`Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
`
`Optical properties of human skin
`
`Tom Lister
`Philip A. Wright
`Paul H. Chappell
`
`Petitioner Apple Inc. – Ex. 1033, cover p. 1
`
`
`
`Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Biomedical-Optics on 11/1/2018
`Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
`
`Journal of Biomedical Optics 17(9), 090901 (September 2012)
`
`REVIEW
`
`Optical properties of human skin
`
`Tom Lister,a,b Philip A. Wright,a and Paul H. Chappellb
`aWessex Specialist Laser Centre, Salisbury District Hospital, Salisbury, SP2 8BJ, United Kingdom
`bUniversity of Southampton, School of Electronics and Computer Science, Southampton, SO17 1BJ, United Kingdom
`
`Abstract. A survey of the literature is presented that provides an analysis of the optical properties of human skin,
`with particular regard to their applications in medicine. Included is a description of the primary interactions of light
`with skin and how these are commonly estimated using radiative transfer theory (RTT). This is followed by analysis
`of measured RTT coefficients available in the literature. Orders of magnitude differences are found within published
`absorption and reduced-scattering coefficients. Causes for these discrepancies are discussed in detail, including
`contrasts between data acquired in vitro and in vivo. An analysis of the phase functions applied in skin optics,
`along with the remaining optical coefficients (anisotropy factors and refractive indices) is also included. The survey
`concludes that further work in the field is necessary to establish a definitive range of realistic coefficients for
`clinically normal skin. © 2012 Society of Photo-Optical Instrumentation Engineers (SPIE). [DOI: 10.1117/1.JBO.17.9.090901]
`
`Keywords: skin optics; radiative transfer theory; absorption; scatter; melanin; hemoglobin.
`Paper 12354V received Jun. 6, 2012; revised manuscript received Jul. 31, 2012; accepted for publication Jul. 31, 2012; published online
`Sep. 24, 2012.
`
`1 Introduction
`The color of human skin has long been used as a subjective
`adjunct to the detection and diagnosis of disease. More recently,
`the introduction of skin color measurements has extended this
`to include the potential for objective determination of skin fea-
`tures,1 including melanin and hemoglobin concentrations,2–6 the
`depth and diameter of blood vessels,7–9 the depth of pigmented
`skin lesions,10,11 the maturity and depth of bruises,12,13 and
`keratin fiber arrangements.14
`Such advances have proved invaluable for the advancement
`of skin laser treatments15 and photodynamic therapy,16–18 and
`have contributed to further advances in the diagnosis of cancer-
`ous and noncancerous skin lesions.10,19–21
`However, the success of these methods depends entirely
`upon adequate knowledge of the behavior of light as it impinges
`upon, and travels through, the skin. This article presents a
`description of the major interactions of visible light with skin
`and the principal skin features that contribute to these. This
`is followed by an analysis of published optical coefficients
`used in simulations of light transport through skin.
`
`2 Background
`
`2.1 Absorption
`
`Absorption describes a reduction in light energy. Within the
`visible region, there are two substances generally considered
`to dominate the absorption of light in skin: hemoglobin and
`melanin.
`Hemoglobin is the dominant absorber of light in the dermis.
`Normal adult hemoglobin (Hb A) is a protein consisting four
`polypeptide chains, each of which is bound to a heme.22 The
`heme in Hb A is named iron-photoporphyrin IX23,24 and is
`responsible for the majority of light absorption in blood. The
`free-electron molecular-orbital model describes this absorption
`
`as an excitation of loosely bound “unsaturation electrons” or
`“π-electrons” of the heme.25 Within the visible region, Hb A
`contains three distinctive peaks. The dominant peak is in the
`blue region of the spectrum and is referred to as the Soret
`peak or Soret band. Two further peaks can be distinguished
`in the green-yellow region, between 500 and 600 nm, that in
`combination with the Soret band cause Hb A to appear red.
`These are known as the α and β bands, or collectively as the
`Q-band, and have intensities of around 1% to 2% of the
`Soret band.26 The excitation levels of π-electrons vary, and
`therefore the positions and intensities of these bands vary
`with the ligand state of the heme (Fig. 1).
