2.1.

Tissue Optical Clearing Concept

During short-time OC treatments, the physical dimensions of tissue scatterers [scatterer radius $a$, in Eq. (1)] and their RI ($n_s$) are not expected to change. Consequently, as the partial replacement of interstitial water by the OCA will produce some volume changes in the sample, ${\rho }_s$ will suffer a variation. For a slab-form sample that has a small thickness relative to its superficial area, the time dependence of ${\rho }_s$ [or scatterer volume fraction–$f_s(t)$] can be calculated from the time dependence of sample thickness [$d(t)$]. By replacing ${\rho }_s$ by $f_s(t)$ in Eq. (1), the time dependence for ${\mu }_s^{\prime }$ can be calculated, provided that we can obtain the time dependence for ${\bar{n}}_0$.

If total and mobile water contents in the natural tissue are known, the variation of ${\bar{n}}_0$ can be calculated during treatment according to Eq. (3):³⁰

When spectral data are available for ${\mu }_s^{\prime }$ [${\mu }_s^{\prime }(\lambda )$] [from inverse adding doubling (IAD) method applied to integrating sphere measurements of tissue sample optical properties], for the RI of scatterers [$n_s(\lambda )$] and for the RI of ISF [$n_0(\lambda )$], it is possible to try some $a$ values in Eq. (2) to match data for ${\mu }_s^{\prime }$_.⁹ Once the correct $a$ value that allows such matching is obtained, the same equation can be used to calculate the time dependence for ${\mu }_s^{\prime }$, provided that independent determination of the time dependencies for ${\bar{n}}_0$ and $f_s$ was made. If $f_s(t=0)$, $f_0(t=0)$ and the sample volume of natural tissue are known, $f_s(t)$, $f_0(t)$ can be calculated from $d(t)$. ${\overline{n}}_0(t)$, on the other hand, can be calculated with Eqs. (3) and (4). Once ${\mu }_s^{\prime }(t)$ is calculated with Eq. (2) and ${\mu }_s(t)$ is calculated with Eq. (4), they can be used in Eq. (6) to obtain $g(t)$:⁹

2.2.

Tissue Samples

Due to the high incidence in human populations worldwide, colorectal cancer is a major health concern.³⁵ At an early stage of development, adenomatous polyps are formed in the inner layer of the colorectal wall, the mucosa—see Fig. 1.

Fig. 1

(a) Histological cross section of human colorectal wall showing distinct layers and (b) the gross features of invasive colorectal adenocarcinoma (left) and adjacent mucosa viewed in a surgical specimen.

If not removed at an early stage, these polyps will evolve into adenocarcinoma that intrudes the other colorectal wall layers, first the submucosa, followed by the muscularis propria, and eventually reaching the peritoneal lining or adjacent organs.³⁶ Early stage detection and removal of colorectal polyps is very important and since they start developing in the mucosa, noninvasive endoscopic optical techniques are highly desired. Due to the distinct tissue layers in the colorectal wall, which have different optical properties, OC treatments might be a strong option to improve the development of such techniques for early or later stage polyp removal.

All normal and pathological tissue samples used in the present study were obtained from adult patients undergoing colorectal surgery at Portuguese Oncology Institute of Porto, Portugal. A signed consent allowing for the subsequent use of surgical specimens for diagnostic and research purposes was obtained previously to surgical procedures from all patients. This procedure has been approved by the Ethics Committee of Portuguese Oncology Institute of Porto.

After collecting the samples from surgical resection specimens, normal and pathological colorectal mucosa samples were separated and preserved frozen at −80°C for a period of 12 h. A cryostat from Thermo Scientific^™ (Waltham, Massachusetts), model Micron HM 550 was used to prepare samples to use in study with a circular form ($\phi =1\mathrm{cm}$) and 0.5-mm thickness. To mimic natural hydration in tissue samples, they were kept in saline for 10 min before initiating studies. Both types of tissues were prepared and processed according to this standard protocol, meaning that any discrepancy related to the freezing procedure should be minimal.

A total of 16 normal and 16 pathological samples were used in measurements described in Secs. 2.3 and 2.4.

2.3.

