The mechanism of tissue optical properties change induced by tissue deformation

(1)

Joy Willemse

10411070

The mechanism of tissue optical

properties change induced by tissue

deformation

Bachelorproject Natuur- en Sterrenkunde

15 EC

Carried out from 04-04-2016 to 04-07-2016

Handed in on 04-07-2016

Supervisors:

Dr. ir. D.J. Faber

X. Zhang Msc

Second examiner:

Prof. dr. ing. M.C.G. Aalders

University of Amsterdam

Faculty of Science

Natuur- en Sterrenkunde

Academisch Medisch Centrum

Biomedical

engineering

and

physics

(2)

Abstract

Single fiber reflectance spectroscopy (SFR) is a promising method as a diagnostic medical tool for cancer detection. When an optic probe is pressed to tissue, the tissue deforms which changes the tissue optical properties. This could lead to a false diagnosis.

In this report SFR was performed in the visible and near-infrared part of the spectrum on the outer forearm with controlled indentation. A semi-empirical SFR model was utilized to extract the tissue optical properties. Long-term and large deformation induce significant distortions in measured spectra. When applying in-dentation from 0 to 10 mm the oxygen saturation and blood volume fraction was observed to decrease. Water content also decreased with an average of 15% in all measurements. Blood volume fraction dropped with an average of 40%.

A correlation has been found between the changes in the scattering and the changes water content. It shows that because of the decrease in water content caused by indentation the structure and composition of tissue changes. The tissue becomes more dense with scattering particles which results in an increase of the reduced scattering coefficient.

(3)

Populair wetenschappelijke samenvatting

Als licht op huid valt wordt het geabsorbeerd of verstrooid. Dit is afhankelijk van de samenstelling in de huid. Single Fiber Reflectance Spectroscopy is een techniek die gebruik maakt van deze eigenschap voor het maken van diagnoses. Bij kankercellen bijvoorbeeld wordt het licht anders geabsorbeerd en verstrooid dan bij gezonde huid. Hiervoor wordt een heel dunne glasvezel gebruikt, ook wel probe genoemd. Deze wordt op de huid gezet om licht op de arm te schijnen en het verstrooide licht op te vangen. Hierdoor wordt de huid een beetje ingedrukt waardoor het van vorm verandert. Dit heeft invloed op de samenstelling van de huid. Het duwt water en bloed weg en perst de lagen van de huid samen.

Om het effect van deze druk op deze metingen te onderzoeken heb ik drie experi-menten uitgevoerd. In het eerste experiment is de probe gezet op de arm van een vrijwilliger en is in de tijd gemeten wat de veranderingen zijn in de samenstelling van het weefsel, zoals en water concentratie. Het bleek dat de gevonden bloed-waarden, zoals geschatte diameter van de aderen en zuurstofinhoud, varieerden in de tijd. De water concentratie in de huid nam af naarmate de probe langer op de huid stond. In het tweede experiment is de probe op 5 mm diepte in de huid gezet, gemeten, en weer losgelaten. Vervolgens is de meting herhaald, waarbij de probe weer op dezelfde diepte werd gezet. Hierdoor kan bekeken worden hoe reproduceer-baar de metingen zijn. De gevonden waarden waren verschillend van elkaar.

In de derde meting is de probe aangebracht en is de druk in stapjes verhoogd door de probe steeds een halve millimeter harder op de arm te drukken. Het bleek dat hoe hoger de druk was, hoe minder bloed en water er aanwezig was in de huid. De bloedconcentratie nam gemiddeld af met 40% als de probe 10 mm in de huid werd gedrukt. De waterconcen-tratie nam gemiddeld af met 15%. Bij een diepte van meer dan 12 mm nam de waterconcentratie af terwijl het bloed al weggedrukt was. De verandering in waterconcentratie door het indrukken van de huid zorgt voor verandering in struc-tuur en compositie van het weefsel.

Een van de problemen die ik tegengekomen ben is dat de meeste formules in het model dat gebruikt wordt om de verschillende concentraties te bepalen, berusten op een numerieke benadering in plaats van op natuurkundige wetten. Dit zorgt ervoor dat de exacte waarden van de concentraties soms niet betrouwbaar zijn.

De betrouwbaarheid van de meting wordt ook be¨ınvloed door beweging van de arm van de vrijwillger. Het feit dat deze methode zo precies is kan zowel een voordeel als een nadeel zijn. Voor medische applicaties, zoals het opsporen van kankercellen, is het fijn dat de locatie heel precies bepaald kan worden. Aan de andere kant zijn de metingen niet erg reproduceerbaar.