`Melanins are ordinarily contained within the epidermis and
`produce an absorption spectrum that gradually decreases from
`the ultraviolet (UV) to the infrared (IR) regions. In contrast to
`hemoglobin, the variation and complexity of melanins means
`that their detailed structures are not yet fully understood, despite
`intense research over the last five decades, and this broadband
`absorbance spectrum is still a topic of scientific debate.5,28,29 At
`present, the scientific consensus appears to gravitate towards a
`chemical disorder model.5,28–33 This model proposes that mela-
`nins consist of a collection of oligomers or polymers in various
`forms arranged in a disordered manner. This results in a number
`of absorption peaks that combine to create a broadband absor-
`bance effect28,30 (Fig. 2).
`Further absorption of light may be attributed to chromophores,
`such as bilirubin and carotene,34 lipids,35 and other structures,
`including cell nuclei and filamentous proteins36,37 Although
`the individual contributions from these secondary chromophores
`may be considered separately,13,38,39 most simulations group them
`into a single overarching value.40,41
`Despite its abundance in all tissues, water is not a significant
`absorber of light in the visible region, although its contribution
`has been considered when simulating skin color.42
`
`Address all correspondence to: Tom Lister, Wessex Specialist Laser Centre, Salis-
`bury District Hospital, Salisbury, SP2 8BJ, United Kingdom. Tel: 01722780104;
`E-mail: tom.lister@soton.ac.uk
`
`0091-3286/2012/$25.00 © 2012 SPIE
`
`Journal of Biomedical Optics
`
`090901-1
`
`September 2012 (cid:127) Vol. 17(9)
`
`Petitioner Apple Inc. – Ex. 1033, p. 1
`
`
`
`Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Biomedical-Optics on 11/1/2018
`Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
`
`Lister, Wright and Chappell: Optical properties of human skin
`
`collagen is the principal filamentous protein of the dermis and
`occupies approximately 18% to 30% of its volume.45 Further
`scatter is attributed to melanosomes in the epidermis, cell nuclei,
`cell walls and many other structures in the skin that occur in
`smaller numbers.46
`Scatter from filamentous proteins has been approximated
`using a Mie solution to Maxwell’s equations applied to data
`from in vitro skin samples.47,48 This approach provides an
`increase in simulated scattering probability with increasing
`fiber diameter and decreasing wavelength. The dependence of
`scatter on fiber diameter suggests that the protein structures
`of the dermis, which may be 10 times as large as those in
`the epidermis,45,49 possess a greater scattering cross-section.
`This in part compensates for the lower number densities of fila-
`mentous proteins in the dermis. The scattering events that occur
`are mainly in the forward direction, meaning that on average,
`light that returns to the surface undergoes a large number of scat-
`tering events.50 One implication of the wavelength dependence
`of scatter is that blue and green light that has returned to the
`surface of the skin will have, on average, travelled less deeply
`than red light. This is considered the primary reason why blood
`vessels and pigmented nevi that are situated deeper within the
`skin are only able to absorb light from the red end of the
`spectrum and therefore appear bluer than their superficial
`equivalents.51,52
`The volume fraction of melanosomes in the epidermis varies
`typically from 1% in pale skin to 5% in darker skin,53 although
`one group has suggested greater values.54,55 However, despite
`their
`low numbers relative to keratins, melanosomes are
`approximately 10 times the diameter of the largest keratin struc-
`tures in the epidermis56 and possess a greater refractive index57
`(and therefore a greater difference in refractive index at their
`interface with skin). Melanin has been shown to contribute sig-
`nificantly to the degree of scatter within the epidermis.58,59 As
`well as the volume fraction, the distribution and size of melanin
`structures in the epidermis also vary with skin type. Thus, the
`total amount of scatter that occurs as a result of melanin in the
`epidermis can vary substantially between individuals,60,61
`although this is not always taken into account when simulating
`the effects of varying melanin concentration on skin color,42,62
`or when simulating laser treatments,15,63 for example.
