This is a brief summary of the major points from a

tutorial lecture by Steven L. Jacques at the

Graduate summer school:
Bio-Photonics '03
15-21 June, Ven, Sweden
9.00-12.20 Session 1 - Optical properties of tissue

Steven Jacques, Oregon Health & Science University, USA

http://omlc.ogi.edu
http://www.bme.ogi.edu/biomedicaloptics

9.00-9.40 Tissue optics - part I

Introduction to optical properties

9.50-10.30 Tissue optics - part II

Introduction to light transport

Coffee (bring back to lecture room)

10.50-11.30 Tissue optics - part III

Measurements of optical properties

11.40-12.20 Tissue optics - part IV

Optical fiber reflectance spectroscopy
The definition and units of the absorption coefficient µa [cm-1] are
illustrated in the following pages.

The absorption of tissue is dominated by protein, DNA, and other

“dry” material in the ultraviolet, by water in the infrared, and by
hemoglobin and melanin in the visible.
absorption cross-sectional area
efficiency
geometrical area

σa = Qa A
[cm2] [-] [cm2]

A σa = Q aA
geometrical effective
cross-section cross-section
 absorption coefficient
density
cross-sectional area

µa = ρa σa efficiency
area
[cm-1] [cm-3][cm2]
σa = Q a A
[cm2] [-][cm2]

T = exp(-µaL)
]
-1

UV Vis IR
10,000
Absorption coefficient [cm
ArF Er:YAG
excimer whole blood
1,000 KrF melanosome
excimer
100 epidermis CO
2
XeCl
excimer
10 75% water
aorta Ho:YAG
dye laser
1 argon
skin Nd:YAG

100 1,000 10,000

Wavelength [nm]
The definition and units of the scattering coefficient, µs [cm-1], are
similar to those of the absorption coefficient, and are illustrated in
the following pages.
The anisotropy, g [dimensionless], describes the efficiency of
scattering, and its definition is described as follows. (The Henyey-
Greenstein function approximates single photon scattering by tissue.)

The “reduced scattering coefficient”, µs´ [cm-1] = µs(1-g), is a

lumped parameter that characterizes photon diffusion as a random
walk with step size 1/µs´.
The scattering properties of tissue are dominated by the lipid-
water interfaces presented by membranes in cells (especially
membranes in mitochondria), by nuclei, and by protein fibers such
as collagen or actin-myosin.
 scattering cross-section
efficiency
cross-sectional area

σs = Qs A
[cm2] [-] [cm2]

A σs = Q sA
geometrical effective
cross-section cross-section
 Anisotropy, g

š
g = <cosθ> = ∫ p(θ)cosθ 2š sinθ dθ
0
š
where ∫ p(θ) 2š sinθ dθ = 1
0
Henyey-Greenstein function
26°
equivalent Random Walk
The small-scale structure of tissue components, such as the
membranes of mitochondria or the water-protein periodicity of
collagen fibrils, are much smaller than the wavelength of light
(λ). Such small structure gives rise to so-called “Rayleigh limit”
scattering which presents isotropic scattering (scatters in all
directions) and scales with wavelength as 1/λ4 (stronger
scattering at short wavelengths).
The large-scale structures such as a whole mitochondrion or a
collagen fiber bundle are much bigger than the wavelength of
light (λ) and give rise to “Mie scattering” which is anisotropic
scattering (forward directed) and scales roughly as 1/ λb where b
is typically somewhere in the 0.5-1.5 range.
 Reduced scattering coefficient
of dermis

Reduced Scattering, µs(1-g)

100 Mie
+
Rayleigh
[cm -1 ]

10
Mie
Rayleigh
3
0 500 1000 2000
Wavelength (nm)
mitochondria

0.5 µm
Collagen fibrils, fibers
Monte Carlo simulation is computational method for
calculating the movement of photons within a scattering
tissue.
Launching millions of photons and observing where the
photons propagate enables one to map the fluence rate
distribution of photons within a tissue.
The following pages illustrate the above.
Monte
MonteCarlo
Carlosimulation
simulation
of
ofphoton
photonmigration
migration

air
tissue

5mmx5mm

esophagus @ 630 nm wavelength

 1 = 1 mm
µ s (1-g) Monte Carlo

W
cm2
z [mm] per W
incident
power

r [mm]
 1 = 1 mm
µ s (1-g)

