898 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 2, NO.

4, DECEMBER 1996

Three-Dimensional Computation
of Light Scattering From Cells
Andrew Dunn and Rebecca Richards-Kortum

Abstract— Using the finite-difference time-domain method, makeup of the tissue and will influence the amplitude of the
three-dimensional scattering patterns are computed for cells scattered light. A larger difference in the index of refraction
containing multiple organelles. The scattering cross section and between a cell component and its surrounding will result in
average cosine of the scattering angle are computed for cells as
a function of volume fraction of melanin granules and mitochon- increased scattering. A better understanding of the relationship
dria. Results show that small organelles play a significant role in between the biochemical and morphological structure of cells
light scattering from cells, and the volume fraction of organelles and light scattering will aid in the development of diagnostic
affects both the total amount of scattered light and the angular applications.
distribution of scattered light. Mie theory has been used to approximate tissue scattering at
the cellular level by assuming cells are homogeneous spheres
I. INTRODUCTION of a single size [5]. While Mie theory has been successful in
describing light propagation in bulk tissue, it cannot describe
R ECENTLY, the potential of diagnostic optical imag-
ing has generated considerable interest in the optical
properties of tissues and cells at near infrared wavelengths,
the light interaction with complex structures such as individual
cells containing multiple organelles. Mie theory can describe
where scattering is dominant over absorption. Both direct and the scattering of individual spherical organelles, but it cannot
indirect imaging applications rely on the scattering properties account for a combination of organelles within a cell.
of tissue to infer information about the physiological state of To account for scattering that arises from inhomogeneous
the tissue. Confocal microscopy [1] and optical coherence objects of arbitrary shape, a more complex model is needed.
tomography (OCT) imaging [2], which are direct imaging In this paper we present the application of the finite-difference
techniques, provide high-resolution images of human tissue in time-domain (FDTD) method to model light scattering from
vivo by constructing a three-dimensional (3-D) backscattering cells containing multiple organelles. The FDTD technique
map that is a direct image of tissue morphology. Indirect is a powerful computational method that is a full vector,
techniques, such as near infrared spectroscopy techniques, 3-D solution of Maxwell’s equations. Complex objects of
relate changes in the amount of diffusely backscattered light any dielectric structure can be modeled, and the scattered
to the optical properties of tissue [3], [4]. Attempts to relate electromagnetic fields can be calculated for cells containing
directly or indirectly measured changes in tissue scattering a nucleus and other organelles. The angular distribution of
to cellular physiological differences, such as those between scattered light, or scattering pattern, anisotropy value, , and
normal and cancerous cells, have been difficult since the scattering cross section are computed. Simulations of cells
mechanisms by which cellular changes affect light scattering containing various volume fractions of various organelles are
are not fully understood. This is primarily due to the difficulty performed to assess the effect of these organelles on tissue
in describing scattering at the cellular level since cells are scattering. Specifically, mitochondria and melanin are studied
complex structures with a spatially varying index of refraction. since it has been suggested that each plays an important role in
There are two important parameters that influence the scat- tissue scattering. In the first section the FDTD method is briefly
tering properties of tissue: tissue morphology and biochem- outlined, and sufficient detail is provided to implement the
istry. Tissue morphology, or the relative size of organelles technique for cell scattering calculations, then some example
and cells with respect to the wavelength of light, affects the results are presented.
angular distribution of the scattered light. Generally, a tissue
component whose size is small compared to the wavelength II. FDTD METHOD
will scatter light more isotropically than larger components,
but the total amount of scattered light will be less for the
smaller component due to its smaller size. The index of A. Yee’s Algorithm
refraction of each component is determined by the biochemical The FDTD method, first described by Yee [6], is a direct
solution of Maxwell’s equations in the time domain. It has
Manuscript received September 26, 1996. This work was supported in part
by the National Science Foundation under Grant BCS-9 157 202 and the Center been used in a wide range of applications including electro-
for High Performance Computing at the University of Texas. magnetic absorption of tissue in hyperthermia [7], scattering
The authors are with the Biomedical Engineering Program, Department of cross-section calculations of arbitrary objects [8], scattering
Electrical and Computer Engineering, University of Texas, Austin, TX 78731
USA. from frequency-dependent materials [9], as well as many other
Publisher Item Identifier S 1077-260X(96)09666-9. applications [10].
1077–260X/96$5.00  1996 IEEE