`Blood normally occupies around 0.2% to 0.6% of the physical
`volume of the dermis2,6,54,64–66 depending upon its anatomical
`location. The vessel walls surrounding this blood, in addition
`to the walls of vessels that remain vacant, may occupy a similar
`volume. Dermal vessels vary in thickness and structure from
`capillaries of around 10 to 12 μm diameter at
`the epider-
`mal junction to terminal arterioles and post-capillary venules
`(approximately 25 μm in diameter) in the papillary dermis and
`venules (approximately 30 μm) in the mid-dermis.67 Furthermore,
`blood vessels occur in higher densities at particular depths, giving
`rise to so-called blood vessel plexi.67 The contribution to light
`scatter by these structures, inclusive of refraction effects, may
`be significant 68–70 and varies with location and depth, as well
`as between individuals.* Larger, deeper vessels may also
`contribute to the color of skin.
`
`*Assuming a reduced scattering coefficient of 0.5 mm−1 for blood and 40 mm−1
`for vessel walls at 633 nm, a 0.5% volume fraction of each contributes approxi-
`mately 0.2 mm−1 to the dermal reduced scattering coefficient, measured at
`around 1-5 mm−1 (Fig. 9). The contribution will be larger within blood vessel
`plexi.
`
`Absorption Spectra for Four Ligand States of
`Human Adult Haemoglobin
`
`Hb
`
`Hb02
`
`HbCO
`
`MetHb
`
`18
`
`16
`
`14
`
`12
`
`10
`
`8
`
`6
`
`4
`
`2
`
`Millimolar Absorptivities (L.mmol-1.cm-1)
`
`0
`450
`
`500
`
`550
`
`650
`600
`Wavelength (nm)
`
`700
`
`750
`
`Fig. 1 Absorption spectra of deoxyhemoglobin (Hb), oxyhemoglobin
`(HbO2), carboxyhemoglobin (HbCO), and methemoglobin (MetHb) in
`the visible region, from Ref. 27.
`
`Fig. 2 Colored lines show individual absorption spectra of tetramer
`subunits within melanin extracted from human epidermis. The average
`absorption spectrum of these is shown by thick black line, shifted up
`1.5 units for clarity. Thin black lines shifted down by one unit represent
`absorption spectra from monomer subunits. a:u: ¼ arbitrary units.
`Reprinted figure with permission from Ref. 30.
`
`2.2 Scattering
`
`As well as absorption, scattering contributes significantly to the
`appearance of skin. Scattering describes a change in the direc-
`tion, polarization or phase of light and is commonly portrayed as
`either a surface effect (such as reflection or refraction) or as an
`interaction with a small region whose optical properties differ
`from its surroundings (particulate scatter).
`It has been estimated that 4% to 7% of visible light is
`reflected from the surface of the skin, independent of wave-
`length and skin color.43,44 The remaining light is refracted as
`it passes from air into the skin.
`The primary sources of particulate scatter within the skin are
`filamentous proteins. Keratins are the filamentous proteins of
`the epidermis and form this layer’s major constituent, whereas
`
`Journal of Biomedical Optics
`
`090901-2
`
`September 2012 (cid:127) Vol. 17(9)
`
`Petitioner Apple Inc. – Ex. 1033, p. 2
`
`
`
`Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Biomedical-Optics on 11/1/2018
`Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
`
`Lister, Wright and Chappell: Optical properties of human skin
`
`Scattering from the remaining structures of the skin, includ-
`ing cell walls, nuclei, and organelles,36 hairs and glands, is rarely
`of central interest to a study of skin optics. As a result, the con-
`tributions from these structures to the total measured scattering
`coefficients are not routinely considered separately.71
`
`4 Optical Coefficients of Skin
`A considerable amount of work has been carried out to deter-
`mine appropriate values of the RTT coefficients. Cheong et al.74
`described both direct (in vitro) and indirect (in vivo) methods of
`measuring absorption and scatter. A comprehensive analysis of
`the literature involving each method is presented here.
