D and 3D shows the same trends in amplitude and phase as Monte Carlo simulations. In the cylindrical
imaging geometry, the model calculations of phase sensitivity to n slightly overestimate what we observed with our
imaging system. The model calculations and measurements of amplitude reveal almost no variation. In the future we
plan further development of this model. Refractive index clearly has an effect upon measurements in diffuse optical
tomography, but it is a difficult parameter to measure independently. Experiments are required that will elucidate this
affect.

5. ACKNOWLEDGEMENTS

Funding for this work has been provided from National Cancer Institute through research grants PO1CA80139 and
RO1CA69544.

6. REFERENCES
1
Boulnois, J.-L.,Photophysical processes in recent medical laser developments: A review Lasers in Medical
Science 1, 47 (1986).
2
Johansson, T., Thompson, M. S., Stenberg, M., Klinteberg, C., Andersson-Engels, S., Svanberg, S., and
Svanberg, K.,Feasibility study of a system for combined light dosimetry and interstitial photodynamic treatment
of massive tumors Applied Optics 41, 1462 (2002).
3
Arridge, S. R.,Optical tomography in medical imaging Inverse Problems 15, R41 (1999).
4
Arridge, S. R., van der Zee, P., Cope, M., Delpy, D. T.,Reconstruction methods for infrared absorption imaging
Proc. SPIE 1431, 204 (1991).
5
Pogue, B. W., Zhu, H., Nwaigwe, C., McBride, T. O., Osterberg, U. L., Paulsen, K. D., Dunn, J. F.,Hemoglobin
Imaging With Hybrid Magnetic Resonance and Near-Infrared Diffuse Tomography Adv,. Expt. Med. Biol., (in
press) (2000).
6
Jiang, S., Pogue, B. W., McBride, T. O., and Paulsen, K. D.,Quantitative Analysis of Near-Infrared
Tomography: Sensitivity to the Tissue-Simulating Pre-Calibration Phantom J. Biomed. Opt. (in press) (2002).
7
Hebden, J. C., Veenstra, G., Dehghani, H., Hillman, E. M. C., Schweiger, M., Arridge, S. R., and Delpy, D.
T.,Three dimensional time-resolved optical tomography of a conical breast phantom Applied Optics 40, 3278
(2001).
8
Fantini, S., Schutz, O., Edler, J., Franceschini, M. A., Heywang-Krbrunner, S., Gotz, L., Siebold, H., in SPIE
Technical Abstract Digest, edited by B. Chance (SPIE Press, San Jose, CA, 1999), p. 103.
9
Bremer, C., Ntziachristos, V., Mahmood, U., Tung, C. H., and Weissleder, R.,[Progress in optical imaging]
Radiologe 41, 131 (2001).
 10
Boas, D., Gaudette, T., Arridge, S.,Simultaneous imaging and optode calibration with diffuse optical
tomography Opt. Exp. 8, 263 (2001).
11
Flock, S. T., Wilson, B. C., and Patterson, M. S.,Monte Carlo modeling of light propagation in highly
scattering tissues--II: comparison with measurements on phantoms IEEE Trans. Biomed. Eng BME-36, 1169
(1992).
12
Patterson, M. S., Moulton, J. D., Wilson, B. C., Berndt, K. W., Lakowicz, J. R.,Frequency-domain reflectance
for the determination of the scattering and absorption properties of tissue Appl. Opt. 30, 4474 (1991).
13
Dehghani, H., Pogue, B. W., Poplack, S. P., and Paulsen, K. D.,Multiwavelength Three-Dimensional Near-
Infrared Tomography of the Breast: Initial Simulation, Phantom, and Clinical Results Appl. Opt. 42, 135
(2003).
14
Dehghani, H., Pogue, B. W., and Paulsen, K. D., in OSA Biomedical Topical Meetings, Miami Beach, Florida,