Authorized licensed use limited to: University of Texas at Austin. Downloaded on February 24,2025 at 00:14:01 UTC from IEEE Xplore. Restrictions apply.
DUNN AND RICHARDS-KORTUM: THREE-DIMENSIONAL COMPUTATON OF LIGHT SCATTERING 899

Yee’s algorithm begins with Maxwell’s two curl equations

(1)

and discretizes them in space and time, resulting in a set

of six explicit finite-difference equations. The electric and
magnetic fields are spatially and temporally offset on a 3-
D grid. Examples of the finite-difference equations for the
component of the electric field, , and component of the
magnetic field, , are

Fig. 1. Geometry of the FDTD simulation demonstrating how the cell is

constructed by assigning each component a different index of refraction.

(2)
paper a sinusoidal plane wave source of the form is
propagated in the direction.
As illustrated in Fig. 1, the cell is constructed within the
grid by assigning a value of the permittivity to each region, or
organelle, within the cell. In this way, the cell is constructed
as a dielectric object that can have any number of components
of varying shapes and sizes.
The total size of each problem is limited, however, by
(3) computational considerations since the grid spacing must
be less than [11]. This is due to the discretization of the
Similar equations exist for the remaining electric and magnetic fields, and step sizes greater than this lead to unstable results.
field components [11]. In (2) and (3), is the location For each simulation, values of the six field components and
of a grid point, and and are the spatial and temporal step permittivity must be stored in memory at each point in the
sizes. The time step is indicated by the superscript , and computational domain. Thus the size of the computational
and are the electric permittivity and magnetic permeability, domain, and therefore storage requirements, are dictated by
respectively. The six finite-difference equations are stepped the wavelength and size of the object. In three dimensions,
in time, alternately updating the electric and magnetic field even moderately sized problems will exhaust the capabilities
components at each grid point. of current personal computers. However, the FDTD method is
To simulate propagation in an unbounded medium, special well suited for parallel processing [17], making 3-D problems
boundary conditions must be applied to the tangential elec- more tractable.
tric field components along the edges of the computational
boundaries at each time step, as indicated in Fig. 1. A variety B. Far-Field Transform
of boundary conditions exist [12] and are usually based
on either a solution of the wave equation [13], [14] to The FDTD method computes the fields in a region around
prevent artificial reflections at the edges of the computational the cell that lies in the near field. To determine the far-field
domain or an artificial material absorber around the edges to scattering pattern, the near-field data can be transformed to the
effectively damp out any reflections [15]. The second-order far field by weighting it with the free-space Green’s function
Mur absorbing boundary condition was used in this paper [13]. and integrating over a surface , surrounding the cell (Fig. 1)
The incident wave is not included in the finite-difference [8], [18].
equations for the field components and therefore is applied The near-field time-domain values are first converted to
separately. The total field/scattered field formulation [16] can the frequency-domain values with a discrete Fourier transform
be used to propagate a variety of excitation waveforms, [19], and the equivalent electric and magnetic surface current
including a continuous sinusoid and a Gaussian pulse. In this densities, and , are defined on the surface as [20]
formulation the linearity of Maxwell’s equations is used to (4)
split the total fields into incident and scattered fields. In the
interior of the computational domain, the total fields (incident
and scattered) are calculated, while around the edges the where is a point on and is a unit vector normal to
scattered and incident fields are treated separately. In this , where the origin of the far-field transform is located in the

