You are on page 1of 11



RANGE 4002500 NM
Andre Roggan, Moritz Friebel, Klaus Dorschel, Andreas Hahn,* and Gerhard Muller,

Institut fur Medizinische/Technische Physik und Lasermedizin, Universitatsklinikum Benjamin

Franklin, Freie Universitat Berlin, Krahmerstr. 6-10, D-12207 Berlin, Germany; Laser- und Medizin-
Technologie gGmbH, Berlin, Germany; *Stockert Instrumente GmbH, Munchen, Germany
(Paper ODB-007 received Dec. 29, 1997; manuscript accepted for publication Nov. 6, 1998.)

Knowledge about the optical properties m a , m s , and g of human blood plays an important role for many
diagnostic and therapeutic applications in laser medicine and medical diagnostics. They strongly depend on
physiological parameters such as oxygen saturation, osmolarity, flow conditions, haematocrit, etc. The inte-
grating sphere technique and inverse Monte Carlo simulations were applied to measure m a , m s , and g of
circulating human blood. At 633 nm the optical properties of human blood with a haematocrit of 10% and an
oxygen saturation of 98% were found to be 0.21060.002 mm1 for m a , 77.360.5 mm1 for m s , and 0.994
60.001 for the g factor. An increase of the haematocrit up to 50% lead to a linear increase of absorption and
reduced scattering. Variations in osmolarity and wall shear rate led to changes of all three parameters while
variations in the oxygen saturation only led to a significant change of the absorption coefficient. A spectrum
of all three parameters was measured in the wavelength range 4002500 nm for oxygenated and deoxygen-
ated blood, showing that blood absorption followed the absorption behavior of haemoglobin and water. The
scattering coefficient decreased for wavelengths above 500 nm with approximately l 21.7; the g factor was
higher than 0.9 over the whole wavelength range. 1999 Society of Photo-Optical Instrumentation Engineers.

Keywords blood; optical properties; extracorporal circulation; absorption; scattering; phase function.

1 INTRODUCTION oxygen transport. The cell concentration in blood

under physiological conditions is approximately
As given by transport theory, the optical param-
531012 L1.
eters are defined by the absorption coefficient m a in
However, only few theoretical and experimental
mm1, the scattering coefficient m s in mm1 and the
studies about the optical properties of blood have
mean cosine of the scattering angle g (anisotropy
led to reasonable results because the optical prop-
factor) in combination with a scattering phase func-
erties of blood mainly depend on physiological and
tion p(s,s8 ). 1,2 Knowledge about the optical param-
biochemical parameters such as haematocrit, flow,
eters of blood plays an important role for many di-
osmolarity, haemolysis, and oxygen saturation.
agnostic and therapeutic applications in laser
Haematocrit (hct) is the volume fraction of cells
medicine and medical routine diagnosis. To calcu-
late the light distribution in blood perfused tissues, within the whole blood volume and ranges from
knowledge about the optical properties of blood is 36.8% to 49.2% under physiological conditions. The
required for a number of optical methods, such as haemoglobin concentration ranges from 134 to 173
optical tomography, optical biopsy, diaphanoscopy, g/L for whole blood and from 299 to 357 g/L for
photodynamic therapy, laser induced thermo- RBCs. Flow induced shear stress can influence phe-
therapy, or portwine stains and haemangioma nomena such as sedimentation, reversible agglom-
treatment. Normal human whole blood consists of eration, axial migration or cell deformation, and
about 55 vol % plasma (90% water, 10% proteins) orientation. The flow parameters depend on the
and 45 vol % cells (99% red blood cells erythro- blood viscosity and they are influenced by the fact
cytes, 1% leukocytes and thrombocytes). A red that blood is not a Newton fluid. A change in os-
blood cell (RBC) has a characteristic flat biconcave molarity induces a variation of the RBC volume
form with a diameter of 7 to 8 mm and a thickness due to water exchange and therefore has an impact
of 2 mm (Figure 1). The RBC has a mean volume of on the haemoglobin concentration within the RBC.
90 mm3 and contains 30 pg haemoglobin that allows Haemolysis is coupled with the damage of RBC
membranes by mechanical or chemical impact with
Address all correspondence to Andre Roggan, PhD. Tel: 49-30-8449-2327;
Fax: 49-30-8449-2399; E-mail: 1083-3668/99/$10.00 1999 SPIE


JANUARY 1999 d
VOL. 4 NO. 1
Downloaded from SPIE Digital Library on 12 Jun 2010 to Terms of Use:

Fig. 1 Human erythrocyte.

consecutive release of haemoglobin into the plasma.