(4)

Chapter 1 Introduction

1.1 Probe indentation effects on fiber reflectance

spectroscopy

Fiber reflectance spectroscopy is often used as a medical tool for examining human skin (Atencio et al., 2009; Cerussi et al., 2009). By sending light into human tissue and collecting the scattered light, properties of the tissue can be evaluated. This is key to diagnostic applications (Jacques, 2013). These applications include detecting breast- and skin cancer, making cancer treatment decisions and performing optical biopsies (Cugmas et al., 2013). In these applications an optic fiber probe is used to collect the light from the tissue. The probe is pressed to the tissue to ensure opti-cal contact. This often uncontrolled indentation can change the acquired spectra, because it significantly affects the optical properties of the tissue (Cugmas et al., 2013). This could lead to a false diagnosis. When the tissue deforms, the internal structure and chromophore density in the interrogated volume changes (Cugmas et al., 2013). Chromophores in tissue include water, fat, collagen, oxygenated and deoxygenated hemoglobin and melanin. It is hypothesized that during the increase indentation on tissue the blood volume fraction decreases because blood is being pushed out (Atencio et al., 2009). When applying indentation for a longer time the oxygen saturation level of the blood (StO2) of the tissue will decrease, because cells

consume oxygen faster than it gets replenished. (Brooks et al., 2015).

Reported changes in µ0_s when applying increasing pressure vary between studies. Chen et al. (2005) found that in the spectral range of 1,100 to 1,700 nm the µ0_s of palm skin decreased with higher pressure. Atencio et al. (2009) reported a pivoting behaviour in the change of µ0_s, with the pivot wavelength ranging in a relative narrow spectral region (590–615 nm). Above this point diffuse reflectance intensity was reported to decrease when probe pressure was increased while below this point diffuse reflectance increases as the probe pressure is increased.

Furthermore, Lim et al. (2011) reported an increase in µ0_sat all wavelengths when increasing the pressure from 9 to 152 mN/mm2 _{when measuring on the forehead.}

They found a decrease in µ0_s when measuring on the neck. This difference in measur-ing sites is used by Cugmas et al. (2013). They tried to classify tissue accordmeasur-ing to their response on increasing pressure. They measured on different sites of the hand palm; above bone, muscle and near the veins. They reported that on all sites the applied pressure (20–100 kPa) increased the concentrations of water, hemoglobin and lipids. The scattering was found to decrease when increasing pressure and the

(6)

1.2. SINGLE FIBER METHOD CHAPTER 1. INTRODUCTION

changes in absorption properties were found to be site-specific.

In this study the outer forearm skin is measured with increasing indentation. Here the tissue can easily deform, which indicates that blood and water are being pushed out. When the light absorption from blood and water decreases, the scatter-ing of the light will increase, so it is expected that µs will go up at all wavelengths.

1.2 Single fiber method

There are multiple types of fiber reflectance spectroscopy, depending on the distance of the fibers that emit and collect the light. In this work, single fiber reflectance spectroscopy was used. This means that the light is emitted and collected by the same fiber. Only the light that is scattered a few times can be collected by the same fiber, resulting in a very small sampling volume (Gamm, 2013). The advantage of using one fiber is that the probe is smaller than systems with a large source-detector separation. This makes it more convenient to use as a medical tool. It also has a smaller sampling volume, which makes it more likely that the sampled tissue is homogeneous. When extracting chromophore concentrations, a homogeneous sam-pling volume is assumed. In some diagnostical applications it is more suitable to probe only shallow depths in tissue (Gamm, 2013). Per example 85% of premalig-nant lesions arise in the epithelium, the most upper skin layer (Sokolov et al., 2002). Probes that measure a smaller volume are more sensitive to changes in these tissues.

1.3 Theoretical background

When light enters tissue it can be absorbed or scattered. The relationship between the intensity of the incident light and the scattered light is known as the Lambert-Beer law:

I = I0e−µa<L> (1.1)

In this equation I0 is the intensity of the incident light and I is the intensity of the

reflected light from the tissue. µais the absorption coefficient of tissue and < L > is

the effective path length of the transmitted light in the tissue (Chen et al., 2005; van Leeuwen-van Zaane et al., 2013). When performing SFR, I0 and I can be measured,

but < L > is dependent on the scattering coefficient µ0_s. µ0_s is given by:

µ0_s = µs(1 − g1) (1.2)

where g1 is the first moment of the scattering phase function. When using small

source-detector separated systems, an additional phase function dependent variable is needed, because light can be scattered in large angles. This variable γ is defined as:

γ = 1 − g2 1 − g1

(1.3) where g2 is the second Legendre moment of the phase function (van Leeuwen-van

Zaane et al., 2013).

The analysis of SFR spectra is performed using a semi-empirical model, tested by Monte Carlo simulations (e.g. Kanick et al. (2011a)) or tissue phantoms (e.g. Gamm et al. (2012)). From this, the tissue absorption can be determined without

(7)

1.3. THEORETICAL BACKGROUND CHAPTER 1. INTRODUCTION

prior knowledge of the scattering coefficient and γ (van Leeuwen-van Zaane et al., 2013). In the model the phase function in tissue is assumed to be constant, so γ is assumed to be wavelength independent. The semi-empirical model is further described in section 2.2.