Total Water Content Evaluation

To obtain the total water content in normal and pathological tissues, we have performed a dehydration study, where 10 samples of each type were used. At the beginning of this study, all samples were weighted, before being placed inside a Memmert drying oven, model ULM 500 (Schwaback, Germany) for 24 h at a temperature of $103°\mathrm{C}\pm 2°\mathrm{C}$. The weight scale used was a Kern, model ALJ 220-4NM (Balingen, Germany) with a precision of $\pm 0.0001\mathrm{g}$. After total dehydration, the samples were placed inside a desiccator for 15 min to remove any residual humidity at room temperature. Finally, the samples were weighted again. The average temperature for weight measurements before and after dehydration was $22°\mathrm{C}\pm 2°\mathrm{C}$.

As the initial volume of the samples was known before dehydration ($V=\pi \times {0.5}^2\times 0.05{\mathrm{cm}}^3$), and considering that water has a density of $1\mathrm{g}/{\mathrm{cm}}^3$ at room temperature, the weight difference for each sample allowed us to calculate total water content. After averaging data from 10 normal and from 10 pathological tissue samples, we obtained a total water content of $73.96\%\pm 0.02\%$ for normal mucosa and $73.33\%\pm 0.10\%$ for pathological mucosa.

2.4.

Thickness and $T_{\mathrm{c}}$ Measurements

From our previous studies,¹⁹ we have estimated the mobile water content in normal and pathological colorectal mucosa as 59.4% and 64.4%, respectively. Comparing between the total and mobile water contents of normal and pathological mucosa tissues, we see a decrease of 0.63% in total water and an increase of 5% in mobile water from normal to pathological. According to the literature,³⁷ bound water can be converted into mobile water when cancer cells invade the tissue. The significant increase in mobile water and some total water loss from normal to pathological mucosa was already seen for other tissues, as demonstrated by Fig. 7 of Ref. 38. The 5% difference in mobile water of mucosa tissues means that a glucose–water solution that will produce a smooth $T_{\mathrm{c}}$ time dependence during a 30 min OC treatment needs to have a glucose concentration of $\sim 35\%$ (for pathological mucosa) and $\sim 40\%$ (for normal mucosa). Thickness and $T_{\mathrm{c}}$ measurements were previously performed for these particular treatments at a mean temperature of $20°\mathrm{C}\pm 0.5°\mathrm{C}$,¹⁹ and the mean time dependencies resulting from those measurements were used in calculations performed in the present study. For thickness measurements, we used three normal and three pathological samples and for the $T_{\mathrm{c}}$ measurements the same number of samples was used to average final results. Figure 2(a) shows the setup used to measure sample thickness and Fig. 2(b) presents the $T_{\mathrm{c}}$ setup.

Fig. 2

Measurement setups for (a) thickness and (b) $T_{\mathrm{c}}$.

Considering the thickness measurements, the sample is placed inside two microscope glasses (thickness $d_{\mathrm{G}}=1\mathrm{mm}$) and a digital micrometer from Mitutoyo^™ (Japan), model 293 MDC-MX lite with a precision of 0.001 mm, measures the global thickness ($D$).¹⁹ The OC solution is injected in-between the glasses to begin treatment, and measurements are taken at every 15 s within the first 2 min and at every min after that. As the solution presents some viscosity, it stays within the glasses during all treatment, leading to a constant glucose flow into the tissue. At the end, the time dependence for sample thickness is obtained as the difference between the measurements and the fixed thickness of the glasses: $d(t)=D(t)-(2\times d_G)$. Three studies were made for the treatment of pathological tissues with 35% of glucose and the other three for the treatment of normal tissues with 40% of glucose. Mean thickness time dependence was calculated for each case. These results are presented in Ref. 19.

Considering $T_{\mathrm{c}}$ measurements, an Avalight-D(H)-S lamp from Avantes^™ (The Netherlands) was used to irradiate the sample through an optical fiber cable and collimating lens. Light enters the sample cuvette [rectangular-box in Fig. 2(b)] through a glass at the bottom. The sample is fixed at the center of the cuvette by some wires, which are shown in Fig. 2(b) as black bars. A set of pinholes [not represented in Fig. 2(b)] guarantees a beam diameter of 1 mm between the illuminating and collecting optics. At the top of the cuvette, a second optical fiber cable and collimating lens collect the transmitted beam for delivery to the spectrometer. The $T_{\mathrm{c}}$ spectrum of the natural sample is measured first. Then, the OC solution [represented as small dots in the cuvette of Fig. 2(b)] is injected into the cuvette by a syringe through a lateral hole. Spectra are measured during treatment at every 5 s for a period of 30 min. Three sets of $T_{\mathrm{c}}$ studies were made for the treatment of pathological mucosa with 35% of glucose and the other three were made for the treatment of normal mucosa with 40% of glucose. Mean time dependencies for $T_{\mathrm{c}}$ spectra were calculated in each case and can be found in Ref. 19.