Finally, oxygenation of haemoglobin leads to char-
acteristic changes in its absorbing behavior. This ef-
fect is used in clinically available blood oxymeters.
All mentioned parameters have to be precisely
controlled during optical measurements and the Fig. 2 Turbulence free flow cuvette with a thickness of 97 mm.
blood has to be kept moving to avoid sedimenta-
tion and clustering. A physiological flow is corre-
lated with a shear rate of approximately 500 s1 in exception of experiments based on changing oxy-
the capillary system. Therefore an experimental gen states, all investigations were derived with an
setup was realized to provide measurements of the oxygen saturation over 98%. The blood temperature
optical blood properties in the wavelength range was kept constant at 20 C.
4002500 nm under flow conditions using a spe-
cially designed flow cuvette in combination with an 2.2 EXPERIMENTAL SETUP
extracorporal circulation and oxygenation unit. The optical properties of scattering and absorbing
media cannot be measured directly. As a result, the
2 MATERIAL AND METHODS optical blood properties were determined by a
double integrating sphere technique, evaluating the
2.1 BLOOD PREPARATION diffuse backscattering R d , the total transmission
Fresh erythrocyte concentrates from human donors T t , and the collimated, i.e., nonscattered, transmis-
were centrifuged three times and washed in phos- sion T c of thin samples (Figure 3).36 The inner sur-
phate buffer solution (300 mosmol/L, pH 7.4). By faces of the integrating spheres (B152 mm) were
doing this the blood samples contained no plasma coated with a highly reflecting layer of Spectralon ,
proteins, leukocytes, or thrombocytes. Light mi- allowing the total radiation in the hemispheres in
croscopy control of the samples ensured that they front and behind the sample to be collected. Reflec-
were neither haemolysed nor were the cells de- tion standards were used to calibrate the system
formed. The haematocrit (hct) was used as a mea- (Labsphere). All measurements with respect to
sure for the concentration of red blood cells and
was adjusted by dilution with phosphate buffer.
Blood samples were investigated with a haemat-
ocrit value between 2.5% and 70%. Dilution with
buffers of different osmolarities allowed the adjust-
ment of various blood osmolarities from 225 to 450
mosmol/L. In addition, a quantitative haemolysis
was induced by diluting one part of the blood
sample with distilled water, readjusting the osmo-
larity to 300 mosmol/L by adding buffer solution,
and mixing the haemolyzed sample with the intact
blood sample. Blood circulation and predetermined
oxygen states were adjusted with an extracorporal
circulation unit (Stockert Instrument GmbH, Ger-
many). The blood was gently stirred and kept flow-
ing through a specially designed turbulence free cu-
vette with a laminar flow and an optical path of 97
mm (Figure 2). Oxygen saturation was adjusted by
continuous flow of a mixture of O2, N2, and CO2
through a modified membrane oxygenator. The
oxygen saturation was continuously controlled by a
blood oxymeter (OxySAT-Meter SM 0200; Baxter).
For complete deoxygenation the erythrocyte con-
centrates were diluted with phosphate buffer con- Fig. 3 Experimental setup for the determination of optical proper-
taining 0.3% sodium dithionite (Na2S2O4). With the ties of human blood (Refs. 46).


JANUARY 1999 d
VOL. 4 NO. 1 37
Downloaded from SPIE Digital Library on 12 Jun 2010 to Terms of Use:

varying physiological blood parameters were car-

ried out at a fixed wavelength of 632.8 nm, using a
heliumneon laser with 1.2 mW output power
(Spindler & Hoyer). The spot size in the sample
plane was B7 mm; the distance between sample
and T c detector was 1330 mm; The aperture in front
of the T c detector was B5 mm, the beam diameter
in the aperture plane was B1.2 mm. A mercury
high pressure short arc lamp served as light source
for the spectral overview of optical properties (Os-
ram, 100 W, 23106 cd/cm2). The lamp was con-
nected to a monochromator (AMKO, f5200 mm)
with a resolution of 8 nm [full width at half maxi-
mum (FWHM)] in the range 4001140 nm and 16
nm (FWHM) in the range 11402500 nm. The circu-
lar monochromator exit slit (B1 mm) was focused
on the flow cuvette, providing a spot size of B4
mm. The distance between sample and T c detector
was 430 mm, the aperture in front of the T c detector
was B30 mm, and the beam diameter in the aper-
ture plane was B25 mm. The beam was mechani-
cally chopped at 220 s1 and part of the beam was
used for the reference intensity compensation. Sili-
con photodiodes (AMKO, 09-SiU04-C, 4001140
nm) and lead sulphide photodiodes (AMKO, 09-
Pb01-C, 11402500 nm) were used as detectors. The Fig. 4 Scheme of the inverse Monte Carlo simulation for the cal-
signals were finally recorded with a lock-in tech- culation of optical properties from measured data (Refs. 46).
nique (ITHACO 3981).