The reflectance of oxygenized- and deoxygenized blood has been investigated in earlier research, per example by Amelink et al. (2009). They performed Diffuse Path-length Spectroscopy (DPS) on oxygenated and deoxygenated blood samples. Their results are shown in figure 1.1.

Figure 1.1: Reflectance of (a) oxygenized blood and (b) deoxygenized blood. Taken from Amelink et al. (2009)

When applying indentation on tissue, the tissue deforms. Blood is pushed out and the blood supply is blocked, so it is expected that the reflectance spectra shape shifts from the shape of oxygenated blood (a) to deoxygenated blood (b), and with higher pressure the blood absorption dip will disappear completely. Water has a large absorption dip at 1,400 to 1,500 nm. When blood is being pushed out, the water absorption dip will also decrease in amplitude.

The purpose of this work is to investigate this influence and variation of probe indentation on human skin reflectance spectroscopy. For this purpose, multiple SFR measurements are made with different indentation on outer forearm skin of volun-teers. From the results the variation in the composition of absorbing components in tissue when applying indentation with an optic probe is observed.

(8)

Chapter 2 Methods

2.1 Experimental Setup

During measurements, white light emitted by a Deuterium-Halogen light source [Oceanoptics DH2000] is directed through a bifurcated fiber with a diameter df iber

of 1000 nm into the tissue sample as shown in figure 2.1. The scattered light from the sample is collected by the same fiber and directed through another bifurcated fiber into two different spectrographs (SG), one for the visible part (VIS) of the spectrum [Avantes Startline 2048], and one for the near-infrared part (NIR) of the spectrum [OceanOptics NIRQ512]. The spectrographs are connected to a computer for data acquisition. From 400 - 900 nm the VIS spectrograph was used, from 900 - 1700 nm the NIR spectrograph was used. The VIS region of the reflectance showed more variation than the NIR region of the reflectance. This is due to the difference in quality of the spectrographs. The NIR spectrograph has an internal cooling system resulting in a more stable output.

The fiber tip was polished at an angle of 15 degrees to minimize the internal reflection due to the refractive index mismatch between the fiber and the sample. The fiber optical probe is integrated to a motorized linear stage. Tissue deforma-tion was generated and controlled by indentadeforma-tion perpendicular to skin surface over approximately 500 mm2 area. The indentation range was from 0 to 20 mm.

(9)

2.2. SINGLE FIBER REFLECTANCE MODEL CHAPTER 2. METHODS

Figure 2.1: Setup for single fiber reflectance spectroscopy, including a Deuterium-Halogen light source, a linear stage, a fiber optic probe, sample, a NIR spectrograph, a VIS spectrograph and a computer. Adapted from Zonios and Dimou (2006).

For the calibration of the system a reference and a dark measurement were performed. The reference measurement was done with spectralon with a Lambertian reflecting surface. This means the light of the reference was scattered equally in all directions. Dark measurements were performed in a container of water. The reflectance of the sample is then calculated according to:

RSF =

I − D I0− D

(2.1) I is the measured intensity of the sample, D the measured intensity of the dark measurement and I0 the intensity of the reference measurement.

2.2 Single Fiber Reflectance model

Analysis of SFR applies a modified Beer-Lambert law, which accounts for the ab-sorption from tissue chromophores:

RSF = R0SFe

−µa<L> _(2.2)

with RSF the measured reflectance spectrum, < L > the optical path length and

R0

SF the SFR intensity without absorption (Gamm et al., 2011). If < L > and RSF0

are known the concentrations of the chromophores can be derived. R0_SF has been modeled by Kanick et al. (2011a), assuming that the sensitivity of R0

SF depends on

the dimensionless reduced scattering (µ0_sdf iber) resulting in:

R_SF0 = (1 + p3e−p1µ 0 sdf iber₎ (µ 0 sdf iber)p2 p1+ (µ0sdf iber)p2 (2.3) Kanick et al. (2009a) derived an empirical relationship between the effective path length < L > and the combined effect of the absorption coefficient µa, and the

(10)

2.2. SINGLE FIBER REFLECTANCE MODEL CHAPTER 2. METHODS

reduced scattering coefficient, µ0_s:

< L >= Cpfdf iberp4 (µ0

s· df iber)p5(p6+ (µa· df iber)p6

(2.4) The effective path length becomes smaller with smaller detector-source separation, so it is dependent on the fiber diameter. The parameter set [p4, p5, p6] is fitted using Monte Carlo simulations, resulting in p4 = 1.54, p5 = 0.18 and p6 = 0.64

(Kanick et al., 2009b).