2.5.

Reconstruction of ${\mu }_s^{\prime }$ for Natural Tissue

To obtain the time dependencies for the optical properties of normal and pathological colorectal mucosa with accuracy, a complete study with different types of measurements should be made during treatments. Such study is complex and it involves a large number of tissue samples (normal and pathological) to be studied under different measurement procedures. As explained above when we introduced Eq. (2), such an equation was validated only for specific conditions and for samples whose scatterers are spherical. This is definitely not the case for the tissues we used in this study as it has been reported that normal and pathological colorectal mucosa contains scatterers–like collagen fibrils and myofibroblasts in stroma,³⁹ and different types of proteins,⁴⁰ such as glycoproteins.⁴¹As we wanted to test Eq. (2) outside the limits where it was validated, to confirm the theoretical predictions of scattering coefficients decrease and anisotropy increase and possibly differentiate between treatment response of normal and pathological tissues, we had to select an estimation method. In the absence of a more accurate model and to avoid a complex experimental study to estimate the time dependencies for the optical properties of human colorectal mucosa, we selected to perform the calculations based on Eq. (2).

The first calculation consisted of reconstructing the wavelength dependence for ${\mu }_s^{\prime }$ through Eq. (2) with the objective of estimating $a$ for each tissue. In this calculation, we are able to determine also the correct ${\bar{n}}_0(\lambda )$ and $n_s(\lambda )$ for the natural tissues, which are necessary for further calculations. The following reconstruction procedure was used for both tissues.

Considering as reference the ${\mu }_s^{\prime }(\lambda )$ data that were previously estimated via IAD simulations⁴² for natural tissues (presented in Ref. 43), we performed the following steps:

The above-described procedure was made by trial and error until the reconstruction of ${\mu }_s^{\prime }(\lambda )$ was obtained correctly. In the first trials, the calculated ${\mu }_s^{\prime }(\lambda )$ was too far from the values estimated through IAD simulations. To correct this, we had to adjust the VFs in the calculation of the dispersions for scatterers and ISF (steps II and III above). These adjustments also produce a change in $m(\lambda )$. When the calculated ${\mu }_s^{\prime }$ ($\lambda$) was approaching the IAD estimated data, we changed $a$ for a better matching.

The estimated ${\mu }_s^{\prime }(\lambda )$, ${\overline{n}}_0(\lambda )$, $n_s(\lambda )$, corresponding VFs for scatterers and ISF in natural tissue and the $a$ value were considered in further calculations described in the following subsections.

2.6.

Calculation of Kinetics for μ_s, ${\bar{n}}_0$, and n_tissue

To calculate the kinetics for ${\mu }_s$, we needed the wavelength dependence for ${\mu }_a$, which is presented in Ref. 43 for both tissues. Considering this data and the $T_{\mathrm{c}}$ measurements made from natural tissues and during treatments, we used Eq. (4) to obtain ${\mu }_s(\lambda ,t)$.

Once these data were calculated, we have used Eq. (3) to calculate ${\bar{n}}_0(\lambda ,t)$. In this calculation, the mean $d(t)$ measurements obtained for each tissue/treatment were also used.

To calculate the time dependence for the RI of the whole tissue [$n_{\text{tissue}}(\lambda ,t)$] with Eq. (5), we needed $f_s(t)$ and $f_0(t)$ in each tissue. Considering the thickness measurements made from normal and pathological tissues during corresponding treatments, these time dependencies were obtained as follows:

2.7.

Calculation of Kinetics for ${\mu }_s^{\prime }$ and g

To calculate the kinetics for ${\mu }_s^{\prime }$, we used Eq. (9), but now considering the time dependence for the VF of scatterers, $f_s(t)$, instead of its value for natural tissue. Furthermore, instead of using ${\bar{n}}_0(\lambda )$ and $m(\lambda )$ for the natural tissues, we used their time dependencies, ${\bar{n}}_0(\lambda ,t)$ and $m(\lambda ,t)$, as calculated with the procedure described in Sec. 2.6.