detected at the T c detector, deviations from a Lam-

bertian distribution of the diffuse reflectance, and
The extraction of the optical parameters m a , m s , an incomplete separation of the radiation fields in
and g from the measurements is a complex task as the integrating spheres. The potential disadvantage
there are no analytical models available with suffi- of long calculation times was compensated for by
cient precision. The method of choice was therefore the use of currently available fast computer sys-
the Monte Carlo simulation, which as a statistical tems.
method calculates the trajectories of a great number One of the most important questions when deal-
of photons (33105 in our experiments) and as a ing with Monte Carlo simulations of photon trans-
result presents the remission and transmission port in turbid media is the choice of a scattering
characteristics of a sample for a given set of optical phase function that fits best to the investigated me-
parameters. In order to solve the opposite situation, dium. Therefore the HenyeyGreenstein phase
i.e., to determine the unknown optical parameters function was most often used in biomedical optics
from the measured macroscopic values, the Monte because of its good correspondence to goniophoto-
Carlo simulation had to be inverted46 (Figure 4). metric measurements of tissues like skin or paren-
For this the measured data were simulated taking chymatous organs:817
an estimated set of start parameters m a , m s , and g
from KubelkaMunk theory.7 Then the measured 12g 2HG
quantities were calculated and compared with the p HG~ s,s8 ! 5 .
4 p ~ 11g 2HG22g HG cos Q ! 3/2
real measurements. In case of a significant devia-
tion all three parameters were varied slightly and Here p HG(s,s8 ) is the normalized probability for a
three new forward simulations were performed. photon to be scattered from a direction s into a di-
After this procedure a gradient matrix was built up, rection s8 , and g HG is the free parameter which
allowing the calculation of new optical parameters equals the mean cosine of the scattering angle,
which fit better to the measured quantities. This known as the anisotropy factor g. On the other
procedure was repeated until the deviation be- hand, it is known that the HenyeyGreenstein
tween measured and calculated values was within phase function does not fit to the scattering of RBCs
the error threshold (fixed to 0.1%). Thus the associ- because of their large diameter. Hence Yaroslavsky
ated set of optical parameters could be accepted. As proposed the GegenbauerKernel phase function
well as a high degree of accuracy, the method of- for blood samples.18 This is a two parameter func-
fered a simple means of compensating systematic tion with a GK.0 and 21<g GK<1. The
errors such as radiation losses, scattered photons GegenbauerKernel phase function equals the


JANUARY 1999 d
VOL. 4 NO. 1
Downloaded from SPIE Digital Library on 12 Jun 2010 to Terms of Use:

Fig. 5 Comparison of various phase functions for red blood cells.

HenyeyGreenstein phase function for a GK50.5. tions while the factor g GK was the free fit param-
The anisotropy factor g has to be calculated numeri- eter. The anisotropy factor g was finally calculated
cally: numerically from a GK and g GK .


p GK~ s,s8 ! 5 2 a GK 2 a GK
p ~~ 11g GK! 2 ~ 12g GK! ! For each experimental question a total of three in-
dependent measurement series were carried out us-
~ 12g 2GK! 2 a GK
3 . ing erythrocyte concentrates from different human
~ 11g 2GK22g GK cos Q ! a GK11 donors. One measurement series at a fixed wave-
length of 633 nm with varying haematocrit, shear
The importance of a correct phase function becomes
rate, osmolarity, haemolysis, or oxygenation re-
evident with regard to photons which travel
quired approximately 60 min. The evaluation of the
through the measuring cuvette and which are scat-
optical properties applying inverse Monte Carlo
tered only once or twice. These photons have a high
simulations required a total time of 1020 h for all
probability to be collected in the T c detector be-
three series of one experiment. Mean values and
cause of the extreme forward scattering of RBCs.
standard deviations were calculated as shown in
Thus these photons had to be considered as a T c
the subsequent graphs. The spectral overview re-
offset in the Monte Carlo routine, but the amount of quired approximately 5 h for each spectrum which
this offset strongly depended on the selected phase was also measured three times in order to calculate
function. This is shown in Figure 5 where three mean values of optical properties. Before and after
phase functions with the same anisotropy factor of each measurement the blood samples were checked
g50.9924 were compared. Mie calculations of the for their haematocrit as well as for possible
RBC phase function at 633 nm served as a reference haemolysis and cell deformation.
where a mean diameter of 5.56 mm was selected as
a spherical RBC equivalent with a volume of 90
mm3. The refractive index was set to n51.333 for 3 RESULTS AND DISCUSSION
the surrounding medium and to n51.402 for the 3.1 HAEMATOCRIT
RBC. That the Mie calculation for a spherical
equivalent is a valid assumption for RBCs could be Isotonic blood samples (300 mosmol/L) of various
shown in the subsequent measurements at various haematocrits were measured at a wavelength of 633
plasma osmolarities (see Secs. 3.3 and 3.4 for de- nm. Figure 6 shows the mean values (n53) of m a ,
tailed information). Evidently the Henyey m s , g, and m s8 depending on the haematocrit rang-
Greenstein phase function overestimates the RBC ing from 0% to 70% at a constant shear rate of 500
phase function by a factor of approximately 30 s1. The absorption coefficient m a increased linearly
within the first degrees of the scattering angle that with the haematocrit over the whole measured hae-
are critical for the T c detection. On the other hand, matocrit range. The scattering coefficient m s also in-
the GegenbauerKernel phase function with a GK creased proportionally to the haematocrit but only
51.0 mimics the Mie function without significant for values of hct<10%. The anisotropy factor g was
deviations in the critical range. Therefore the almost constant in that haematocrit range (0.994
GegenbauerKernel phase function was imple- .g.0.992). On the other hand, m s appeared to be
mented in the Monte Carlo model and a fixed factor independent of blood concentration at values hct
of a GK51 was applied for all subsequent evalua- .10% and the g factor decreased continuously