`
`4.1 Absorption Coefficients
`
`4.1.1 In vitro absorption coefficients
`
`to produce repeat
`Direct measurements have the potential
`measurements of a predetermined volume or section of skin
`and, unlike in vivo measurements, can include transmission
`data. However, the processes necessary to extract and prepare
`a skin sample cannot be carried out without altering its optical
`properties.
`The in vitro studies presented in Fig. 3 vary significantly in
`tissue-processing methodologies, measurement setup and the
`interpretation of data. For example, Jacques et al.’s work75
`included three methods of tissue preparation. The epidermis was
`separated from the dermis using a micro-cryotome for one set
`of skin samples, after mild thermal treatment in a water bath in
`another set, and was not separated in a third set. The same mild
`thermal treatment was used to separate the dermis and epidermis
`in Prahl’s work55 and a micro-cryotome was also applied in
`Salomatina et al.’s study.59 No separation of the epidermis
`was reported by Chan et al.76 or Simpson et al.77 Although
`Salomatina’s work shows greater absorption from the in vitro
`epidermis when compared to the dermis, the studies analyzed
`here do not demonstrate a clear distinction in absorption coeffi-
`cients reported between the methods of separation described,
`nor between those that separated the epidermis and those that
`did not.
`The level of hydration is likely to have varied considerably
`between the studies analyzed. Prahl55 and Jacques et al.75 soaked
`samples in saline for at least 30 min before carrying out mea-
`surements, during which the samples were placed in a tank of
`saline. Salomatina et al.59 also soaked their skin samples prior to
`measurement and sealed them between glass slides to maintain
`hydration. Chan et al.76 and Simpson et al.77 did not soak their
`samples prior to or during measurement. Jacques et al. reported
`that soaking the sample increases back-scattered reflectance,
`although the effects on the calculated absorption are not
`described. Chan et al. commented that dehydration may elevate
`the measured absorption coefficient. However,
`the greatest
`reported absorption coefficients are those from rehydrated tissue
`samples. From the information available, the effect of tissue
`hydration on the measured absorption coefficients is not clear.
`Data was interpreted using Monte Carlo simulations by
`Salomatina et al.,59 Simpson et al.77 and Graaf et al.,78 an
`adding-doubling technique by Prahl et al.,55 and by direct inter-
`pretation in Chan et al.’s76 and Jacques et al.’s75 studies. Both the
`methods described in the Monte Carlo simulations and Prahl’s
`adding-doubling technique are based upon assumptions of opti-
`cally homogeneous tissue layers, uniform illumination and no
`time dependence, and both are essentially discrete solutions
`to the radiative transport equation. The methods described con-
`trast in their approach to internal reflection for beams exiting the
`skin model and the adding-doubling method relies upon accu-
`rate representation of the angular distribution of beams exiting
`the thin layer upon which the model is built. It is not directly
`
`3 Simulating Light Transport Through Skin
`Optical simulations involving mathematical models of healthy
`human skin generally approximate the surface as perfectly
`smooth, although some computer graphics models have applied
`calculations of directional reflectance from rough surfaces.72
`Surface scattering effects (reflection and refraction) can be
`calculated for smooth surfaces using Fresnel’s equations and
`Snell’s law, respectively:
`
`
`ða − cÞ2
`ða þ cÞ2
`
`1 þ ½cða þ cÞ − 12
`½cða − cÞ þ 12
`
`:
`
`(1)
`
`R ¼ 1
`
`2
`
`Fresnel reflection (R) of unpolarized light from air to skin,
`where c ¼ cosðθiÞ, θi is the angle of incidence, a ¼ n2 þ c2 − 1
`and n is the refractive index of skin.
`
`
`
`(2)
`
`sin θi
`
`:
`
`1 n
`
`θt ¼ arcsin
`
`Angle of refraction (θt) at the skin’s surface calculated using
`Snell’s law.