Authorized licensed use limited to: University of Texas at Austin. Downloaded on February 24,2025 at 00:14:01 UTC from IEEE Xplore. Restrictions apply.
900 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 2, NO. 4, DECEMBER 1996

Fig. 2. Scattering patterns for a cell with only a nucleus and cytoplasm (solid line) and a cell with mitochondria [8.5% volume fraction (dotted line)].

center of the grid. The vector potentials and are computed TABLE I
by numerically integrating and over the surface INDEX OF REFRACTION VALUES USED IN THE FDTD SIMULATION

(5)

where is a point in the far field and is the angle between

and . The far-field electric and magnetic field components,
in spherical coordinates, are then given by of refraction of cells and organelles, and the values listed
in Table I were taken from the literature and used in these
(6) simulations. The cell shown in Fig. 1 is ellipsoidal with a
mean diameter of approximately 11 m and the diameter of
the nucleus is 3 m, while the diameter of organelles, such as
mitochondria, is uniformly distributed between 0.4 and 1 m.
The simulation program was verified by computing the
scattering pattern of homogeneous spheres ranging in diameter
from 1 to 12 m and comparing the results to Mie theory. The
where . The far-field scattering pattern is grid spacing in the simulations was /20 ( 900 nm) and a
defined by sinusoidal source, propagating in the direction, was stepped
in time until sinusoidal steady state of the scattered fields was
(7) reached, which typically required 3 to 4 passes through the
grid. After completion of the time stepping, the scattering
where and are the angles measured from the and pattern, , was computed in one-degree increments for
axes, respectively, in spherical coordinates. Physically, this both and . Simulations were run on a Cray J90 computer,
represents the scattered intensity at any point in the far field. and each simulation typically required approximately 100
Other parameters such as the anisotropy and scattering cross Mword of storage and 30 min of system CPU time for a 12- m
section can be computed from the scattering pattern diameter object. The FDTD computed patterns agreed with the
(see Section IV). Mie theory patterns to within 1%–2% at all angles.

III. SIMULATION PARAMETERS IV. RESULTS

An FDTD code was written to compute the far-field scat- The scattering patterns for two cells are plotted in Fig. 2.
tering pattern of cells containing multiple organelles. In Fig. 1 The lower curve is the pattern for a cell with only a nucleus,
a cross section of a typical cell is shown, where the shaded while the other is for the same cell containing a nucleus and
regions represent different values of permittivity, or index of mitochondria with a volume fraction of 8.5%. The scattering
refraction. There have been few published reports of the index pattern , has been averaged over the azimuthal angle

Authorized licensed use limited to: University of Texas at Austin. Downloaded on February 24,2025 at 00:14:01 UTC from IEEE Xplore. Restrictions apply.
DUNN AND RICHARDS-KORTUM: THREE-DIMENSIONAL COMPUTATON OF LIGHT SCATTERING 901

Fig. 3. Scattering patterns for a cell with only a nucleus and cytoplasm (solid line) and a cell with melanin [8.5% volume fraction (dotted line)].

and is therefore only a function of pattern was normalized to obtain a phase function

(9)