It was shown by Kanick et al. (2011b) that the absorption coefficients of the different chromophores can be obtained without prior knowledge of the tissue scat-tering properties if an optimized set of [p1, p2, p3, CP F] is used. This optimized

set was obtained using Monte-Carlo simulations, resulting in p1 = 6.82, p2 = 0.97,

p3 = 1.55, and CP F = 0.944.

The scattering coefficient µ0_sis expressed as a 4th power polynomial (van Leeuwen-van Zaane et al., 2013):

µ0_s = a1( λ λ0 )−1+ a2( λ λ0 )−2+ a3( λ λ0 )−3+ a4( λ λ0 )−4 (2.5)

λ0 is set at either 700 nm for the visible part of the spectrum or at 1000 nm

for the near-infrared part of the spectrum. The parameter set [a1, a2, a3, a4] is fitted by minimizing the residual error between measured and predicted reflectance spectrum.

The absorption coefficient µaof tissue is determined by the different chromophores

in the tissue. It is dependent on the volume fraction (fi) and the absorption of the

chromophore µa,i): µa= n X i=1 fiµa,i (2.6)

The absorption of the chromophores µa,i are known from the literature. The main

absorbers in the visible part of the spectrum are oxy- and deoxyhemoglobin and melanin, so µa can be expressed as:

µa,V IS = Ccorrbvf · (StO2 · µa,HbO2 + (1 − StO2) · µa,Hb) + mvf · µa,melanin (2.7)

where bvf is the blood volume fraction assuming the concentration of hemoglobin in whole blood to be 150 g/L, StO2is the blood oxygen saturation and µaHbO2, µaHband

µamelanin are the absorption coefficients of oxygenated hemoglobin, deoxygenated

hemoglobin and melanin respectively (van Leeuwen-van Zaane et al., 2013). The Ccorr is a correction factor for the inhomogenous distribution of blood in the tissue.

It is dependent on the vessel diameter because the blood is confined to this area. The correction factor is given by:

Ccorr =

1 − e−µa,V IS∗D

µa∗ D

(2.8) where D is the vessel diameter.

For the near-infrared part of the spectrum µa consists mainly of water, fat and

collagen and can be expressed as:

(11)

2.3. EXPERIMENTAL DESIGN CHAPTER 2. METHODS

with cwater, cf at and ccollagen the volume fractions and µa,water, µa,f at and µa,collagen

the absorption coefficients of water, fat and collagen respectively.

Limitations of this model will be described in section 4.3. In summary, when Rsf is measured, bvf , StO2, mvf , D, cwater, cf at and ccollagen can be fitted with the

equations described above.

2.3 Experimental design

Multiple types of measurements have been done on the outer forearm skin to see the variation in reflectance when deforming tissue, varying location, indentation and time of the measurement. These measurements are done to see the mechanism behind the change in optical properties when applying indentation. The optical probe was positioned at the upper part of the arm near the elbow of a volunteer. Water was applied on the arm as an index matching fluid. The fluid has to have a refractive index close to the one of the probe so the light does not refract as much as in air. This way the internal reflection of the light in the probe is lower.

Experiment 1: Time duration dependence of the

measure-ment

In the first experiment the reflectance was measured at the same indentation for 300 seconds. Indentation of 5 mm was applied, measuring the reflectance every ten seconds. The integration time of the VIS spectrograph was set at 40 ms averaging 250 times. The integration time of the NIR spectrograph was set at ten seconds without averaging. With this experiment the variation in time of the reflectance has been observed. It also showed the variation in the fitted parameters of the model.

Experiment 2: Constant indentation

In the second experiment the variation of the reflectance and therefore the different fitted parameters when releasing and reapplying indentation was measured. This was done by measuring the reflectance multiple times at the same indentation. The indentation was released and reapplied after waiting one minute for the tissue to recover. This procedure was done for 2 mm, 5 mm and 8 mm indentation. These measurements took approximately 20 seconds, and the probe was released after that. As an addition to this, another measurement was done were multiple spectra were taken at same indentation of 2 and 5 mm. Five spectra were measured at the same indentation before indentation was released and reapplied. These measurements took approximately 120 seconds. Again, the indentation was released and reapplied after waiting one minute for the tissue to recover. This procedure was repeated 5 times, so 25 spectra were taken at same indentation. The results of this set of measurements are more certain than the first set of measurements in this experiment but these also took longer to obtain, approximately 100 seconds. The results could differ from the first set because of the possible time dependence investigated in experiment 1.

(12)

2.3. EXPERIMENTAL DESIGN CHAPTER 2. METHODS

Experiment 3: Increasing indentation in steps

In the last measurement starting with gentle contact, the indentation of the probe was increased with steps of 0.5 mm. At every step, five spectra were measured. From this the standard deviation was calculated. Indentation was increased until it exceeded the comfort tolerance as indicated by the volunteer. In this experiment the changes in the tissue when increasing indentation can be observed.