`Within the skin, both absorption and scatter must be consid-
`ered simultaneously. These may be described in the classical
`approach by Maxwell’s equations, which consider the interac-
`tions between the electric and magnetic fields of light with mat-
`ter. However, an exact solution to Maxwell’s equations requires
`precise knowledge of each structure within the medium and
`becomes prohibitively complex for the case of human skin.
`The most commonly used approximation to Maxwell’s equa-
`tions in the field of skin optics is radiative transfer theory
`(RTT).73 This considers the transport of light in straight lines
`(beams),with absorption simulated as a reduction in the radiance
`of a beam and dependent upon the absorption coefficient (μa).
`The degree of scattering is described by the scattering coeffi-
`cient (μs), which considers both a loss of radiance in the direc-
`tion of the beam and a gain from beams in other directions, and
`the phase function (p), the probability that an individual beam
`will scatter in any particular direction. The reduced-scattering
`s ¼ μsð1 − gÞ,
`
`coefficient (μ 0s) combines these variables, i.e., μ 0
`where g is the anisotropy factor, the average cosine of the
`scattering angle θ:
`Z
`
`g ¼
`
`4π
`
`pðcos θÞ cos θdω;
`
`(3)
`
`where dω is a differential solid angle.
`In order for RTT to be valid, it must be assumed that any
`cause for increasing or decreasing the radiance of a beam other
`than that described by the absorption and scattering coefficients,
`including inelastic scatter (fluorescence or phosphorescence)
`and interactions between beams (interference), is negligible. The
`skin model must also consist of volumes that are homogeneous
`with regards to μs, μa and p, and that do not change over time.
`
`Journal of Biomedical Optics
`
`090901-3
`
`September 2012 (cid:127) Vol. 17(9)
`
`Petitioner Apple Inc. – Ex. 1033, p. 3
`
`
`
`Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Biomedical-Optics on 11/1/2018
`Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
`
`Lister, Wright and Chappell: Optical properties of human skin
`
`Summary of Published Absorption Coefficients Measured In Vitro
`
`Summary of Published Absorption Coefficients Measured In Vivo
`
`10
`
`1
`
`0.1
`
`Dognitz 1998*
`
`Toricelli 2001* (mean)
`
`Meglinski 2002 epidermis†
`
`Meglinski 2002 dermis†*
`
`0.01
`
`Graaf 1993
`
`Zonios 2006
`
`Absorption coefficient (mm-1)
`
`Jacques 87
`Prahl 1988*
`Chan 1996*
`Salomatina 2006 epidermis*
`Salomatina 2006 dermis*
`Graaf 1993
`Simpson 1998*
`
`10
`
`1
`
`0.1
`
`0.01
`
`Absorption coefficient (mm-1)
`
`0.001
`350
`
`400
`
`450
`
`600
`550
`500
`Wavelength (nm)
`
`650
`
`700
`
`750
`
`0.001
`350
`
`Doornbos 1999† arm
`
`Doornbos 1999† forehead
`
`Svaasand 1995 epidermis
`
`Svaasand 1995 dermis
`
`Bosschaart 2011
`
`400
`
`450
`
`600
`550
`500
`Wavelength (nm)
`
`650
`
`700
`
`750
`
`Fig. 3 Summary of absorption coefficients available in the literature. In vitro data represents absorption coefficients from exsanguinated skin whereas
`dermal in vivo data is inclusive of blood absorption. *Data obtained from graphical presentation. †Data presented was not complete and required input
`of hemoglobin or water optical properties obtained from.35 The raw data is provided in the Appendix (Table 1).
`
`clear if or how these differences may have contributed to the
`higher absorption coefficients reported by Prahl et al.
`It is also of interest that Prahl et al.’s,55 Chan et al.’s76 and
`Salomatina et al.’s59 studies, whose samples varied in thickness
`between 60 and 780 μm, did not demonstrate a clear correlation
`between sample thickness and published absorption coefficients
`but Simpson et al.’s published absorption coefficients, which are
`an order of magnitude smaller than the other values analyzed
`involved much thicker samples (1500 to 2000 μm
`here,
`thick). Thus, differences between the published absorption
`coefficients across these studies may have resulted from varia-
`tions in the regions of skin investigated or the ability of the simu-
`lations to correctly account for boundary effects at the lower
`boundary.