Once the kinetics of ${\mu }_s^{\prime }$ was obtained, we used Eq. (6) to calculate the kinetics for $g(\lambda )$.

All the results of these calculations for normal and pathological tissues are presented in Sec. 3.

3.1.

Data Estimation for Natural Tissue

To calculate the kinetics for the optical properties of tissues under study, we needed to start by estimating the spectral data for the RI of scatterers and ISF and the mean scatterer radius. To obtain these data, we have initiated our calculations by reconstructing ${\mu }_s^{\prime }(\lambda )$, as indicated in Sec. 2.5. Such reconstruction was made by adjusting $n_s(\lambda )$, ${\overline{n}}_0(\lambda )$, and $a$. The dispersions that provided the best reconstruction of ${\mu }_s^{\prime }(\lambda )$ for both tissues are shown in Fig. 3.

Fig. 3

Dispersions for water,⁴⁵ colorectal mucosa, its scatterers and ISF at 20°C. (a) Normal mucosa and (b) pathological mucosa.

The dispersions obtained for normal mucosa that is shown in Fig. 3(a) are described by the following equations [whole tissue—Eq. (13), ISF—Eq. (14), and scatterers—Eq. (15)]:

Fig. 4

Relative index for natural tissue. (a) Normal mucosa and (b) pathological mucosa.

Using all the calculated dispersions for scatterers and ISF and the VF of scatterers in Eq. (9), we obtained a good reconstruction for ${\mu }_s^{\prime }(\lambda )$ data of both tissues, by considering the correspondent scatterer radius, $a$, as: 46-nm (normal mucosa) and 158-nm (pathological mucosa).

Considering that collagen fibrils are the main scatterers in colorectal mucosa stroma underlying the epithelium,⁴⁶ we performed a simple calculation to check the accuracy of these results. Collagen fibrils have a cylinder-shape, and making length and diameter measurements from electronic micrographs of normal mucosa tissue in Ref. 47, we estimated the mean fibril length as $0.119\mu \mathrm{m}$ and the mean radius as $0.039\mu \mathrm{m}$. Considering these values, the mean volume can be calculated:

Using this volume, we can calculate the mean radius for an equivalent sphere model as

Comparing this value with the mean estimated radius for normal mucosa (46 nm), we see that calculations using Eq. (2) are not precise, but they show that the spherical scatterer model is not too bad to approximate real fibril scatterers in tissues. The values measured from Ref. 47 refer to collagen fibrils in oral mucosa, but smaller collagen diameters can be found in other tissues like in Fig. 3 of Ref. 48 for the eye sclera.

Regarding pathological tissues, we could not compare our estimated sizing data with real data from the literature, but as indicated by some authors,^49–51 when cancer invades mucosal stroma, collagen fibrils increase in the length, diameter, and density. This means that the estimations we made with Eq. (2) are sensitive to the mean scatterer size increase from normal to pathological tissue.

Previously estimated ${\mu }_a(\lambda )$ and ${\mu }_s^{\prime }(\lambda )$ data and reconstructed ${\mu }_s^{\prime }(\lambda )$ curves with Eq. (9) are shown in Fig. 5.

Fig. 5

Estimated ${\mu }_a(\lambda )$ data for (a) normal mucosa and (b) pathological mucosa. Estimated and calculated ${\mu }_s^{\prime }(\lambda )$ data for (c) normal mucosa and (d) pathological mucosa.

From Fig. 5, we see similar wavelength dependencies for normal and pathological tissues but with pathological tissues showing more blood content. From graphs in Figs. 5(c) and 5(d), we see that the calculated curves provide good fittings of the data previously estimated with IAD simulations. For the case of pathological mucosa, some estimated points between 500 and 600 nm are not in agreement with the calculated curve, meaning that some cross talk between absorption and scattering near hemoglobin bands changes the smooth decreasing behavior as calculated with Eq. (9). We do not see significant evidence of such behavior in normal mucosa, indicating that pathological tissues have higher blood content, which is accordingly to the reddish color seen in Fig. 1(b) for the adenocarcinoma. A different equation that accounts for the Rayleigh and Mie scattering terms can be used to describe the curves shown in graphs of Figs. 5(c) and 5(d), respectively. Such equation is described by Steven Jacques in his paper from 2013,⁵² and if we fit the calculated data (curves) in graphs of Figs. 5(c) and 5(d) with that equation, we obtain optimized fitting ($R^2=1$) for both types of tissues:

3.2.