(13)

Chapter 3 Results

3.1 Measured spectra

A typical example of the reflectance measured on the outer forearm is shown in figure 3.1. This particular measurement was made at 5 mm indentation. The average and standard deviation were calculated from five serial measurements. The blood absorption dip can be seen at 600 nm. It takes the shape of a W because of the presence of oxygenated hemoglobin. The water absorption dip can be seen at 1,400 nm. The reflectance is decreasing with longer wavelengths.

Figure 3.1: Reflectance measurement at 5 mm indentation with standard deviation

3.2 Experiment 1: Time dependence measurement

When measuring on the same spot for 300 seconds, the changes in reflectance has been observed. This leads to changes in the fitted values of blood volume fraction, vessel diameter, oxygen saturation and water volume fraction. These fitted values are shown in figure 3.2. The water volume fraction is observed to go down in time. Total decrease from t=0 to t=300 was 4%.

(14)

3.3. EXPERIMENT 2: VARIATION FITTED PARAMETERS AT SAME

INDENTATION CHAPTER 3. RESULTS

Figure 3.2: Blood volume fraction, vessel diameter, oxygen saturation and water volume fraction at constant indentation of 5 mm.

3.3 Experiment 2: Variation fitted parameters at

same indentation

When applying the same pressure multiple times, the variation in blood volume frac-tion, water volume fracfrac-tion, oxygen saturafrac-tion, collagen content and vessel diameter was determined. Also the changes in µ0_s were observed at different wavelengths. Re-sults are shown in table 3.1.

Table 3.1: Average fitted parameters with standard deviation at different indenta-tion levels. 2 mm 5 mm 8 mm bvf (1.4 ± 0.3) · 10−4 (6.8 ± 1.9) · 10−4 (2.9 ± 1.2) · 10−4 vessel diameter (µm) 13.1 ± 2.0 10 ± 5 1.7 ± 2.4 StO2 0.82 ± 0.14 0.78 ± 0.08 0.54 ± 0.08 collagen content 0.1255 ± 0.0015 0.120 ± 0.004 0.120 ± 0.003

water volume fraction 0.212 ± 0.009 0.173 ± 0.009 0.182 ± 0.012

µ0_s(500 nm) (mm−1) 0.1811 ± 0.0020 0.1936 ± 0.0022 0.200 ± 0.003 µ0_s(1100 nm) (mm−1) 1.730 ± 0.012 1.776 ± 0.006 0.182 ± 0.012

Individual parameters are plotted in figure 3.3. At higher indentation, the blood volume fraction is lower and the variation is small. The calculated vessel diame-ter showed larger spreads. The largest spread was observed at 5 mm; 4.9 - 17.1 µm. Oxygen saturation and water volume fraction variation was small, and values overlapped with different indentation.

(15)

3.3. EXPERIMENT 2: VARIATION FITTED PARAMETERS AT SAME

INDENTATION CHAPTER 3. RESULTS

Figure 3.3: Blood volume fraction, vessel diameter, oxygen saturation and water volume fraction at constant indentation of 2 mm, 5 mm or 8 mm.

The results of the second set of measurements of experiment 2 are shown in figure 3.4. The average of the five measurements at same indentation without releasing the pressure are shown together with the standard deviation.

Figure 3.4: Blood volume fraction, vessel diameter, oxygen saturation and water volume fraction at constant indentation of 2 mm and 5 mm with standard deviation.

(16)

3.4. EXPERIMENT 3: INCREASING INDENTATIONCHAPTER 3. RESULTS

3.4 Experiment 3: Increasing indentation

Results from a typical measurement 3 where the indentation was increased in steps of 0.5 mm are shown in fig 3.5.

Figure 3.5: Reflectance of arm measured with different indentations

The blood absorption dip was observed to disappear with increase of the inden-tation. The reflectance went up at all wavelengths with higher indeninden-tation. The water absorption dip at 1,400 to 1,500 nm was observed to decrease gradually with increasing indentation. This is visualized in figure 3.6. Here the reflectance at 0 mm is taken as a reference, and the ratio is calculated as:

Ratio (mm, λ) = reflection (mm, λ)

reflection (0, λ) (3.1)

Figure 3.6: Reflectance ratio of different indentation measured

The standard deviation of the reflectance measurements decreased when increas-ing the indentation from 0-10 mm.

(17)

In figure 3.7 the data and fit of the reflectance spectrum at 10 mm indentation are shown. The blue line is the SFR intensity without absorption (R0

SF).

Figure 3.7: Data and fit of the reflectance at 10 mm indentation.

The parameters bvf , vessel diameter, StO2 and cwater were extracted from the

(18)

Figure 3.8: Blood volume fraction, vessel diameter, oxygen saturation and water volume fraction increasing indentation.

The blood- and water content is often observed to decrease with increasing in-dentation. In the particular measurement shown in figure 3.8 water volume fraction decreased from 21% to 18%, oxygen saturation decreased from 80% to 10%. Col-lagen content did not follow a clear pattern (not shown). The other chromophore contents fat and melanin were approximately zero. Overall water volume fraction dropped with an average of 15%, and the blood volume fraction dropped with an average of 40%.