W
cm2
per W
z [mm] incident
power

r [mm]
Launching collimated
photons into a optical fiber
scattering medium in scattering
creates an apparent
center for diffusion that medium
is 1/µs’ in front of the
launch point.
1
µs(1-g)
An isotropic collector consists of a small ball of scattering
material (eg., Teflon) into which is inserted an optical fiber.
Light from any direction that strikes the ball has an equal chance
of entering the fiber and being detected.
An experiment with a fish tank illustrates how adding milk to
clear water causes the light concentration to pile up near the front
of the tank that is illuminated with a broad collimated laser
beam.
 2
Fluence rate F [W/cm ]

mirror
The scattering in tissue
redirects light toward an
object, just as the mirrors
shown here.
Hence, the fluence rate F
E [W/cm2] can exceed the
delivered irradiance E
[W/cm2].

mirror
Estd

Pdetected.std
F

P
+ milk/ink

Clear water
add water to match
front boundary

+matching
A simple expression approximates the penetration of light into a
scattering tissue illuminated with a broad collimated beam. This is
a 1-dimensional solution, only a function of depth z.
The solution is not correct near the surface, but is correct within
the tissue. It is useful for predicting how deeply light will
penetrate into tissue.
Simple 1-D approximation that applies
to a broad flat-field beam of
irradiance (Eo [W/cm2])
The backscatter term k [dimensionless]

k
To a first approximation, the total diffuse reflectance Rd
behaves as if there were a single pathlength L = 7δ for
photons, where δ is 1/e the optical penetration depth.
Fluence rate (F [W/cm2] is a parameter that is proportional to the
concentration of photons (C [J/cm3]): F = cC where c is the
speed of light [cm/s].
If a cube of space is irradiated with 1 W/cm2 fluence rate (at 488
nm wavelength), the concentration of photons in that cube is
1.1x1010 photons/cm3, or 1.8x10-13 moles/liter.
Time-resolved diffusion of photons simply describes the
diffusion of photon concentration (or photon energy
concentration) as a function of distance (r) from a point source
and time (t) after the impulse of optical energy deposition. The
diffusivity of photons is
χ = cD [cm2/s]
where D = 1/(3(µa + µs’)) [cm].
 Fluence rate = speed of light x photon concentration
F=cC
(c [cm/s])(44
[cm / s])(44 ps)
ps)
2
11W
W / cm cm-2
1 cm

1 cm

1 cm
En λ
Cph = = 1.1×1010 photons / cm 3
c hc
En λ 1000
= = 1.8×10−13 moles / liter
c hc N av
for λ = 488 nm, n = 1.33
fluence rate concentration
F =
F = ccC
C

[W cm-2] [J cm-3]

[cm/s]
speed of light
 time-resolved diffusion
diffusivity, χ = cD [cm2/s]

r2
2
exp−r
(- /(4cDt ) )
e 4cDt
C(r,t),t)== UU
C(r oo
πcDt)3/2
(4(4πcDt) 3/ 2

[J cm-3] [J] [cm-3]

Fick’s 1st Law of diffusion
∂C
J = −χ
∂x
Diffusivity
[cm2/s]
χ = cD
1 1
D=
3 µa + µs
'
Fick’s 2nd Law of diffusion
∂C ∂ C 2
=χ x
∂t ∂x
 time-resolved fluence rate

2
−r /(4cDt )
e exp (-
r2
) −µ a ct
F(r ,t) = cU
F(r,t) = c Uo o
4cDt e
3 /exp(-µ act)
(4 πcDt)(4πcDt)3/2
2

[W cm-2] [cm/s][J] [cm-3] absorption

C(r,t)
Fluence rate F [W/cm2] times time t [s] equals the radiant
exposure H [J/cm2].
Therefore, integration of F over time yields the time-
independent overall exposure H.
Note that the diffusion length D [cm] equals µaδ2.