Kinetics for ${\mu }_s(\lambda )$, ${\bar{n}}_0$, and n_tissue

Using the procedure outlined in Sec. 2.6, we started by calculating the time dependence for the VFs of ISF and scatterers. Figure 6 presents these variations for both tissues.

Fig. 6

Time dependence for the VFs of colorectal mucosa during treatment. (a) Normal mucosa treated with 40% of glucose and (b) pathological mucosa treated with 35% of glucose.

Fig. 7

Kinetics of ${\mu }_s$ for (a) normal mucosa and (b) pathological mucosa.

Graphs in Fig. 6 show a fast and strong variation in VFs for both tissues within the first two minutes, indicating the occurrence of the dehydration mechanism, which dominates at early stage treatment. After 3 min, VFs show smooth and low-magnitude variations, meaning that they are now driven by the RI matching mechanism that occurs in the interstitial locations. Figure 6 shows evidence of the two OC mechanisms and give an idea of their duration. Similar behavior is observed for both mucosa tissues, but the magnitude of the variations that correspond to the dehydration mechanism is higher for pathological mucosa.

Using ${\mu }_a(\lambda ,t=0)$, $T_c(\lambda ,t)$, and $d(t)$ data in Eq. (4), we calculated the time dependencies for ${\mu }_s(\lambda )$ in both tissues. Those graphs are shown in Fig. 7.

In this case, we see that both tissues present a decreasing behavior over the time of treatment as expected for OC treatments. There is some transient decreasing behavior within the first seconds of treatment for long wavelengths in both tissues, possibly due to interaction of OCA with superficial layers of tissues. After that and within the first 2 min, both tissues show an increase in ${\mu }_s$, which indicates the approximation of scatterers due to the loss of water in the interstitial locations. Such increase seems to have higher magnitude near the absorption bands of hemoglobin and in particular for pathological mucosa. The water and blood contents in natural tissues originate these differences between normal and pathological tissues for the kinetics of ${\mu }_s$: natural pathological mucosa has less water content than normal mucosa (see Fig. 3) and higher blood content than normal mucosa (see Fig. 5).

Regarding the subsequent smooth decrease, both types of tissues present similar behavior, where the longer wavelengths show high magnitude decrease. For smaller wavelengths, the decrease in ${\mu }_s$ is modest due to the presence of the absorption bands of blood. The data shown in graphs of Fig. 7 indicate that the simple model we used confirms the expectation of a global decreasing in ${\mu }_s$ during OC treatments. It shows also that normal and pathological tissues have similar variations in ${\mu }_s$ during treatment.

Using the obtained ${\mu }_s$ data in Eq. (3), we calculated the kinetics of ${\bar{n}}_0$ for both tissues and combining that data with the dispersion of scatterers in Eq. (5), we calculated the kinetics for tissue RI for both tissues. Figure 8 shows these graphs.

Fig. 8

Kinetics of ${\bar{n}}_0$ for (a) normal mucosa and (b) pathological mucosa, and kinetics of $n_{\text{tissue}}$ for (c) normal mucosa and (d) pathological mucosa.

As expected, the RI of ISF increases with treatment, showing the evidence of the RI matching mechanism. For the case of normal mucosa, we see a smooth behavior for all wavelengths and during all treatment. For pathological mucosa, the overall effect and magnitude of the increase are similar to normal mucosa, but a significant change of form in dispersion is seen for lower wavelengths at the beginning of treatment.

The graphs for $n_{\text{tissue}}$ in Figs. 8(c) and 8(d) present three stages: within the first min, a strong increase in $n_{\text{tissue}}$ corresponds to the dehydration mechanism; a transition between mechanisms is seen in the following 3 min; and the RI matching mechanism dominates over the following 26 min. Although the overall behavior is similar between tissues, for the case of pathological mucosa we see a higher magnitude increase at the beginning, which is conditioned by the water content in natural pathological mucosa. As the water content in untreated pathological mucosa is smaller than in untreated normal mucosa (see Fig. 3), it takes less time for the dehydration mechanism to occur in pathological tissues.

3.3.