(8)
The average cosine , of , was then computed

(10)
The pattern is highly peaked in the forward direction, and
the largest differences in amplitude between the two cells occur
at angles above 90 . Below 90 the relative difference in This value of was then used for the Henyey–Greenstein
amplitude is less pronounced. phase function to compare the phase functions. Fig. 4(a) and
In Fig. 3 the scattering patterns , for a cell with a (b) shows the comparison of the Henyey–Greenstein and
nucleus only and a cell containing a nucleus and melanin with computed phase functions for the cell containing mitochon-
a volume fraction of 8.5%, are plotted. There is a significant dria [Fig. 4(a)] and melanin [Fig. 4(b)]. The value of for
increase in the amount of scatter at all angles relative to the the cell containing mitochondria is 0.992, indicating almost
cell without melanin. complete forward scatter. The Henyey–Greenstein phase func-
To investigate the effect of the nucleus, the scattering tion for 0.992 shows significant differences from the
patterns of a cell with and without a nucleus were compared cell phase function at almost all angles. In the cell with
(data not shown). There was little difference in the patterns melanin [Fig. 4(b)], the value of is smaller (0.944) and the
except for a slight increase in amplitude at small angles ( Henyey–Greenstein phase function matches somewhat better,
10 ). To compare the relative contributions of the nucleus and although there are still significant differences. The lower
small organelles, the scattering patterns of a cell containing value is the result of increased scattering at higher angles
organelles and a nucleus, and a cell containing only organelles (Fig. 3).
were compared. There is very little difference in the two The volume fraction of a particular organelle in cells will
patterns indicating that at near infrared wavelengths, small vary depending on tissue type. To investigate the effect of
organelles play a more important role than the nucleus in organelle volume fraction on scattering, the average cosine
scattering from a cell, and the effect of the nucleus on the and the scattering cross section were computed for cells
scattering patterns for a cell with organelles is overshadowed containing differing amounts of organelles. The scattering
by the scatter due to the organelles. cross section was computed by integrating the scattering
The Henyey–Greenstein phase function [21] is often used pattern over all angles
to characterize the angular distribution of scattered light by
tissue and is characterized by the average cosine of the
(11)
scattering angle . Since the Henyey–Greenstein phase func-
tion is a probability density function, it is normalized to an
area of one. To compare the computed scattering pattern to In Fig. 5, and are plotted as a function of organelle vol-
the Henyey–Greenstein phase function , the scattering ume fraction for organelles representing mitochondria. When

Authorized licensed use limited to: University of Texas at Austin. Downloaded on February 24,2025 at 00:14:01 UTC from IEEE Xplore. Restrictions apply.
902 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 2, NO. 4, DECEMBER 1996

(a)

(b)
Fig. 4. Comparison of the Henyey–Greenstein phase function and the FDTD computed phase functions. (a) Cell with mitochondria (8.5% volume fraction).
(b) Cell with melanin (8.5% volume fraction).

the organelle volume fraction is zero, the cell consists only both the amplitude and angular distribution of scattered light.
of a nucleus and cytoplasm. The scattering cross section Recently, it was demonstrated that the shape of the phase
increases with organelle volume fraction, while the anisotropy function can play a significant role in reflectance measure-
value decreases, indicating that the organelles contribute to ments when the source and detector are positioned close to
high angle scattering. In Fig. 6, the scattering cross section each other [22]. While Mie theory can provide some insight
and anisotropy are plotted for increasing amounts of melanin into the scattering from cells, it cannot predict the influence
within the cell. The scattering cross section increases with of organelles on the scattering patterns of cells. In Fig. 2
volume fraction, but the anisotropy remains relatively constant the major difference in the scattering patterns between a cell
with increasing volume fraction after an initial decrease. with and without organelles lies at angles greater than 90 .
At angles up to about 50 the patterns have the same basic
shape, indicating that the cytoplasm, or overall cell size and
V. DISCUSSION shape, determines the forward scattering properties. At higher
The scattering pattern of tissue structures, such as cells, is angles, the role of the small organelles is more pronounced,
important to understand because it contains information about which could be predicted qualitatively since they are small

Authorized licensed use limited to: University of Texas at Austin. Downloaded on February 24,2025 at 00:14:01 UTC from IEEE Xplore. Restrictions apply.
DUNN AND RICHARDS-KORTUM: THREE-DIMENSIONAL COMPUTATON OF LIGHT SCATTERING 903

Fig. 5. Scattering cross section, C , and anisotropy value, g , as a function of organelle volume fraction (nrel = 1:03).