In figure 3.9 the water volume fraction and blood volume fraction are plotted together. From 0 to 9 mm indentation the blood volume fraction and water volume fraction are both decreasing. From 9 to 12.5 mm the water volume fraction is decreasing further, while blood volume fraction is not.

(19)

Figure 3.9: Water volume fraction versus blood volume fraction at different inden-tations

A linear correlation was found between the water volume fraction and the reduced scattering coefficient. This is shown in figure 3.10.

Figure 3.10: Water volume fraction versus reduced scattering coefficient at different indentations

(20)

Chapter 4 Discussion

4.1 Variation in measurements

The spectra showed significant changes with increasing indentation. This variation in results was also reported by Cugmas et al. (2013). They speculate that because of the higher penetration depth from the deformation, the light is gradually transferred to the deeper tissue layers.

Lim et al. (2011) reported that when inducing long-term or large pressure, the reflectance changed significantly. When applying small pressure (< 0.009N/m2_{) for}

a short time (< 2s) the spectra did not change significantly. This difference can be important depending on the type of investigation. When the physiological properties of tissue under pressure are investigated, long-term large pressure changes will give a better insight in the structure, identification and flexibility of tissue. Cugmas et al. (2013) found that pressure induced changes in tissue are site specific. Changes depended on whether the measurement was done on bone, muscle or veins.

However, when using this technique to differentiate between normal and diseased tissue, the changes under pressure will negatively influence the repeatability of the measurement. In this area, pressure should be applied for only a short period with small pressure. In this study a large variation in blood volume fraction, vessel diameter and oxygen saturation was found when no indentation was applied. This is caused by movement and changes in optical contact. An indentation of 5 mm showed less variation within these fitted parameters than smaller indentation values.

4.2 Observed trends in blood- and water content

In nearly all measurements blood was seen to be pushed out. The W-shaped hemoglobin dip between wavelengths of 500 - 600 nm of the spectrum in the re-flectance is disappearing with higher indentation. However, there were also excep-tions where the acquired spectra did not show any change in blood content, or the blood content fluctuated as shown in figure 4.1a.

(21)

4.3. LIMITATIONS OF THE SFR MODEL CHAPTER 4. DISCUSSION

Figure 4.1: Blood volume fraction and water volume fraction with increasing inden-tation for one particular measurement

In this measurement the blood volume fraction fluctuated between 7.0 ·10−4 and 1.8·10−3. These values are higher than the blood volume fraction measured in other measurements. So it is possible that the measurement was done on a bigger vein where blood was not pushed out.

Even though there was no clear trend seen in the blood volume fraction, the water content did have a clear trend, as is also shown in figure 4.1b. The dip in reflectance caused by water absorption around 1,450 nm is always observed to become smaller with higher indentation.

4.3 Limitations of the SFR model

4.3.1 Assumed shape of µ

0_s

In the SFR model a fourth order polynomial for µ0_s is assumed. This allows suf-ficient degrees of freedom to correct for the physically incorrect assumption that [p1, p2, p3, CP F] in equations 2.3 and 2.4 are independent of γ. The extracted values

of µ0_s are incorrect because of the correlation between γ and µ0_s.

The non-physical form of µ0_s is needed to correct for the assumption that the phase function is constant. The wavelength dependence of γ in tissue is not known, leading to this unfortunate fitting necessity. More details on this matter can be found in the work of Kanick et al. (2011b) and van Leeuwen-van Zaane et al. (2013). In further research Multi Diameter Single Fiber Reflectance could be used to determine the exact value of µ0_s to compare with the data found in this study.

4.3.2 Competition in the fitted parameters

When looking at the results from figure 3.3 it seems that the blood volume frac-tion, vessel diameter and oxygen saturation follow the same pattern. Since these parameters result from the same fit, they might compensate for each other, while they should be independent. A similar error could occur because of the competition between melanin and the reduced scattering coefficient as described in section 4.4.3.

(22)

4.4. COMPARISON OF FITTED VALUES OF CHROMOPHORESCHAPTER 4. DISCUSSION

4.4 Comparison of fitted values of chromophores

From the problems described in the previous paragraph, the fitting procedure might result in unrealistic values of the volume fraction of the chromophores. Water volume fraction and vessel diameter did yield realistic values. Brooks et al. (2015) found values in the same order of the vessel diameter. Oxygen saturation levels were measured to be very low. This can be true because under pressure cells might consume oxygen faster than it gets replenished. Brooks et al. (2015) found an oxygen saturation level around 30% for the arm while they did not measure under high indentation. Fat volume fraction, blood volume fraction and melanin volume fraction yielded less realistic values as will described below.