2001), Vol. 2002 Technical Digest.
15
Jiang, H., Paulsen, K. D., in Proc. SPIE: Optical Tomography, Photon migration and spectroscopy of tissue
and model media, edited by B. Chance, Alfano, R. R., San Jose, CA, 1995), Vol. 2389, p. 608.
16
Cheong, W., Prahl, S. A., and Welch, A. J.,A review of the optical properties of biological tissues IEEE journal
of quantum electronics 26, 2166 (1990).
17
Durduran, T., Choe, R., Culver, J. P., Zubkov, L., Holboke, M. J., Giammarco, J., Chance, B., and Yodh, A.
G.,Bulk optical properties of healthy female breast tissue Phys. Med. Biol. 47, 2847 (2002).
18
Li, H. and Xie, S.,Measurement method of the refractive index of biotissue by total internal reflection Applied
Optics 35, 1793 (1996).
19
Bolin, F. P., Preuss, L. E., Taylor, R. C., and Ference, R. J.,Refractive index of some mammalian tissues using a
fiber optic cladding method Appl. Opt. 28, 2297 (1989).
20
Bartlett, M. A. and Jiang, H.,Effect of refractive index on the measurement of optical properties in turbid media
Appl. Opt. 40, 1735 (2001).
21
Arridge, S. R., Schweiger, M.,Sensitivity to prior knowledge in optical tomographic resconstruction Proc. SPIE
2389, 378 (1995).
22
Takatani, S. and Graham, M.,Theoretical analysis of diffuse reflectance from a two-layer tissue model IEEE
Trans. Biomed. Eng 26, 656 (1979).
23
Keijzer, M., Star, W. M., and Storchi, P. R. M.,Optical diffusion in layered media Appl. Opt. 27, 1820 (1988).
24
Kienle, A., Patterson, M. S., Dognitz, N., Bays, R., Wagnieres, G., and van den Bergh, H.,Noninvasive
determination of the optical properties of two-layered turbid media Appl. Opt. 37, 779 (1998).
25
Schmitt, J. M., Zhou, G. X., Walker, E. C., and Wall, R. T.,Multilayer model of photon diffusion in skin J. Opt.
Soc. Am A 7, 2141 (1990).
26
Farrell, T. J., Patterson, M. S., and Wilson, B. C.,A diffusion theory model of spatially resolved, steady-state
diffuse reflectance for the noninvasive determination of tissue optical properties in vivo Medical Physics 19,
879 (1992).
27
Groenhuis, R. A. J., Ferwerda, H. A., ten Bosch, J. J.,Scattering and absorption of turbid materials determined
from reflection measurements. I.Theory. Appl. Opt. 22, 2456 (1983).
28
Haskell, R. C., Svaasand, L. O., Tsay, T. T., Feng, T. C., McAdams, M. S., and Tromberg, B. J.,Boundary
conditions for the diffusion equation in radiative transfer J. Opt. Soc. Am. A 11, 2727 (1994).
29
Hielscher, A. H., Jacques, S. L., Wang, L. V., and Tittel, F. K.,The influence of boundary conditions on the
accuracy of diffusion theory in time-resolved reflectance spectroscopy of biological tissue Phys. Med. Biol. 40,
1957 (1995).
30
Lapidus, L. L. and Pinder, G. F., Numerical Solution of Partial Differential Equations in Science and
Engineering (John Wiley & Sons, Inc., New York, 1999).
31
McBride, T. O., Pogue, B. W., Jiang, S., Osterberg, U. L., and Paulsen, K. D.,Development and Calibration of
a Parallel Modulated Near-Infrared Tomography System for Hemoglobin Imaging In Vivo Rev. Sci. Instr. 72,
1817 (2001).
32
Vishwanath, K., Pogue, B. W., and Mycek, M. A.,Quantitative Fluorescence Lifetime Spectroscopy in Turbid
Media: Comparison of Theoretical, Experimental and Computational Methods Phys. Med. Biol. 47, 3387
(2002).

View publication stats