Fig. 6. Scattering cross section, C , and anisotropy value, g , as a function of melanin volume fraction.

relative to the wavelength and will act as isotropic scatterers. those of epithelial tissues [24] but are slightly underestimated
To predict the increase in high angle scatter quantitatively, because the internal structure of each organelle has not been
however, the electromagnetic fields must be calculated. taken into account.
The macroscopic scattering coefficient due to a collection The role of melanin in determining light distribution in
of cells can be estimated from the scattering cross section tissues has been largely that of an absorber. However, in
by [23] individual cells, its high index of refraction relative to the
surrounding cell components causes a large increase in the
(12) scattering pattern, as demonstrated in Fig. 3. The absorption
of the melanin granules can be accounted for in the FDTD
where is the volume fraction of the cells relative to the simulations by assigning each granule a conductivity , al-
whole tissue and is the volume of a single cell. If we though this has very little effect on the scattering patterns for
assume that the cells occupy 70% of the total tissue volume cells with melanin since the pathlength is small. The large
( 0.7), then the predicted scattering coefficient based on the increase in high angle scatter has been observed in in vivo
FDTD computed scattering cross section values in Fig. 5 falls confocal [1] and OCT [25] images of the skin, which measure
between 60 and 105 cm . These values of are similar to single scattering.

Authorized licensed use limited to: University of Texas at Austin. Downloaded on February 24,2025 at 00:14:01 UTC from IEEE Xplore. Restrictions apply.
904 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 2, NO. 4, DECEMBER 1996