The H in response to a sequence of impulses of value Uo and

frequency f, which present an average power P = fUo, yields
the time-independent steady-state expression for fluence rate
in response to a steady-state point-source of power.
 radiant exposure
∞

∫
∞
H(r) = F(r,t) dt
H(r) =
0
∫ c C(r,t) exp(-µ ct) dt
0 a

[J cm-2] [W cm-2] [s]

−r /δ −r /δ
e
exp(- r) e
H(r) == UUo =U
µa/D
4 πµ4πDr
δ 4 πDr
2 o
o
a r
 frequency

[W] [J] [s-1]

in the limit
P=U f
Po = Uo f
o
→∞∞
ff ∅
Uoo ∅
→00
steady-state fluence rate

−r /δ
e
F(r ) = P exp(- µa/D r)
F(r) = Po o 4 πDr
4πDr

[W cm-2] [W] [cm-2]

Memorize this equation !!!

Use it.
To measure optical properties:
Any two measurements that are sufficiently different that they
behave differently as a function of tissue optical properties will
allow two optical properties to be specified.
For example, measurements of light by two fibers at different
distances (r1, r2) from a source fiber will yield two
measurements, T1 and T2, that map uniquely into a grid of two
optical properties, µa and µs’, thereby specifying µa and µs’.
If the distances r1, r2 become two similar, then the grid collapses
into a single line and the measurements can no longer
distinguish µa and µs’.
 Measurement vs properties grid

1.5

1 Collection fiber
Collection
fibers

z [cm]
0.5 r2 = 1.0 cm

0 Source fiber Collection fiber

Source fiber r1 = 0.3 cm

-0.5
-1 -0.5 0 0.5 1
x [cm]
black x = µs’ < 10µa blue = const µ s', red = const µ a
1
10
r1 = 0.1 cm 100
r = 1.0 cm
2
µs '
10

0
10 1000
T2 [1/cm 2]

1.0
0.01
-1
µa 0.1
10

1.0

-2
10 0 2 4
10 10 10
2
T1 [1/cm ]
black x = µs’ < 10µa blue = const µ s', red = const µ a
1
10
r1 = 0.3 cm 100
r = 1.0 cm
2
µs '
10

0
10 1000
T2 [1/cm 2]

1.0
0.01
-1
µa 0.1
10

1.0

-2
10 0 2 4
10 10 10
2
T1 [1/cm ]
black x = µs’ < 10µa blue = const µ s', red = const µ a
1
10
r1 = 0.6 cm 100
r = 1.0 cm
2
µs '
10

0
10 1000
T2 [1/cm 2]

1.0
0.01
µ-1 a0.1
10

1.0

-2
10 0 2 4
10 10 10
2
T1 [1/cm ]
 blue = const µ s', red = const µ a
1
10
r1 = 0.9 cm 100
r = 1.0 cmµ '
2 s
10

0
10 1000
T2 [1/cm 2]

1.0
0.01
µa0.1
-1
10

1.0

-2
10 0 2 4
10 10 10
2
T1 [1/cm ]
To calibrate a particular device that measures photon transport
from source to detector optical fibers, a matrix of phantoms
(gels with added absorber and scatterer) is created and
measurements taken.
For example full spectral measurements (200 wavelenths) of
white light transmission from a set of 6 source fibers
surrounding a single collector fiber were made on an 8x8
matrix, yielding 8x8x200 = 12800 data values.
These data values were analyzed to yield a transport function
M = getM(µa, µs’)
where M = Mgel/Mstd and Mstd is a measurement on a known
standard. While similar to transport theory, this function
accounts for the details of the device geometry.
 optical fiber probes

µa

Mtissue

µs’

gel

Array of gels Mstd

Fixed
Diffuse 99+% reflectance standard distance
 M tissue S Ttissue ηc D Ttissue ηc
M= = =
M std S Tstd Gstd D Tstd Gstd