`
`4.1.2 In vivo absorption coefficients
`
`Indirect measurements do not suffer from such changes in the
`properties of the interrogated skin volume, although care must
`be taken to consider variations in blood perfusion, for example,
`which may result from sudden changes in ambient temperature,
`the use of some drugs and even contact between the skin and the
`measurement device.79
`In general, absorption coefficients measured in vivo may be
`expected to be higher than in vitro values where the highly
`absorbing pigments from blood are removed from the samples.
`This is particularly true in the blue-green regions of the visible
`spectrum. Assuming a value of 0.5% blood volume in the
`−1 at 410 nm
`dermis, this would contribute approximately 8 cm
`
`−1 at 500 nm and 1.4 cm−1 at 560 nm
`(Soret band), 0.6 cm
`−1 at 700 nm (values calcu-
`(Q-band), but only around 0.05 cm
`lated from35). This contribution is not reflected in the literature.
`Absorption coefficients obtained from in vivo work show greater
`variation, but are not consistently higher than those obtained
`from in vitro work (Fig. 3).
`Absorption coefficients from Svaasand et al.,80 Zonios
`et al.,81 and Meglinski and Matcher42 clearly demonstrate the
`effect of blood on the measured absorption coefficients. Each
`
`study shows an absorption peak between 400 and 450 nm
`corresponding to the Soret band and a double peak at approxi-
`mately 540 and 575 nm corresponding to the α and β bands
`of oxyhemoglobin (see Fig. 1). There are, however, notable
`differences between the absorption coefficients produced from
`the three studies. Meglinski and Matcher and Svaasand et al.
`considered epidermal absorption coefficients separately to der-
`mal values. The reported values from Svaasand et al. are greater,
`and show a different spectral curve to those from Meglinski and
`Matcher. This is a direct result of Svaasand et al.’s inclusion of
`0.2% blood by volume in the calculation of epidermal absorp-
`tion coefficients, representing blood infiltrating the modeled
`epidermal layer from the papillae. Compared to Meglinski and
`Matcher’s dermal values and Zonios et al.’s absorption coeffi-
`cients for their skin model consisting a single layer, both of
`which also included the influence of blood, Svaasand et al.’s
`reported dermal absorption coefficients were consistently high.
`This is despite using a dermal blood volume fraction of 2%,
`compared to an average of 12% from Meglinski and Matcher’s
`study and a value of 2.6% in Zonios et al.’s work. The cause of
`this discrepancy is the variation in magnitude of the blood
`absorption coefficients applied across the three studies (Fig. 6,
`Appendix). Bosschaart et al.82 employed a diffusion approxima-
`tion technique to their data collected from neonates, effectively
`applying a single value of absorption across the skin volume.
`Their data is in close agreement to Meglinski and Matcher’s
`dermal absorption coefficients in the 530- to 600-nm range,
`but the contribution of melanin produces a relative increase
`in Bosschaart et al.’s values at shorter wavelengths.