The calculated anisotropy factor, , for a cell with organelles [3] D. Boas, M. O’Leary, B. Chance, and A. Yodh, “Scattering of diffuse
(Fig. 5) is higher than the values of typically measured for photon density waves by spherical inhomogeneities within turbid media:
Analytic solution and applications,” in Proc. National Academy Scis.,
tissues which lie in the range of 0.8–0.97 [24], [26] because USA, 1994, vol. 91, pp. 4887–4891.
the simulation calculates the scattering pattern for a single [4] E. Sevick, J. Frisoli, C. Burch, and J. Lakowicz, “Localization of
cell, whereas measurements typically involve tissue sections absorbers on scattering media using frequency domain measurements of
time-dependent photon migration,” Appl. Opt., vol. 33, pp. 3562–3570,
containing multiple cells as well as other tissue structures such 1994.
as collagen and elastin. In addition, the organelles in this work [5] I. Saidi, S. Jacques, and F. Kittel, “Mie and Rayleigh modeling of
were assumed to be homogeneous dielectric objects. Actual visible-light scattering in neonatal skin,” Appl. Opt., vol. 34, pp.
7410–7418, 1995.
organelles are inhomogeneous with spatially varying dielectric [6] K. Yee, “Numerical solutions of initial boundary value problems in-
properties that will cause an increase in the amount of scattered volving Maxwell’s equations in isotropic media,” IEEE Trans. Antennas
light. The FDTD method is able to account for this internal Propagat., vol. AP-14, pp. 302–307, 1966.
[7] D. M. Sullivan, D. T. Borup, and O. P. Gandhi, “Use of the finite-
structure if it is known. Therefore, an understanding of the difference time-domain method in calculating EM absorption in human
3-D dielectric structure of cells is required for more accurate tissues,” IEEE Trans. Biomed. Eng., vol. 34, pp. 148–157, 1987.
[8] A. Taflove and K. Umashankar, “Radar cross section of general three-
calculations. However, few detailed studies of the index of dimensional structures,” IEEE Trans. Electromagn. Compat., vol. 25,
refraction of cells and organelles have been undertaken. pp. 433–440, 1983.
The FDTD method has large computational requirements [9] R. Luebbers and H. Hunsberger, “FDTD for Nth-order dispersive
media,” IEEE Trans. Antennas Propagat., vol. 40, pp. 1297–1301, 1992.
for 3-D problems. The amount of storage and CPU time limits [10] K. Shlager and J. Schneider, “A selective survey of the finite-difference
the method to single cell calculations at the present time. The time-domain literature,” IEEE Antennas Propagat. Mag., vol. 37, pp.
computational requirements of scattering from cells in two 39–56, 1995.
[11] A. Taflove and M. Brodwin, “Numerical solution of steady-state elec-
dimensions using the FDTD technique [27] is considerably tromagnetic scattering problems using the time-dependent Maxwell’s
less than that of three dimensions, and multiple cells can equations,” IEEE Trans. Microwave Theory Tech., vol. MTT-23, pp.
be modeled. However, approximating cells as 2-D structures 623–630, 1975.
[12] J. G. Blaschak and G. A. Kriegsmann, “A comparative study of absorb-
alters the shape of the phase functions. For example, the ing boundary conditions,” J. Computat. Phys., vol. 77, pp. 109–139,
anisotropy value , for a 3-D cell containing only a nucleus 1988.
and cytoplasm was computed to be 0.99, while the two- [13] G. Mur, “Absorbing boundary conditions for the finite-difference ap-
proximation of the time-domain electromagnetic-field equations,” IEEE
dimensional (2-D) phase function of the same cell, computed Trans. Electromagn. Compat., vol. EMC-23, pp. 377–382, 1981.
by modeling a cross section of the 3-D cell, yields a value of [14] Z. P. Liao, H. L. Wong, B. P. Yang, and Y. F. Yuan, “A transmitting
0.96. In a cell containing melanin, the values of were 0.96 boundary for transient wave analysis,” Sci. Sin., Ser. A, vol. 27, pp.
1063–1076, 1984.
and 0.65 for 3-D and 2-D cells, respectively. However, the [15] J.-P. Berenger, “A perfectly matched layer for the absorption of elec-
2-D model [27] did exhibit the same general trends as the 3-D tromagnetic waves,” J. Computat. Phys., vol. 114, pp. 185–200, 1994.
model—small organelles contributed more than the nucleus, [16] K. R. Umashankar and A. Taflove, “A novel method to analyze elec-
tromagnetic scattering of complex objects,” IEEE Trans. Electromagn.
and the anisotropy decreased with organelle volume fraction. Compat., vol. 24, pp. 397–405, 1982.
Therefore, the 2-D approximation can be used to qualitatively [17] R. S. David and L. T. Wille, “Massively parallel finite-difference time-
predict the influence of organelles, but the 3-D model must be domain methods for electromagnetic scattering problems,” in Proc. 10th
Annu. Rev. Progress Appl. Computational Electromagnetics, Monterey,
used to quantitatively compute scattering cross sections and CA, Mar. 1994, vol. 1, pp. 495–502.
anisotropy values of cells. [18] R. Luebbers, S. Kunk, M. Schneider, and F. Hunsberger, “A finite-
difference time-domain near zone to far zone transformation,” IEEE
Trans. Antennas Propagat., vol. 39, pp. 429–433, 1991.
VI. CONCLUSION [19] C. M. Furse and O. P. Gandhi, “Why the DFT is faster than the FFT
for FDTD time-to-frequency domain conversions,” IEEE Microwave
The FDTD method has been applied to light scattering Guided Wave Lett., vol. 5, no. 10, pp. 326–328, 1995.
from cells containing different amounts of organelles. The [20] C. Balanis, Advanced Engineering Electromagnetics. New York: Wi-
FDTD method has advantages over the Mie theory since it ley, 1989.
[21] L. G. Henyey and J. L. Greenstein, “Diffuse radiation in the galaxy,”
accounts for multiple organelles within the cell simultaneously Astrophys. J., vol. 93, pp. 70–83, 1941.
and is relatively easy to implement. The results showed that [22] J. Mourant, J. Boyer, A. Hielscher, and I. Bigio, “Influence of the
scattering phase function on light transport measurements in turbid
small organelles play a significant role in light scattering from media performed with small source-detector separations,” Opt. Lett.,
cells, and the volume fraction of organelles affects both the vol. 21, pp. 546–548, 1996.
total amount of scattered light and the angular distribution [23] A. Ishimaru, Wave Propagation and Scattering in Random Media, vol.
1. New York: Academic, 1978.
of scattered light. The results also demonstrate the need for [24] W. Cheong, S. Prahl, and A. J. Welch, “A review of the optical
more accurate measurements of the index of refraction of cells properties of biological tissue,” IEEE J. Quantum Electron., vol. 26,
and cell components for more accurate determination of the pp. 2166–2185, 1990.
[25] V. Gelikonov, A. Sergeev, G. Gelikonov, and F. Feldchtein, “Charac-
scattering patterns. terization of human skin using optical coherence tomography,” in Proc.
BiOS Europe, Sept. 1996.
REFERENCES [26] S. Jacques, C. Alter, and S. Prahl, “Angular dependence of HeNe laser
light scattering by human dermis,” Lasers Life Sci., vol. 1, pp. 309–333,
[1] M. Rajadhyaksha, M. Grossman, D. Esterwitz, R. Webb, and R. An- 1987.
derson, “ In-vivo confocal scanning laser microscope of human skin: [27] A. Dunn, C. Smithpeter, A. J. Welch, and R. Richards-Kortum, “Light
Melanin provides strong contrast,” J. Invest. Dermatol., vol. 104, pp. scattering from cells,” in OSA Tech. Dig.—Biomedical Optical Spec-
946–952, 1995. troscopy and Diagnostics. Washington, DC: Opt. Soc. Amer., 1996,
[2] D. Huang, E. Swanson, C. Lin, J. Schuman, W. Stinson, W. Chang, pp. 50–52.
M. Hee, T. Flotte, K. Gregory, C. Puliafito, and J. Fujimoto, “Optical [28] J. Maier, S. Walker, S. Fantini, M. Franceschini, and E. Gratton,
coherence tomography,” Science, vol. 254, pp. 1178–81, Nov. 1991. “Possible correlation between blood glucose concentration and the