M( µ a ,µ s' ) = M o e −µ a L
'2
M o = a0 + a1µ + a2 µ + ... ≈ a0 + a1µ s'
'
s s
'2
L = b0 + b1µ + b2 µ + ... ≈ b0 + b1µ
'
s s
'
s

Central collection fiber

~200 µm
6 surrounding source fibers
spacing
 function M = getM(mua, musp)
5
M 1.8

4.5
1.6
4
1.4
a = 0.0036;
3.5

3 1.2
b = -0.1576;
µ
a 2.5
1
c = 0.1136;
2
0.8
d = 0.6927;
1.5

1 0.6

0.5
0.4
L = -(a*musp + b);
2 4
µ'
s
6 8 10 Mo = c*musp + d;

M = Mo.*exp(-mua.*L);
After calibration, any measurement of skin or other tissue is
normalized by the measurement of the standard.
A least squares fitting routine then iteratively chooses values of µa
and µs’ and uses getM(µa, µs’) to generate predicted M values for
comparison with the experimental value. This process specifies µa
and µs’.
In our lab, we usually fit for values that specify the absorption and
reduced scattering coefficient over a spectral range:
µadry = absorption due to background material (proteins, bilirubin,
distributed porphyrin absorption, …)

fwater = volume fraction of water in tissue (~0.85)

fblood = volume fraction of blood in tissue (~0.005-0.15)
S = oxygen saturation of mixed arterio-venous blood
a, b where µs’ = aλ-b , the scattering coefficient
 Absorption coefficient:

µa = µadry + f waterµawater + fblood (Sµaoxy + (1− S )µadeoxy )

4
Library
10
oxy
deoxy
dry
2
10 water

0
10

-2
10

-4
10
500 600 700 800 900
Wavelength (nm)

Scattering coefficient:

µs '= aλ−b + cλ−d + ...+ wλ−4 ≈ aλ−b

 Circles = experimental data
Solid line = fit
Dashed black line = bloodless fit

Normal esophagus
2
a
epoxy bloodless
upper view

1.5
in vivo
exit hole stainless steel tubing teflon tubing
1
600 µm optical fiber silicone

side view
45o polished stainless steel rod
a = 9941
Endoscopic two-fiber probe 0.5 b = -1.115
fv = 0.0056
S = 0.511
const = 1.118
0
500 550 600 650 700 750 800 850
Wavelength [nm]

Note: above fv = fb
 Circles = experimental data
Solid line = fit
Dashed black line = bloodless fit

Esophageal tumor
2
b
upper view epoxy

1.5 bloodless

exit hole stainless steel tubing teflon tubing

1
600 µm optical fiber silicone

side view
45o polished stainless steel rod
a = 17145
Endoscopic two-fiber probe 0.5 in vivo b = -1.297
fv = 0.0277
S = 0.276
const = 1.811
0
500 550 600 650 700 750 800 850
Wavelength [nm]

Note: above fv = fb
 Circles = experimental data
Solid line = fit
Dashed black line = bloodless fit

Esophageal tumor
2
c
upper view epoxy

1.5 bloodless
100% oxy
exit hole stainless steel tubing teflon tubing
1 in vivo
600 µm optical fiber silicone

side view
45o polished stainless steel rod
a = 25833
Endoscopic two-fiber probe 0.5 b = -1.297
100% deoxy fv = 0.1482
S = 0.567
const = 1.314
0
500 550 600 650 700 750 800 850
Wavelength [nm]

Note: above fv = fb
In summary,
light transport through a tissue can be modeled by
diffusion theory or Monte Carlo simulations to indicate the
relationship between transport and tissue optical properties.
Two such measurements can be interpreted to yield values of
absorption and reduced scattering coefficients. (3 measurements can
yield µa, µs, and g.)

Practical calibration of specific light transport

measurement devices uses a grid of phantoms with varying
amounts of added absorber and scatterer to yield an empirical
description of light transport: M = getM(µa, µs’).
Measurements are interpreted in terms of the blood
content, oxygen saturation of the mixed arterio-venous blood,
the tissue water content, the tissue scattering properties, and
any additional absorbers (eg., bilirubin or drugs).