JANUARY 1999 d
VOL. 4 NO. 1 39
Downloaded from SPIE Digital Library on 12 Jun 2010 to Terms of Use:

Fig. 6 Mean values 6 SD ( n 53) of m a , m s , g, and m s8 vs haematocrit (p 5300 mosmol/L, g 5500 s1, SatO2 .98%, l5633 nm).

down to 0.975 at hct 70%. One could conclude from flow on m a and m s8 (Figure 7). The underlying
investigation of the measured quantities that the mechanism for cell deformation and axial migration
collimated transmission dropped below the detec- was the predominating shear force within the flow
tion threshold for hct.10%. Therefore the calcula- cuvette. Therefore the shear rate was calculated
tion of three independent optical properties was not from the ratio of flow velocity and cuvette cross
possible at high hct levels and consequently the re- section. The range of the adjusted shear rate corre-
duced scattering coefficient m s8 5 m s (12g) was sponded to the physiological range in the human
evaluated. In contrast to m s a linear increase was vascular system. A low shear rate of 50 s1 led to a
found for m s8 up to hct 45%. Above 45% a slight decrease of m a to 95% compared to a short flow
saturation effect was found which might be the re- stop. Additional increase of the shear rate resulted
sult of either interfering scattering events or the for-
in a continuous but smaller decrease of m a down to
mation of RBC clusters. This nonlinear scattering
dependence on haematocrit was already predicted 88% at g 53000 s1. The reduced scattering coeffi-
by Reynolds from theoretical investigations.19 Due cient m s8 showed a similar behavior with increasing
to the incomplete separation of m s and g only m s8 shear rates. m s8 decreased to 93% at g 550 s1 and to
was calculated for blood samples at haematocrits 80% at g 53000 s1. It could be clearly shown that
above 10%. the flow velocity and the related shear rates had a
significant impact on the optical properties of
3.2 FLOW VELOCITY blood. However, from the measured data it was not
The optical properties were measured under vari- possible to differentiate if the shear rate depen-
ous flow conditions to derive the effects of blood dence was a result of cell deformation or axial mi-

Fig. 7 Mean values 6 SD ( n 53) of m a and m s8 vs shear rate (hct541%, p 5300 mosmol/L, g 5500 s1, SatO2.98%, l5633 nm).


JANUARY 1999 d
VOL. 4 NO. 1
Downloaded from SPIE Digital Library on 12 Jun 2010 to Terms of Use:

Fig. 8 Mean values 6 SD ( n 53) of m a , m s , m s8 , and g at short flow stop vs plasma osmolarity (hct57.5% at 300 mosmol/L, g
50 s1, SatO2.98%, l5633 nm).