Kinetics for ${\mu }_s^{\prime }(\lambda )$ and g

According to the procedure described in Sec. 2.7, we first calculated the time dependencies for the relative index of refraction for both tissues. Then using these data and Eq. (9), we calculated the time dependencies for ${\mu }_s^{\prime }(\lambda )$. Combining these data with the data in Fig. 7 by using Eq. (6), we calculated the time dependencies for $g$-factor for both tissues. Figure 9 shows all these graphs for both tissues.

Fig. 9

Kinetics of $m$ for (a) normal mucosa and (b) pathological mucosa; kinetics of ${\mu }_s^{\prime }$ for (c) normal mucosa and (d) pathological mucosa; kinetics of $g$ for (e) normal mucosa and (f) pathological mucosa.

Considering graphs (a) and (b) of Fig. 9, we see that the relative index of refraction shows similar decreasing behavior for both tissues and a higher magnitude decrease is observed for pathological mucosa. The kinetics of ${\mu }_s^{\prime }(\lambda )$ presented in graphs (c) and (d) of Fig. 9 show a smoother behavior than the one observed for ${\mu }_s(\lambda )$ (see Fig. 7), more similar to the kinetics of $m$, and no significant differences are seen between normal and pathological mucosa in this case. To show this decreasing behavior over the time of treatment, we had to select the presented perspective in graphs (c) and (d) of Fig. 9, which does not show a slight increase in ${\mu }_s^{\prime }(\lambda )$ from $t=0$ to $t=15\mathrm{s}$.

Graphs (e) and (f) of Fig. 9 plot a change in the form of $g(\lambda )$ at the beginning of treatment for both tissues, where longer wavelengths show a significant decrease for normal mucosa. We have seen similar transient behavior for ${\mu }_s$ in Fig. 7, and it is possibly related to the early interaction of glucose molecules with superficial tissue layers. Immediately after, we see a strong and fast increase. As water loss leads to a better packing inside the tissue, forward scattering increases.

Due to the transition between dehydration and RI matching mechanisms, we see a small decrease in $g$ after 2 min for both tissues. Such decrease is originated by the separation of scatterers as glucose molecules place themselves in the interstitial locations.⁹

Using confocal microscopy, a group from the Department of Dermatology of the Oregon Health and Science University has reported in 2010 the increase in $g$ during optical clearing of skin,⁵³ which accords with our calculations. In that same study, no significant changes in ${\mu }_s$ were observed. This study regards the treatment of skin samples with 80% of DMSO, and pure glycerin solutions and measurements were made only before and after 1 h treatment. Considering their results and the ones from the present study, we must consider that skin is a multilayered tissue with a set of optical properties per layer, whereas colorectal mucosa has only a set of optical properties. Another major difference that might lead to insignificant change in ${\mu }_s$ in their study is that they used oversaturated solutions. As we have demonstrated in a previous study,²⁴ when the water in the treating solution matches the mobile water in the tissue (intermediate OCA concentrations in solution), only the RI matching mechanism occurs, as minimal or no water flux is observed. This is approximately the case for the treatments presented here—normal mucosa treated with 40% of glucose and pathological mucosa treated with 35% of glucose.¹⁹ Under these conditions, the unique OCA flux into the interstitial locations of the tissue is smooth and continuous. Such process leads to a reduction of ${\mu }_s$, through the RI matching mechanism and an increase in the $g$-factor as our calculations demonstrate. When oversaturated solutions are used, as in the case of the study presented in Ref. 53, the dehydration mechanism will dominate OC during early stage treatment, as we have also previously demonstrated.²⁴ During this early stage dehydration, the water loss by the tissue forces scatterers to approach each other and to form better packing (scatterers change their random orientations into the same direction). Such scatterer alignment leads to an increase in the $g$-factor. When treating normal and pathological colorectal mucosa samples with 54% of glucose,¹⁹ we observed an initial $T_{\mathrm{c}}$ increase, followed by a decrease. These $T_{\mathrm{c}}$ variations indicate that the glucose molecules initially diffuse into the interstitial space of the tissues and then into the tissue cells. OCA diffusion into tissue cells increases ${\mu }_s$ due to scatterer size increase. For a long treatment with oversaturated solutions, as in the case of Ref. 53, it is expected that the overall variations are some increase in $g$-factor (due to better alignment of scatterers) and minimal change in ${\mu }_s$ (due to mutual compensation of RI matching effect and increased scattering efficiency caused by scatterer density elevation).