`Data selected for analysis in this work involved “Caucasian”
`skin types only. Where stated, these studies involved skin types
`described as Northern European. Where not stated,
`it was
`assumed that such skin types were used except for the studies
`carried out by Zonios et al.81 and Torricelli et al.83 that were
`conducted in Southern Europe. However, the latter two studies
`did not report higher absorption coefficients, as may be expected
`from measurements on darker skin types. In Zonios et al.’s
`
`Journal of Biomedical Optics
`
`090901-4
`
`September 2012 (cid:127) Vol. 17(9)
`
`Petitioner Apple Inc. – Ex. 1033, p. 4
`
`
`
`Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Biomedical-Optics on 11/1/2018
`Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
`
`Lister, Wright and Chappell: Optical properties of human skin
`
`work, this is primarily a result of the low values of blood absorp-
`tion coefficient applied. Torricelli et al.83 was the only group to
`apply time-resolved reflectance spectroscopy. This involves a
`prediction of the temporal spread of a laser pulse using a diffu-
`sion model. The values presented can only be as good as the
`diffusion model, and rely upon a wavelength dependence deter-
`mined from phantom measurements.84
`The absorption coefficients from both Graaf et al.’s78 and
`Doornbos et al.’s85 studies were lower than those from the
`remaining studies. Both studies involved an integrating sphere
`and multifiber probe, respectively, as did the higher values from
`Svaasand et al.40 and Meglinski and Matcher.42 Graaf et al. and
`Doornbos et al. applied a Monte Carlo Simulation and diffusion
`approximation respectively, as did Meglinski and Matcher and
`Svaasand et al. The multiple layered mathematical skin models
`that Svaasand et al. and Meglinski and Matcher applied when
`separately considering the effects of the epidermis and dermis
`may be a more accurate approach than the single homogeneous
`layer used in Graaf et al.’s and Doornbos et al.’s work. Although
`the cause of lower values is not clear, Graaf et al. commented
`that their absorption coefficient at 633 nm was “much smaller
`than expected from in vivo” results. Doornbos et al. did not com-
`ment directly on the cause of their low values, but mentioned
`that their “results resemble those of Graaf et al.”
`
`4.2 Scattering Coefficients
`
`4.2.1 In vitro scattering coefficients
`Of the studies analyzed here, both Prahl’s55 and Jacques et al.’s75
`studies describe a number of processes between tissue extraction
`and measurement that are likely to have had an effect on the
`measured reduced-scattering coefficient, including: exposure
`to a 55°C water bath for 2 min to aid with separating the epi-
`dermis from the dermis; freezing, cutting and stacking of 20-μm
`thick slices of the dermis; and soaking in saline to rehydrate and
`wash away any blood. The bloodless samples were then held
`
`between glass slides in a saline-filled tank and illuminated
`using a 633-nm laser. Freezing and drying, heating to remove
`the epidermis and deformation of skin samples have all been
`reported to change the measured values of scattering and absorp-
`tion coefficients.75,77,78 In particular, experimental work by Pick-
`ering et al.86 suggested that heating tissue to 55°C may increase
`the value of μs. Also, Jacques et al.75 commented that soaking
`the dermis (in saline) will increase the backscattered reflectance,
`and thus may increase the calculated scattering coefficient. In
`contrast, Chan et al.76 and Simpson et al.,77 whose reduced-scat-
`tering coefficients were substantially lower, reported minimal
`tissue processing (although Chan et al.’s specimens had pre-
`viously been frozen).
`A further source of disparity between in vivo data from ear-
`lier studies,55,75,78 which involved more tissue processing than
`the more recent in vivo data presented in Fig. 4,59,76,77 may
`have come from the choice of measurement setup. For example,
`Graaf et al.78 reported that discrepancies may arise when internal
`reflectance is not taken into consideration. Due to a larger dif-
`ference in refractive indices, this will have a greater effect for
`samples in air compared to samples in water or saline solution.
`All samples were placed between glass slides. However, only
`the earlier studies analyzed here, those which produced higher
`values of reduced-scattering coefficient, submerged the sample
`in water or saline.55,75
`Salomatina et al.’s
`study59 determined separately the
`reduced-scattering coefficients of the epidermis and dermis.
`Their data show that the epidermal reduced-scattering coeffi-
`−1 higher than the dermal
`cient was consistently 2 to 3 mm
`reduced-scattering coefficient over the visible spectrum. This
`suggests that studies that excluded the epidermis, such Chan
`et al.’s76 and Simpson et al.’s,77 should provide lower values
`of
`reduced-scattering coefficient
`than data obtained from
`studies in which the epidermis remained, such as Graaf
`et al.’s78 and Prahl’s.55 However, most probably due to a prevail-
`ing effect from the aforementioned influences, this is not the
`case.