gration. Thus, further measurements with different from 0.995 at 225 mosmol/L down to 0.991 at 450
geometrical shapes of RBC were carried out by mosmol/L. The reduced scattering coefficient m s8
varying the osmolarity of the buffer. increased linearly with osmolarity. In contrast to
the absorption properties, the scattering behavior
3.3 OSMOLARITY appeared to be reasonable because the cell shape
Variation in the plasma osmolarity led to consider- and the refractive index of the cell bounded haemo-
able changes in the shape of the erythrocytes. Hy- globin solution varied with osmolarity, both
perosmotic plasma induces cell shrinking (acantho- strongly affecting the scattering performance.
cytes); in hypo-osmotic plasma the cells swell Mie calculations on red blood cells were carried
(spherocytes). In both cases deformability of the out applying the algorithms of Zijp and ten Bosch20
erythrocyte is distinctly diminished. Diluted blood to differentiate between the impact of cell shape
samples (hct57.5% at 300 mosmol/L) were ad- and refractive index. Therefore the cell shape was
justed to osmolarity values ranging from 225 to 450 assumed to be spherical when applying Mie theory
mosmol/L using phosphate buffer solutions of and the sphere diameter was adjusted to mimic the
various osmolarity. Because the number of cells per RBC volume coupled to a certain osmolarity. The
volume was kept constant for these experiments, refractive index of the cell bounded haemoglobin
the haematocrit varied slightly with osmolarity due solution was derived by the equation n solution
to the different cell volumes. Figure 8 shows the 5n water1 b c, where c is the haemoglobin concen-
measured optical properties at a wavelength of 633 tration in g/100 ml and b 50.001 942 at a wave-
nm. length of 589 nm.21 Assuming a mean erythrocyte
The absorption coefficient m a increased continu- volume of 90 mm3 and an inner cell haemoglobin
ously with osmolarity, i.e., cell shrinkage led to an concentration of 350 g/L (35 g/100 ml) for isotonic
increase of m a . At 225 mosmol/L m a amounted to conditions, a refractive index of n 30051.402 and a
60% of the value measured under isotonic condi- sphere equivalent diameter of d 30055.56 m m were
tions while m a was 50% above the isotonic value at calculated. Considering slight variations of the hae-
450 mosmol/L. This effect was unexpected because matocrit at 250 mosmol/L (hct 8.1%, RBC volume
the number of cells, and therefore the overall con- 96.7 mm3, RBC haemoglobin concentration 325 g/L)
centration of absorbing haemoglobin, was kept con- and 400 mosmol/L (hct 6.6%, RBC volume 78.6
stant during all measurements with different osmo- mm3, RBC haemoglobin concentration 400 g/L) the
larities. An explanation will be given in the corresponding refractive indices were n 25051.397
discussion of the measurements on haemolysed and n 40051.412 and the sphere equivalent diam-
blood samples. The scattering coefficient m s showed eters were d 25055.70 m m and d 40055.32 m m. The
a slight decrease with increasing osmolarity of scattering coefficients were calculated by multiply-
about 10%. Also the anisotropy factor g decreased ing the Mie scattering cross section with the num-


JANUARY 1999 d
VOL. 4 NO. 1 41
Downloaded from SPIE Digital Library on 12 Jun 2010 to Terms of Use:

Fig. 9 Measured mean values 6 SD ( n 53) of m s and g compared with Mie calculations for spherical equivalents (hct57.5% at 300
mosmol/L, g 50 s1, SatO2 .98%, l5633 nm).

ber density of 8.3331011 L1 which was related to dynamic decrease of haematocrit as a result of axial
hct57.5% at 300 mosmol/L and which was kept migration (Fahraeus effect), while shear stress in-
constant at all osmolarities. duced cell deformation could be neglected with re-
Figure 9 shows the results of the Mie calculations spect to the optical properties.
for m s and g compared with the measured data. At
250 mosmol/L m s were calculated to be 63.8 mm1 3.4 HAEMOLYSIS
vs 62.861.5 mm1 for the measured value, at 300 Blood samples of different haemolytic states were
mosmol/L 62.2 mm1 vs 61.860.7 mm1 and 57.7 prepared by mixing intact and completely haemo-
mm1 vs 59.161.1 mm1 at 400 mosmol/L. The cal- lyzed blood without removing the membrane re-
culated values of g were 0.994 vs 0.994360.0005 at siduals. Figure 10 shows the measured optical
250 mosmol/L, 0.9924 vs 0.993860.0004 under iso- properties at five different degrees of haemolysis
tonic conditions and 0.990 vs 0.991760.0006 at 400 between 0% and 100%.
mosmol/L. The agreement of measurements and The absorption coefficient m a significantly de-
calculations indicated that the shape of the erythro- creased with increasing haemolysis. When blood
cytes was of minor importance for the scattering was completely haemolyzed, m a amounted to 0.11
properties of blood. Variations in the scattering be- 60.03 mm1 which was only 55% of the value for
havior were due to changes in cell volume and re- intact blood (0.2060.05 mm1). Nevertheless, the
fractive index. In addition, these results showed dependence of haemoglobin absorption on
that the dependence of the absorption and scatter- haemolysis rate was as unexpected as the increase
ing behavior on shear rate was mainly caused by a of absorption with increasing osmolarity because

Fig. 10 Mean values 6 SD ( n 53) of m a , m s , g, and m s8 vs extent of haemolysis (hct57.5%, p 5300 mosmol/L, g 5500 s1, SatO2
.98%, l5633 nm).


JANUARY 1999 d
VOL. 4 NO. 1
Downloaded from SPIE Digital Library on 12 Jun 2010 to Terms of Use:

not be assumed to occur at point scatterers with

absorption taking place in the surrounding (non-
scattering) medium. In blood the scatterers them-
selves have a significant dimension compared to
the whole volume and they contain all or at least
part of the absorbers. Hence it has to be taken into
consideration that the optical path of a photon
Fig. 11 Path of a photon within an erythrocyte.
within a scatterer might be increased due to mul-
tiple reflections at its internal boundary (Figure 11).
This effect of internal reflections is well known
the overall haemoglobin concentration was kept from atmospherical scattering in water drops and is
constant. The fact that absorption was higher in responsible for the occurrence of rainbows. The in-
blood than in a haemoglobin solution with the creased pathlength on its part leads to a higher ab-
same concentration was also found by comparing sorption probability of the whole solution if absorb-
literature data and scaling the reported absorption ers are contained within the scatterers.
coefficients to our concentration of 27 g/L (hct Consequently, this resulted in a decrease of the
57.5%). For an oxygenated haemoglobin solution overall absorption coefficient if more erythrocytes
Gordy,22 Steinke,23 and Kuenstner24 reported a loose their scattering properties due to haemolysis.
mean m a of 0.06460.013 mm1 at 633 nm. On the The described phenomenon also explains the in-
other hand, Reynolds19 and Steinke25 reported a crease of m a if osmolarity increases (see Figure 8).
mean m a of 0.1260.04 mm1 for whole human Here the RBC refractive index increased due to the
blood. The values were slightly smaller than our higher RBC haemoglobin concentration in the hy-
data but the same factor of approximately two perosmolar state. Hence the number of internal re-
could be deduced. flections also increased, resulting in a higher total
The scattering coefficient m s showed a distinct de- absorption coefficient. Nonlinear absorption aspects
crease with increasing haemolysis. At complete as a concentration dependent extinction coefficient
and the Sieve effect can be excluded to describe
haemolysis m s amounted to 18% of the value mea-
the results because they would reduce m a with in-
sured for intact blood. The anisotropy factor g was
constant up to 50% haemolysis and slightly de- creasing inner cell haemoglobin concentration and
not vice versa.
creased at higher haemolysis rates. m s8 decreased
continuously, reaching values of about 16% com-
pared to intact blood. The decrease of scattering
with increasing haemolysis rate followed from the 3.5 OXYGEN SATURATION
fact that destroyed erythrocytes distributed their
haemoglobin into the whole solution, resulting in a To elucidate the influence of oxygen saturation on
refractive index match. Thus the RBCs loose their the optical properties of blood a sample (hct 41%)
scattering properties. Consequently the measure- was investigated at different oxygen saturations be-
ments showed that the residual erythrocyte mem- tween 100% and 30%, adjusted by insufflation of N2
branes have only a minor impact on the scattering and CO2. Figure 12 shows the measured values of
properties that might be understood from their m a and m s8 versus O2 saturation. At a wavelength of
very small wall thickness of about 4 nm. 633 nm m a showed a linear decrease with increasing
Even more remarkable was the significant de- O2 saturation. Completely oxygenated blood
crease in absorption when blood was haemolyzed showed approximately half the absorption of blood
or osmolarity was decreased. This phenomenon with an oxygen saturation of 30%. The reduced
found an explanation in the special scattering and scattering coefficient m s8 remained constant over the
absorbing structure of blood. Here scattering can- measured range of oxygen saturation. As was ex-

Fig. 12 Absorption coefficient m a and reduced scattering coefficient m s8 vs O2 saturation (hct541%, p 5300 mosmol/L, g 5500 s1, l
5633 nm).


JANUARY 1999 d
VOL. 4 NO. 1 43
Downloaded from SPIE Digital Library on 12 Jun 2010 to Terms of Use:

Fig. 13 The absorption spectrum of oxygenated and deoxygenated diluted blood and the values of m s and g which did not significantly
vary with oxygenation (hct55%, p 5300 mosmol/L, g 5500 s1).

pected, oxygenation only changed the absorption tion. In the range of 1170 to 1350 nm blood absorp-
properties of blood. Scattering remained unaffected tion was approximately twice that of water, above
by changes in oxygen saturation. 1350 nm m a followed the water absorption on a
level of approximately 120%130%. Thus the same
3.6 OPTICAL PROPERTIES OF DILUTED effect of increased absorption that was found in the
BLOOD IN THE WAVELENGTH visible region also appeared for the water absorp-
RANGE 4002500 NM tion in the near infrared where other absorbers
were negligible. This might also be the result of the
To obtain a complete optical spectrum of blood,
internal photon reflection, because it does not mat-
measurements in the wavelengths range 4002500
ter whether absorption within a scatterer is caused
nm were conducted in increments of Dl510 nm.
by water or haemoglobin.
Figure 13 shows the mean values (n53) of m a , m s ,
In contrast to the absorption coefficient no signifi-
and g for oxygenated isotonic blood at hct55% and
cant differences between scattering coefficients or
a shear rate of 500 s1. In addition, the measure-
anisotropy factors were found for both oxygen
ments were repeated for completely deoxygenated
blood samples. Significant differences were only saturations over the whole measured wavelength
found for the absorption coefficient in the wave- range. The scattering coefficient m s showed a dis-
length range of 4001200 nm. Here the spectra cor- tinct maximum at 500 nm with values over 40 mm1
related qualitatively with literature data on haemo- which was consistent with theoretical investiga-
globin solutions.2224 The absorption spectra tions of Pittman.26 In the wavelength range 550800
showed the well-known characteristic maximums nm m s slightly decreased to values of 30 mm1.
around 420 and 540 nm and the isosbestic point at Above 800 nm m s decreased proportionally with
805 nm. However, m a was approximately twice the wavelength to l 21.7. At 2500 nm m s amounted to
reported haemoglobin solutions if they were scaled 10% of the value found at 500 nm. The decrease of
to our measured concentration of 18 g/L (hct m s with wavelength was due to a reduction of the
55%). Above 1200 nm the absorption spectra of scattering cross section and was consistent with
oxygenated and deoxygenated blood were not sig- Mie calculations. Considering the measured dilu-
nificantly different and followed the water absorp- tion of the investigated blood sample (hct55%)