4.4.1 Fat volume fraction

In this study, the values for fat were in the order of 10−14. Since SFR only measures the superficial layers of tissue, it is possible there is no fat in this region. No literature has been found on single fiber reflectance spectroscopy performed in the NIR region to compare. In future research the fitting of the fat can be left out.

4.4.2 Blood volume fraction

Typical measured blood volume fractions were in the order of 0.001. The highest blood volume fraction measured was 0.0016. This is lower than expected from literature. Brooks et al. (2015) performed Multi-Diameter Single Fiber Reflectance Spectroscopy (MDSFR) on the arm and found a value around 0.01 for the blood volume fraction. However, in a particular group of volunteers the mean value of the blood volume fraction was 0.002.

The small number could be due to the superficial sampling volume of SFR and is also dependent on the measurement site. The reflectance spectrum on the outer forearm shows less blood absorption compared to the reflectance spectrum measured on the finger tip, lip or hand palm (not reported). Also, the model does not fit total blood content. It fits the oxygen saturation, calculated from oxygenated and deoxygenated hemoglobin; the main absorber in blood. If the concentration of hemoglobin in blood changes under pressure, the measured values might be off.

4.4.3 Melanin volume fraction

Difficulties aroused when trying to fit the melanin absorption spectrum. First, for the absorption coefficient of melanin at different wavelengths different values are found in the literature. Jacques (2013) plotted the different values for the absorption coefficient of melanosomes found in literature together, as shown in figure 4.2.

(23)

4.4. COMPARISON OF FITTED VALUES OF CHROMOPHORESCHAPTER 4. DISCUSSION

Figure 4.2: Absorption coefficient of melanosomes versus wavelength. Taken from Jacques (2013).

The fits for the absorption coefficient in these plots use a power curve: µa,melanosome= (519 cm−1)(

λ 500 nm)

−m

(4.1) The power factor m varies from literature as is shown in figure 4.2.

This assumed power curve for the absorption of melanin leads to a second fit-ting difficulty, namely that is has the same shape as the assumed curve of µ0_s (van Leeuwen-van Zaane et al., 2013):

µ0_s= b1(

λ λ0

)b2 _(4.2)

So the fitted values for the melanin volume fraction and µ0_s might be dependent on each other and there is competition within the fit. Future measurements on white skin without melanin could help verifying whether the melanin is independent of the reduced scattering coefficient. The measurements in this report were done on skin where melanin is visible, so it is not realistic to leave it out of the fit.

The exact values found for melanin volume fraction can not be compared to the literature, because in this model it is not the concentration of melanin that is extracted, but a volume fraction. Melanin is only present in the upper layer of the epidermis, which makes up only a small part of the sampling volume, so no melanin concentration could be determined. In future research, instead of using the absorption coefficient for melanin, the attenuation coefficient might be used, to derive a concentration of melanin.

(24)

4.5. REFRACTIVE INDEX MATCHING CHAPTER 4. DISCUSSION

4.5 Refractive index matching

When performing single fiber reflectance spectroscopy a refractive index matching fluid is applied between the sample and the probe to ensure optical contact. For the VIS part of the spectrum, water is a good refractive index matching fluid. It has a refractive index of 1.33 and does not absorb VIS light. However, for the NIR measurements water is less suitable because it has an absorption peak in this region. For this region the thickness of the layer between the sample and the probe influences the size of the water absorption dip (1,400 –1,500 nm). This might result in a higher water volume fraction. The water layer could also cause changes of the reflectance in this region during measurements because the water evaporates and gets absorbed by the skin. Gel for ultrasound measurements and fiber refractive matching oil was tested as an alternative to water, but they also had an absorption peak in the NIR region and from these fluids it is unknown if they affect the tissue optical properties. Using a fluid with an refractive index close to 1.33 without an absorption peak in the spectrum region 400 - 1,700 nm that does not change the optical properties of the skin could solve this problem. Such a fluid has not been found, this is why water was continued to be used. It is assumed that the thickness of the water layer for the reference measurement is the same as the layer of water on the tissue measurement.

(25)

Chapter 5 Conclusion

In this report the mechanism of tissue optical properties change induced by tissue deformation is investigated. It shows how much variation in the fitted parameters is possible when performing SFR on the outer forearm skin with different indenta-tion. The parameters fitted from the measurements showed variation dependent on indentation, location and time. Long-term and large indentation induce significant distortions in measured spectra.

When indentation increases the oxygen saturation level and blood volume frac-tion is observed to decrease. Water content is also observed to decrease, even when no blood seems to be present anymore. When increasing indentation from 0 to 10 mm water content dropped with an average of 15%, and the blood volume fraction dropped with an average of 40%.

A correlation has been found between the changes in the scattering and the changes in water volume fraction. It shows that because of the decrease in water content caused by indentation the structure and composition of tissue changes. The tissue becomes more dense with scattering particles which results in an increase of the reduced scattering coefficient.