Authorized licensed use limited to: University of Texas at Austin. Downloaded on February 24,2025 at 00:14:01 UTC from IEEE Xplore. Restrictions apply.
DUNN AND RICHARDS-KORTUM: THREE-DIMENSIONAL COMPUTATON OF LIGHT SCATTERING 905

reduced scattering coefficient of tissues in the near infrared,” Opt. Lett., Rebecca Richards-Kortum received the B.S. degree in physics and math-
vol. 19, pp. 2062–2064, 1994. ematics with highest distinction from the University of Nebraska-Lincoln in
[29] A. Brunsting and P. Mullaney, “Differential light scattering from spher- 1985, the M.S. degree in physics and the Ph.D. degree in medical physics
ical mammalian cells,” Biophys. J., vol. 14, pp. 439–453, 1974. from the Massachusetts Institute of Technology, Cambridge, in 1987 and
[30] H. Liu, B. Beauvoit, M. Kimura, and B. Chance, “Dependence of tissue 1990, respectively.
optical properties on soluute-induced changes in refractive index and In 1990, she joined the faculty as an Assistant Professor at the University
osmolarity,” J. Biomed. Opt., vol. 1, pp. 200–211, 1996. of Texas at Austin, where she has established collaborative research projects
[31] I. Vitkin, J. Woolsey, B. Wilson, and R. Anderson, “Optical and ther- which focus on the application of optical spectroscopy for detection of
mal characterization of natural (sepia oficinalis) melanin,” Photochem. precancer and atherosclerosis.
Photobiol., vol. 59, pp. 455–462, 1994.

Andrew Dunn received the B.S. degree in physics from Bates College in
Lewiston, ME, in 1992 and the M.S. degree in electrical engineering from
Northeastern University, Boston, MA, in 1994. Currently he is pursuing the
Ph.D. degree in biomedical engineering at the University of Texas at Austin.
His doctoral research involves developing an understanding of the physics
of light scattering in tissue through numerical simulation and experimental
measurements.

Authorized licensed use limited to: University of Texas at Austin. Downloaded on February 24,2025 at 00:14:01 UTC from IEEE Xplore. Restrictions apply.