JANUARY 1999 d
VOL. 4 NO. 1
Downloaded from SPIE Digital Library on 12 Jun 2010 to Terms of Use:

very high scattering coefficients above 350 mm1 at 9. The membranes of red blood cells had only a
500 nm can be expected for blood with physiologi- minor impact on the scattering behavior.
cal RBC concentrations (hct'41%). The anisotropy 10. Oxygenation only changed the absorption
factor g was larger than 0.98 in a wavelength range properties.
from 400 to 1400 nm with maximum values over
0.99 in the range 400 to 600 nm. Above 1400 nm g The spectral overview in the wavelength range
decreased slightly to values about 0.95 in the near 4002500 nm at hct55% lead to the following con-
infrared. The increasing oscillation of g above 1800 clusions:
nm was caused by the small values of the measured 1. Blood absorption followed haemoglobin ab-
quantities. Especially R d was found to be far below sorption in the range 4001200 nm on a level
0.1% in this range. These results confirm the for- of approx. 200%.
ward scattering characteristic of blood over a wide
range of wavelength. 2. Blood absorption followed water absorption
at wavelength above 1200 nm on a level of
120% to 200%.
4 SUMMARY AND CONCLUSION 3. A distinct difference between oxygenated and
Fundamental knowledge about the optical behavior deoxygenated blood was only found for the
of human blood could be derived in the present absorption coefficient in the range 4001200
study. We found that measurements under flow nm.
conditions were a necessary prerequisite to get ex- 4. Scattering showed a maximum around 500
act and reproducible results on blood samples. The nm and decreased with wavelength at a rate
double integrating sphere technique proved to be a of approximately l 21.7.
reliable method to determine macroscopic optical 5. Anisotropy was higher than 0.99 in the range
properties. It was necessary to apply inverse Monte 400800 nm, higher than 0.98 in the range
Carlo simulations that took into consideration all 8001400 nm and higher than 0.9 between
kinds of systematic errors to calculate precise in- 1400 and 2500 nm.
trinsic optical properties from the measured quan-
tities. One of the most important details was the One can conclude that scattering of blood is
application of an optimized scattering phase func- mainly caused by a high refractive index of the hae-
tion for red blood cells. We found that the moglobin solution within the RBCs. Internal reflec-
tions of photons lead to an increase of the overall
GegenbauerKernel phase function with a GK51 fit
absorption coefficient by increasing the optical path
best to the scattering behavior of blood samples.
length within the cells. This effect explains the dif-
The impact of variations of the most important
ference between normal and haemolysed blood as
physiological blood parameters on the optical prop-
well as the increase of absorption with increasing
erties was measured at 632.8 nm using a HeNe la-
osmolarity and the increased water absorption in
ser. We found that
the near infrared.
1. Scattering and absorption increased linearly The presented data provide the calculation of op-
with haematocrit if hct,50%. tical properties of blood for any given haematocrit
up to 50% because scattering increases linearly with
2. Absorption and scattering decreased slightly haematocrit and anisotropy remains unchanged.
with increasing shear rate. For the adaptation of the absorption coefficient it is
3. Axial migration was the predominant factor necessary to separate between haemoglobin and
on the optical properties with respect to the water absorption.
flow parameters.
4. The deformation of erythrocytes had no sig-
nificant impact on the optical properties if vol-
1. A. Ishimaru, Wave Propagation and Scattering in Random Me-
ume and haemoglobin content was kept con- dia, Vol. 1, Academic, New York (1978).
stant. 2. A. Ishimaru, Wave Propagation and Scattering in Random Me-
5. Increasing osmolarity lead to an increase of dia, Vol. 2, Academic, New York (1978).
3. J. Reichmann, Determination of absorption and scattering
absorption while scattering and anisotropy coefficients for nonhomogenious media 1: Theory, Appl.
decreased. Opt. 12(8), 18111815 (1973).
4. A. Roggan, O. Minet, C. Schroeder, and G. Mueller, Mea-
6. Mie calculations on spherical equivalents pro- surements of optical properties of tissue using integrating
vided a precise estimation of the RBC scatter- sphere technique. SPIE Institute for Advanced Optical Tech-
ing behavior. nologies, in Medical Optical Tomography: Functional Imaging
and Monitoring, G. Muller et al., Eds., Vol. IS11, pp. 149165,
7. Increasing haemolysis led to a reduction in SPIE Press, Bellingham, WA (1993).
absorption and scattering. 5. A. Roggan, K. Dorschel, O. Minet, D. Wolff, and G. Muller,
The optical properties of biological tissue in the near infra-
8. Haemoglobin solutions had a smaller absorp- red wavelength rangereview and measurements, in
tion coefficient than whole blood at the same Laser-induced Interstitial Thermotherapy, G. Muller and A.
haemoglobin concentration. Roggan, Eds., pp. 1044, SPIE Press, Bellingham, WA (1995).