In further research new ways of taking a reference measurement without using water should be investigated. Better calibration and higher repeatability is needed when using this technique in medical studies. In further research Multi Diameter Single Fiber Reflectance could also be used to determine the µ0_s more accurately to compare with the data found in this study.

(26)

Bibliography

Amelink, A., Christiaanse, T., and Sterenborg, H. J. (2009). Effect of hemoglobin extinction spectra on optical spectroscopic measurements of blood oxygen satu-ration. Optics letters, 34(10):1525–1527.

Atencio, J. D., Guill´en, E. O., y Montiel, S. V., Rodr´ıguez, M. C., Ramos, J. C., Guti´errez, J., and Mart´ınez, F. (2009). Influence of probe pressure on human skin diffuse reflectance spectroscopy measurements. Optical Memory and Neural Networks, 18(1):6–14.

Brooks, S., Hoy, C. L., Amelink, A., Robinson, D. J., and Nijsten, T. E. (2015). Sources of variability in the quantification of tissue optical properties by multidi-ameter single-fiber reflectance and fluorescence spectroscopy. Journal of biomed-ical optics, 20(5):057002–057002.

Cerussi, A., Siavoshi, S., Durkin, A., Chen, C., Tanamai, W., Hsiang, D., and Tromberg, B. J. (2009). Effect of contact force on breast tissue optical property measurements using a broadband diffuse optical spectroscopy handheld probe. Applied optics, 48(21):4270–4277.

Chen, W., Liu, R., Xu, K., and Wang, R. K. (2005). Influence of contact state on nir diffuse reflectance spectroscopy in vivo. Journal of Physics D: Applied Physics, 38(15):2691.

Cugmas, B., B¨urmen, M., Bregar, M., Pernuˇs, F., and Likar, B. (2013). Pressure-induced near infrared spectra response as a valuable source of information for soft tissue classification. Journal of biomedical optics, 18(4):047002–047002.

Gamm, U., Kanick, S., Sterenborg, H., Robinson, D., and Amelink, A. (2011). Measurement of tissue scattering properties using multi-diameter single fiber re-flectance spectroscopy: in silico sensitivity analysis. Biomedical optics express, 2(11):3150–3166.

Gamm, U., Kanick, S., Sterenborg, H., Robinson, D., and Amelink, A. (2012). Quantification of the reduced scattering coefficient and phase-function-dependent parameter γ of turbid media using multidiameter single fiber reflectance spec-troscopy: experimental validation. Optics letters, 37(11):1838–1840.

Gamm, U. A. (2013). Quantification of Tissue Scattering Properties by Use of Fiber Optic Spectroscopy.

Jacques, S. L. (2013). Optical properties of biological tissues: a review. Physics in medicine and biology, 58(11):R37.

(27)

BIBLIOGRAPHY BIBLIOGRAPHY

Kanick, S., Sterenborg, H., and Amelink, A. (2009a). Empirical model of the photon path length for a single fiber reflectance spectroscopy device. Optics express, 17(2):860–871.

Kanick, S. C., Gamm, U. A., Sterenborg, H. J., Robinson, D. J., and Amelink, A. (2011a). Method to quantitatively estimate wavelength-dependent scattering properties from multidiameter single fiber reflectance spectra measured in a turbid medium. Optics letters, 36(15):2997–2999.

Kanick, S. C., Robinson, D. J., Sterenborg, H., and Amelink, A. (2009b). Monte carlo analysis of single fiber reflectance spectroscopy: photon path length and sampling depth. Physics in medicine and biology, 54(22):6991.

Kanick, S. C., Robinson, D. J., Sterenborg, H. J., and Amelink, A. (2011b). Method to quantitate absorption coefficients from single fiber reflectance spectra without knowledge of the scattering properties. Optics letters, 36(15):2791–2793.

Lim, L., Nichols, B., Rajaram, N., and Tunnell, J. W. (2011). Probe pressure effects on human skin diffuse reflectance and fluorescence spectroscopy measurements. Journal of biomedical optics, 16(1):011012–011012.

Sokolov, K., Follen, M., and Richards-Kortum, R. (2002). Optical spectroscopy for detection of neoplasia. Current opinion in chemical biology, 6(5):651–658.

van Leeuwen-van Zaane, F., Gamm, U., van Driel, P., Snoeks, T., de Bruijn, H., Mol, I., L¨owik, C., Sterenborg, H., Amelink, A., Robinson, D., et al. (2013). In vivo quantification of the scattering properties of tissue using multi-diameter single fiber reflectance spectroscopy. Biomedical optics express, 4(5):696–708. Zonios, G. and Dimou, A. (2006). Modeling diffuse reflectance from semi-infinite

turbid media: application to the study of skin optical properties. Optics express, 14(19):8661–8674.

The mechanism of tissue optical properties change induced by tissue deformation

Joy Willemse

10411070