`
`100
`
`Summary of Published Reduced Scattering Coefficients Measured
`in Vitro
`
`100
`
`Summary of Published Reduced Scattering Coefficients Measured
`in Vivo
`
`10
`
`1
`
`Dognitz 1998*
`
`Toricelli 2001* (mean)
`
`Reduced Scattering Coefficient (mm-1)
`
`600
`550
`500
`Wavelength (nm)
`
`650
`
`700
`
`750
`
`0.1
`350
`
`Graaf 1993
`
`Doornbos 1999 arm
`
`Doornbos 1999 forehead
`
`Svaasand 1995
`
`Bosschaart 2011
`
`Zonios 2006
`
`400
`
`450
`
`600
`550
`500
`Wavelength (nm)
`
`650
`
`700
`
`750
`
`10
`
`1
`
`Jacques 87
`
`Prahl 1988*
`
`Chan 1996*
`
`Salomatina 2006 epidermis*
`
`Salomatina 2006 dermis*
`
`Graaf 1993
`
`Simpson 1998*
`
`0.1
`350
`
`400
`
`450
`
`Reduced Scattering Coefficient (mm-1)
`
`Fig. 4 Summary of reduced- scattering coefficients available in the literature. In vitro data represents absorption coefficients from exsanguinated skin
`whereas dermal in vivo data is inclusive of blood absorption. *Data obtained from graphical presentation. †Data presented was not complete and
`required input of hemoglobin or water optical properties obtained from Ref. 35. Raw data is provided in the Appendix (Table 2).
`
`Journal of Biomedical Optics
`
`090901-5
`
`September 2012 (cid:127) Vol. 17(9)
`
`Petitioner Apple Inc. – Ex. 1033, p. 5
`
`
`
`Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Biomedical-Optics on 11/1/2018
`Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
`
`Lister, Wright and Chappell: Optical properties of human skin
`
`4.2.2 In vivo scattering coefficients
`In addition to their interpretation of Prahl’s in vitro data,55 Graaf
`et al.78 performed measurements of reflection on five male sub-
`jects with “white” skin at 660 nm using an LED source. Despite
`using a similar wavelength light source to Prahl’s 633 nm,
`reduced-scattering coefficients from Graaf et al.’s in vivo
`measurements were appreciably lower than their interpretation
`of in vitro data (Fig. 4). This is likely to be a result of the
`posthumous tissue processing performed in Prahl’s study as pre-
`viously described. However, this effect is not reflected across
`the literature, as in general reduced-scattering coefficients from
`in vivo studies were not substantially lower than those evaluated
`from in vitro studies, nor did they demonstrate an appreciable
`difference when considering the variation in reduced-scattering
`coefficients across the visible spectrum.
`Any solution involving two independent variables (such as
`μ 0
`s and μa) can suffer from nonuniqueness, where equivalent
`results can be obtained from two or more sets of input values
`(local minima). When applying RTT to skin, a simulated
`increase or reduction in reflection can be attributed to a change
`in either μa or μ0
`s. The reduced- scattering coefficients from
`Svaasand et al.’s study40 were considerably higher than any
`of the other in vivo studies assessed. This is in addition to
`their high values of absorption coefficient discussed in the pre-
`vious section. The paper stated that “the fact that the calculated
`[skin reflectance] values tend to be higher than the measured
`ones might indicate that the used values for the epidermal and
`dermal (reduced) scattering coefficients are somewhat too high.”
`The reduced-scattering coefficient from Svaasand et al.’s work
`was derived from a single data point at 577 nm measured by
`Wan et al.87 fitted to a simple μ0
`s ∝ wavelength
`−1 relationship
`and therefore may not be as reliable as data derived from a series
`of direct measurements. It should also be noted that the remain-
`ing studies that produced the highest reduced-scattering coeffi-
`cients analyzed here also provided the highest absorption
`coefficients, including both in vivo and in vitro data. Similarly,
`those studies presenting the lowest reduced-scattering coeffi-
`cients produced the lowest absorption coefficients (Fig. 4).
`Furthermore, when applying high values of dermal scattering
`to a minimization procedure, Verkruysse et al. demonstrated
`