JANUARY 1999 d
VOL. 4 NO. 1 45
Downloaded from SPIE Digital Library on 12 Jun 2010 to Terms of Use:

6. A. Roggan, Dosimetrie thermischer Laseranwendungen in and W. M. Star, Skin optics, IEEE Trans. Biomed. Eng. 36,
der MedizinUntersuchung der optischen Gewebeeigen- 11461154 (1989).
schaften und mathematisch-physikalische Modellbildung, 17. M. Essenpreis, P. van der Zee, P. S. Jones, P. Gewehr, and T.
PhD Dissertation, Technische Universitat Berlin, in N. Mills, Changes in scattering phase function of rat liver
Fortschritte in der Lasermedizin, G. Muller and H.-P. Berlien, at 1.064 mm and 1.321 mm following photocoagulation, La-
Eds., Vol. 16, Ecomed, Landsberg (1997). sers Surg. Med. 11, Suppl. 3, 5 (1991).
7. P. Kubelka and F. Munk, Ein Beitrag zur Optik der Farban- 18. A. N. Yaroslavsky, I. V. Yaroslavsky, T. Goldbach, and H. J.
striche, Z. Tech. Phys. (Leipzig) 12, 593601 (1931). Schwarzmaier, The optical properties of blood in the near
8. L. G. Henyey and J. L. Greenstein, Diffuse radiation in the infrared spectral range, Proc. SPIE 2678, 314324 (1996).
galaxy, Astrophys. J. 93, 7083 (1941). 19. L. Reynolds, Diffuse reflectance from a finite blood me-
9. H. C. van de Hulst, Multiple Light Scattering, Vol. 1, Aca- dium: applications to the modeling of fiber optic catheters,
demic, New York (1980). Appl. Opt. 15(9), 20592067 (1976).
10. H. C. van de Hulst, Multiple Light Scattering, Vol. 2, Aca- 20. J. R. Zijp and J. J. ten Bosch, Pascal program to perform
demic, New York (1980). Mie calculations, Opt. Eng. 32(7), 16911695 (1993).
11. S. T. Flock, B. C. Wilson, and M. S. Patterson, Total attenu- 21. R. A. MacRae, A. McClure, and P. Latimer, Spectral trans-
ation coefficients and scattering phase functions of tissue mission and scattering properties of red blood cells, J. Opt.
and phantom materials at 633 nm, Med. Phys. 14(4), 835 Soc. Am. 51(12), 13661372 (1961).
841 (1987). 22. E. Gordy and L. Drabkin, Determination of the oxygen
12. S. L. Jacques, C. A. Alter, and S. A. Prahl, Angular depen- saturation of blood by a simplified technique applicable to
dance of HeNe laser light scattering by human dermis, La- standard equipment, J. Biol. Chem. 227, 285299 (1957).
sers Life Sci. 1(4), 309333 (1987). 23. J. M. Steinke and A. P. Shepherd, Effects of temperature on
13. M. Arnfield, J. Tulip, and M. McPhee, Optical propagation optical absorbance spectra of oxy-, carboxy-, and deoxyhe-
in tissue with anisotropic scattering, IEEE Trans. Biomed. moglobin, Clin. Chem. 38(7), 13601364 (1992).
Eng. 35, 372381 (1988). 24. J. T. Kuenstner and K. H. Norris, Spectrophotometry of
14. S. Prahl, Light Transport in Tissue, PhD Dissertation, human hemoglobin in the near infrared region from 1000 to
University of Texas at Austin (1988). 2500 nm, J. Near Infrared Spectrosc. 2, 5965 (1994).
15. R. Marchesini, A. Bertoni, S. Andreola, E. Melloni, and A. E. 25. J. M. Steinke and A. P. Shepherd, Diffusion model of the
Sichirollo, Extinction and absorption coefficients and scat- optical absorbance of whole blood, J. Opt. Soc. Am. A 5(6),
tering phase functions of human tissues in vitro, Appl. Opt. 813822 (1988).
28(12), 23182324 (1989). 26. R. N. Pittman, In vivo photometric analysis of hemoglo-
16. M. J. C. van Gemert, S. L. Jacques, H. J. C. M. Sterenborg, bin, Ann. Biomed. Eng. 14, 119137 (1986).


JANUARY 1999 d
VOL. 4 NO. 1
Downloaded from SPIE Digital Library on 12 Jun 2010 to Terms of Use: