2.1.

Relationship Between RGB Values and Skin Chromophore Concentrations

RGB values of a pixel on a skin surface image acquired by a digital camera can be expressed as

${\mathbf{L}}_{\mathbf{1}}$ is a transformation matrix to convert $XYZ$ values to the corresponding RGB values and exists for each working space (NTSC, PAL/SECAM, sRGB, etc.). In addition, $\lambda$, $E(\lambda )$, and $O(\lambda )$ are the wavelength, the spectral distribution of the illuminant, and the diffuse reflectance spectrum of human skin, respectively, and $\overline{x}(\lambda )$, $\overline{y}(\lambda )$, and $\overline{z}(\lambda )$ are the color matching functions in the CIEXYZ color system. The value of constant $k$ that results in $Y$ being equal to 100 for the perfect diffuser is given by

In Eqs. (2) through (5), the summation can be carried out using data at 10-nm intervals, from 400 to 700 nm. Assuming that the skin tissue consists primarily of the stratum corneum, epidermis containing melanin, and dermis containing oxygenated and deoxygenated blood, the diffuse reflectance of skin tissue $O$ can be expressed as

2.2.

Estimation of Skin Chromophore Concentrations Based on RGB Image

Figure 1 shows the flow of estimation using the proposed method. The proposed method means a solution of the inverse problem to deduce $C_m$, $C_{\mathrm{ob}}$, and $C_{\mathrm{db}}$ from the measured RGB values. The way to solve this is by transforming the measured RGB values to $XYZ$ values with the matrix ${\mathbf{N}}_{\mathbf{1}}$ and assumes a linear relation between $XYZ$ values and $C_m$, $C_{\mathrm{ob}}$, and $C_{\mathrm{db}}$. The linear terms define the matrix ${\mathbf{N}}_{\mathbf{2}}$. First, RGB values in each pixel of the image are transformed into $XYZ$ values by a matrix ${\mathbf{N}}_{\mathbf{1}}$ as

Fig. 1

Flow of estimation process for $C_m$, $C_{\mathrm{ob}}$, $C_{\mathrm{db}}$, and $C_{\mathrm{tb}}$.

We determined the matrix ${\mathbf{N}}_{\mathbf{1}}$ based on measurements of a standard color chart (ColorChecker, $X$-Rite Incorporated, Michigan) that has 24 color chips and is supplied with data for the CIEXYZ values for each chip under specific illuminations and corresponding reflectance spectra. To determine the matrix ${\mathbf{N}}_{\mathbf{2}}$, we calculated 300 diffuse reflectance spectra $O(\lambda )$ in a wavelength range of from 400 to 700 nm at intervals of 10 nm by MCS for light transport³⁴ in skin tissue. We used the skin baseline absorption coefficient³⁵ for that of the stratum corneum. The absorption coefficient of the epidermis depends on the volume concentration of melanin in the epidermis $C_m$. We used the absorption coefficient of melanosome given in the literature³⁶ as the absorption coefficient of melanin ${\mu }_{a,m}$. This corresponds to the absorption coefficient of the epidermis for the case in which $C_m=100\%$. We subsequently derived the absorption coefficients of the epidermis for 10 different lower concentrations of $C_m=1$ to 10% at intervals of 1%, by simply proportioning it to that for $C_m=100\%$, and the absorption coefficients were input for the epidermis. The sum of the absorption coefficient of oxygenated blood for $C_{\mathrm{ob}}$ and that of deoxygenated blood for $C_{\mathrm{db}}$ were considered for the dermis. This summation provides the total blood concentration $C_{\mathrm{tb}}=C_{\mathrm{ob}}+C_{\mathrm{db}}$ and oxygen saturation ${\mathrm{SO}}_2\%=(C_{\mathrm{ob}}/C_{\mathrm{tb}})\times 100$. The absorption coefficients of blood having a 44% hematocrit with $150\mathrm{g}/\text{liter}$ of hemoglobin³⁷ were assumed to be that of the dermis for the case in which $C_{\mathrm{tb}}=100\%$ and were input for the dermis as ${\mu }_{a,\mathrm{ob}}+{\mu }_{a,\mathrm{db}}$. Then the absorption coefficients of the dermis were derived for five different concentrations of $C_{\mathrm{tb}}=0.2$, 0.4, 0.6, 0.8, and 1.0% for six different cases of ${\mathrm{SO}}_2=0$, 20, 40, 60, 80, and 100%. Typical published values for ${\mu }_s(\lambda )$³⁸ and $g(\lambda )$³⁹ were input for the stratum corneum, epidermis, and dermis, which are provided as a function of wavelength. The layer thicknesses of the stratum corneum, epidermis, and dermis were set to be 0.02, 0.06, and 4.92 mm, respectively. The refractive index of the stratum corneum was set to be 1.47.⁴⁰ The refractive index of the epidermis was set to be 1.37, which is the average value of the volar side of the lower arm, the granular layer of the palm of the hand, and the basal layer of the palm of the hand.⁴⁰ The refractive index of the dermis was set to be 1.42, which is the average value of the volar side of the lower arm and the palm of the hand.⁴⁰ The optical parameters used in the MCS for the skin tissue model were summarized in Ref. 33. The $XYZ$ values were then calculated based on the simulated $O(\lambda )$. The above calculations were performed for various combinations of $C_m$, $C_{\mathrm{ob}}$, and $C_{\mathrm{db}}$ in order to obtain the data sets of chromophore concentrations and $XYZ$ values. Multiple regression analysis with 300 data sets established three regression equations for $C_m$, $C_{\mathrm{ob}}$, and $C_{\mathrm{db}}$:

The regression coefficients $a_i$, $b_i$, and $c_i$ ($i=0$, 1, 2, 3) reflect the contributions of the $XYZ$ values to $C_m$, $C_{\mathrm{ob}}$, and $C_{\mathrm{db}}$, respectively, and were used as the elements of a $4\times 3$ matrix ${\mathbf{N}}_{\mathbf{2}}$ as

Transformation with ${\mathbf{N}}_{\mathbf{2}}$ from the tristimulus values to the chromophore concentrations is thus expressed as

The computation times for the MCS on obtaining all the simulated spectra and the matrix ${\mathbf{N}}_{\mathbf{2}}$ in Eq. (11) were 9.2 h and 10 s, respectively. Once we determine the matrices ${\mathbf{N}}_{\mathbf{1}}$ and ${\mathbf{N}}_{\mathbf{2}}$, images of $C_m$, $C_{\mathrm{ob}}$, and $C_{\mathrm{db}}$ are reconstructed without the MCS. The total blood concentration image is simply calculated as $C_{\mathrm{tb}}=C_{\mathrm{ob}}+C_{\mathrm{db}}$.

We perform the particular color conversion from RGB values to $XYZ$ values for applicability of the method to different types of cameras. If the spectral sensitivity of the camera used is available, it will be possible to establish the regression equations that transform directly from RGB values to the chromophore concentrations, $C_m$, $C_{\mathrm{ob}}$, and $C_{\mathrm{db}}$ in the same manner as Eqs. (8) through (10). In such a case, however, the three regression equations for $C_m$, $C_{\mathrm{ob}}$ and $C_{\mathrm{db}}$ must be prepared for every camera because each type of camera has its own spectral sensitivity. On the other hand, $XYZ$ values are independent of types of cameras. Once we adjust the RGB responses of the camera to $XYZ$ values by the color standard, $C_m$, $C_{\mathrm{ob}}$, and $C_{\mathrm{db}}$ can be estimated from the RGB values by only the matrix ${\mathbf{N}}_{\mathbf{2}}$.

2.3.

Calculations of Arterial Inflow, Vascular Resistance, and Venous Capacitance

The limb arterial inflow is usually determined by drawing a line on the recording of $\mathrm{\Delta }V/V$ $\mathrm{mL}/100\mathrm{mL}$ that is tangent to the first few seconds following the cuff inflation. The slope of this line indicates the rate of volume change, which is caused by arterial inflow.⁶ Arterial inflow is expressed as a volume change per unit time, such as AI $\mathrm{mL}/(100\mathrm{mL}·\text{min})$. The mean arterial pressure MAP mmHg is calculated based on the well-known standard equation

Venous capacitance is defined as the percent change in volume of the limb after inflation of the occlusion cuff and can be determined by the difference between the baseline volume established prior to inflation of the cuff and the volume after the 2-min occlusion as VC $\mathrm{mL}/100\mathrm{mL}$.⁶

Figure 2 shows an illustration of a typical response curve of skin blood volume to limb occlusion by inflation of a thigh cuff at 50 mmHg and subsequent deflation of the cuff. We calculate the arterial inflow and venous capacitance in skin as ${\mathrm{AI}}_s$ $\mathrm{mL}/(100\mathrm{mL}·\text{min})$ and ${\mathrm{VC}}_s$ $\mathrm{mL}/100\mathrm{mL}$, respectively, from the response curve of the change in the total blood concentration of skin $\mathrm{\Delta }{C}_{\mathrm{tb}}/C_{\mathrm{tb},c}$ ($\mathrm{\Delta }{C}_{\mathrm{tb}}=C_{\mathrm{tb}}-C_{\mathrm{tb},c}$) to the occlusion at a pressure of 50 mmHg in the same manner as the SPG recording, where $C_{\mathrm{tb},c}$ is the total blood concentration at baseline ($t=0\text{min}$). Vascular resistance in skin ${\mathrm{VR}}_s$ $\mathrm{mm}\mathrm{Hg}·100\mathrm{mL}·\text{min}/\mathrm{mL}$ was calculated by dividing the measured MAP by ${\mathrm{AI}}_s$ (Eq. 14).

Fig. 2

Derivation of skin arterial inflow ${\mathrm{AI}}_s$ and venous capacitance ${\mathrm{VC}}_s$ from a time course of a ${\mathrm{\Delta }C}_{\mathrm{tb}}/C_{\mathrm{tb},c}$, during upper arm occlusion at 50 mmHg.

3.1.

Imaging System

Figure 3 schematically shows the experimental configurations for the 3(a) imaging system and 3(b) in vivo experiments with upper arm occlusion. A metal halide lamplight (LA-180Me-R, Hayashi, Japan) illuminated the surface of a sample via a light guide with a ring illuminator. The light source covered a range from 380 to 740 nm. Diffusely reflected light was captured by a 24-bit RGB CCD camera (DFK-21BF04, Imaging Source LLC, North Carolina) and a camera lens (Pentax/Cosmica, Japan; f 16 mm, $1:1.4$) to acquire an RGB color image of $640\times 480$ pixels. The field of view of the imaging system was $360\times 270\mathrm{mm}$. The lateral resolution of the images was estimated to be 0.56 mm. This indicates the best resolution with a nonscattering object. An IR-cut filter in the camera rejects unnecessary longer-wavelength light ($>700\mathrm{nm}$). A standard white diffuser with 99% reflectance (SRS-99-020, Labsphere Incorporated, North Carolina) was used to correct for the inter-instrument differences in the output of the camera and the spatial nonuniformity of the illumination. The RGB images were acquired at 15 frames per second (fps) and an average of 16 frames was stored in a personal computer at 4-s intervals and analyzed according to the visualization process described above. The standard deviation of RGB values between the 16 frames that are obtained from a subject under the normal condition was 0.15 in average, which indicate no significant difference between the 16 video frames.

Fig. 3

Experimental configurations of (a) the imaging system and (b) the in vivo experiments with upper arm occlusion.

3.2.

Upper Arm Occlusion Experiments

A pressure cuff was applied to the upper arms of 17 subjects (13 men and four women, mean age: $23\pm 1$ years) without any history or physical findings of venous or arterial diseases, as shown in Fig. 3(b). The five male subjects who exercised vigorously for two or more days per week and/or participated in daily physical training for at least six years were regarded as the active group (subject 1, subject 2, subject 3, subject 4, and subject 5). The remaining subjects with no or irregular physical activity (usually exercising less than one day per week) were regarded as the sedentary group. The systolic and diastolic blood pressures of the subjects were measured by the sphygmomanometer except for two of the sedentary male subjects. The data of blood pressure for the two of the sedentary male subject were unavailable owing to the experimental condition. Therefore the mean arterial blood pressure and the vascular resistance were calculated for 15 subjects in this study. The SPG (EC6, D.E. Hokanson, Washington) and a rapid cuff inflator (E-20, D.E. Hokanson) were used to measure in vivo forearm volume change $\mathrm{\Delta }V/V\mathrm{mL}/100\mathrm{mL}$. During the measurement, the subjects sat with their hands placed on a sample stage at approximately heart level. After a rest of 300 s, image acquisition and SPG recording were started and continued for 640 s at 4-s intervals. After 40 s of control, the cuff was inflated to 50 mmHg for 300 s by use of a rapid cuff inflator and subsequently deflated for 300 s. Inflation of the cuff to 50 mmHg prevents blood flow from leaving the measurement site but does not hinder arterial inflow. The SPG data was recorded for only 12 subjects whereas the acquisitions of RGB images were performed for all of the 17 subjects owing to experimental conditions. Analysis of both RGB images and forearm volume change $\mathrm{\Delta }V/V$ were performed offline after measurements were completed. To derive the image of ${\mathrm{AI}}_s$, we performed the linear least squares fitting to the time course of $\mathrm{\Delta }{C}_{\mathrm{tb}}/C_{\mathrm{tb},c}$ ($t=0–16\mathrm{s}$) for each pixel of a sequential image. This derivation process of ${\mathrm{AI}}_s$ image is relatively time consuming. The computation time for the images of ${\mathrm{AI}}_s$, ${\mathrm{VC}}_s$, and ${\mathrm{VR}}_s$ were 1200, 7 and 1200 s, using the Intel Core 2 CPU, 2.66 GHz when the RGB color image of $640\times 480$ pixels was analyzed. Use of a camera with a large number of pixels will improve the spatial resolution of resultant images, but it will increase computation time. A region of interest (ROI) was placed in a part of an image for each resultant image, as shown in Fig. 3(b). Simple linear regression analysis was used to describe the correlation coefficient $R$ between the SPG recordings and the results obtained by the proposed method. An unpaired Student’s $t$-test was used for statistical analysis when comparing the active group and sedentary group. The normality of the averaged value over the ROI for each group was tested by the Shapiro-Wilk test before the Student’s $t$-test. A $P$ value $<0.05$ was considered statistically significant.

4.1.

Responses of the Blood Volume to Cuff Occlusion

Figure 4 shows the forearm volume change $\mathrm{\Delta }V/V$ measured by the SPG for the cuff pressure of 50 mmHg and depicts differences among subjects. In Fig. 4, $\mathrm{\Delta }V/V$ rises quickly after the inflation of the cuff, and the rate of increase in $\mathrm{\Delta }V/V$ then slows. A rapid decrease in $\mathrm{\Delta }V/V$ occurred after deflation of the cuff, and $\mathrm{\Delta }V/V$ then returns to its baseline level. Figure 5 shows an example of the in vivo results obtained from one subject during cuff occlusion at 50 mmHg. The first increase in $C_{\mathrm{ob}}$ appeared after the cuff was inflated, which caused an increase in $C_{\mathrm{tb}}$, probably due to the blockage of venous outflow and the continuous arterial inflow. After peaking, $C_{\mathrm{ob}}$ and $C_{\mathrm{tb}}$ became constant, whereas $C_{\mathrm{db}}$ increased during occlusion. These changes in $C_{\mathrm{ob}}$, $C_{\mathrm{db}}$, and $C_{\mathrm{tb}}$ indicate the decrease of the arterial inflow rate and the deoxygenation of hemoglobin resulting from the consumption of oxygen by the local tissue, respectively. The rapid decreases in $C_{\mathrm{ob}}$, $C_{\mathrm{db}}$, and $C_{\mathrm{tb}}$ immediately after the deflation of the cuff suggest the outflow of venous blood. The tendency of the response in $C_{\mathrm{tb}}$ to the upper arm occlusion at 50 mmHg corresponds to the results for $\mathrm{\Delta }V/V$ shown in Fig. 4. Although there are some artifacts due to the shade originating from the curved and irregular surfaces of the hand, the lateral distribution of $C_{\mathrm{tb}}$ and the response to the venous occlusion were successfully observed. Time courses of $\mathrm{\Delta }$ $C_{\mathrm{tb}}/C_{\mathrm{tb},c}$ averaged over the ROI corresponding to the white box in Fig. 5 are shown in Fig. 6. During the cuff occlusion, $\mathrm{\Delta }$ $C_{\mathrm{tb}}/C_{\mathrm{tb},c}$ increased quickly and then changed slowly. After the cuff was deflated, $\mathrm{\Delta }$ $C_{\mathrm{tb}}/C_{\mathrm{tb},c}$ returned immediately to the baseline levels. This tendency of variations in $\mathrm{\Delta }$ $C_{\mathrm{tb}}/C_{\mathrm{tb},c}$ is similar to the SPG recordings of $\mathrm{\Delta }V/V$ shown in Fig. 4.

Fig. 4

Time courses of forearm volume changes $\mathrm{\Delta }V/V$ measured by the SPG during upper arm occlusion at 50 mmHg ($n=12$).

Fig. 5

Typical images of in vivo measurements during cuff occlusion at 50 mmHg (from top to bottom: RGB image, $C_{\mathrm{ob}}$, $C_{\mathrm{db}}$, and $C_{\mathrm{tb}}$).

Fig. 6

Time courses of $\mathrm{\Delta }{C}_{\mathrm{tb}}/C_{\mathrm{tb},c}$ averaged over the ROIs corresponding to the white box in Fig. 5 ($n=12$).

4.2.

Visualizations of Arterial Inflow, Vascular Resistance, and Venous Capacitance in Human Skin

Figures 7, 8, and 9 show the images of ${\mathrm{AI}}_s$, ${\mathrm{VR}}_s$, and ${\mathrm{VC}}_s$, obtained from the method, respectively. The color coded pixel values over the skin area in each image shown in Figs. 7, 8, and 9 represent the estimated values of ${\mathrm{AI}}_s$, ${\mathrm{VR}}_s$ and ${\mathrm{VC}}_s$, respectively. They are used to evaluate the spatial distribution of the vasodilative indices and the differences among individuals. The average value over the area corresponding to ROI (White box) in Figs. 7, 8, and 9 is used to compare the results from the proposed method to the SPG recordings and to evaluate the difference between the active group and sedentary group. In Figs. 7, 8, and 9, it is clearly demonstrated that ${\mathrm{AI}}_s$, ${\mathrm{VR}}_s$, and ${\mathrm{VC}}_s$ differ among individuals. The spatial heterogeneities can also be seen in the images of ${\mathrm{AI}}_s$, ${\mathrm{VR}}_s$, and ${\mathrm{VC}}_s$, which is indicative of spatial differences in the quantity and density of microvasculature in skin tissue. In the preliminary experiments, the repeatability of the measurements was evaluated for one subject. The results for five repeated measurements were $0.97\pm 0.12\mathrm{mL}/100\mathrm{mL}\cdot \min$, $1.11\pm 0.04\mathrm{mL}/100\mathrm{mL}$, and $81.7\pm 10.4\mathrm{mmHg}\cdot 100\mathrm{mL}\cdot \min /\mathrm{mL}$, for ${\mathrm{AI}}_s$, ${\mathrm{VC}}_s$, and ${\mathrm{VR}}_s$, respectively. We have also confirmed that the measurements are not affected by variations in the orientation of the hand.

Fig. 7

Images of skin arterial inflow ${\mathrm{AI}}_s$ obtained by the proposed method ($n=17$).

Fig. 8

Images of vascular resistance ${\mathrm{VR}}_s$ obtained by the proposed method ($n=15$).

Fig. 9

Images of venous capacitance ${\mathrm{VC}}_s$ obtained by the proposed method ($n=17$).

Figure 10 shows a comparison of the results obtained from the proposed method and measurements from the SPG for 10(a) the arterial inflow, 10(b) the vascular resistance, and 10(c) the venous capacitance. The estimated ${\mathrm{AI}}_s$, ${\mathrm{VR}}_s$, and ${\mathrm{VC}}_s$ are well correlated with the measurements of AI, VR, and VC by the SPG, respectively. The correlation coefficients $R$ between the estimated values by the method and the measurements by the SPG were calculated to be 0.83 ($P<0.001$) for the arterial inflow, 0.77 ($P<0.01$) for the vascular resistance, and 0.77 ($P<0.01$) for the venous capacitance, which revealed a significant relationship between the proposed method and measurements using the conventional SPG.

Fig. 10

Comparison of the estimated values by the proposed method and the measurements of SPG for (a) $\mathrm{AI}$, (b) $\mathrm{VR}$, and (c) $\mathrm{VC}$ ($n=12$).

Figure 11 shows the comparison of mean values between the active group and the sedentary group for (a) ${\mathrm{AI}}_s$, (b) ${\mathrm{VR}}_s$, and (c) ${\mathrm{VC}}_s$. The mean arterial inflow ${\mathrm{AI}}_s$ in the active group [$1.50\pm 0.29\mathrm{mL}/(100\mathrm{mL}\cdot \min )$] was significantly higher than that in the sedentary group [$0.66\pm 0.32\mathrm{mL}/(100\mathrm{mL}\cdot \min )$] ($P<0.001$). The mean vascular resistance ${\mathrm{VR}}_s$ in the active group ($66.1\pm 13.4\mathrm{mmHg}\cdot 100\mathrm{mL}$) was significantly lower than that in the sedentary group ($164.7\pm 90.5\mathrm{mmHg}\cdot 100\mathrm{mL}\cdot \min /\mathrm{mL}$) ($P<0.05$). The mean venous capacitance ${\mathrm{VC}}_s$ in the active group ($0.87\pm 0.15\mathrm{mL}/100\mathrm{mL}$) was significantly higher than that in the sedentary group ($0.69\pm 0.14\mathrm{mL}/100\mathrm{mL}$) ($P<0.05$).

Fig. 11

Estimated values compared by subject group for (a) $\mathrm{AI}$, (b) $\mathrm{VR}$, and (c) $\mathrm{VC}$. Values are $\text{means}\pm \mathrm{sd}$. $*P<0.05$.

Previous studies have demonstrated that the peripheral vascular functions are related to the levels of physical activity and fitness.^2–4 It has been reported that the venous capacitance was reduced in patients with spinal cord injury compared with the able-bodied subjects, which was attributed to the combination of sympathetic denervation and the absence of regular orthostatic challenge.³ Lower venous capacitance was also observed in the sedentary subjects compared with the active subjects, suggesting that the level of activity contributes to the magnitude of venous distensibility by enhancing vasodilatory responsiveness of the vessels.³ The influence of physical activity on the cutaneous blood flow during leg compression has been investigated previously for the active-lifestyle subjects and the sedentary subjects.⁴ A higher arterial inflow was demonstrated in the active subjects compared with the sedentary subjects, which was indicative of the adaptive physiologic change by the venous system to accommodate increased arterial inflow due to exercise.⁴ A significant increase in vascular resistance in subjects with spinal cord injury was demonstrated by using the SPG recording.² The enhanced vascular resistance was discussed in terms of structural changes in vasculature, such as a decrease in the number of arterioles and capillaries and/or a decrease in the diameter of the resistance vessels as well as functional changes due to variations in endothelium-derived factors and/or sympathetic vascular regulation.² In the present study, the arterial inflow and the venous capacitance were significantly higher in the active group compared with the sedentary group, whereas the venous capacitance was significantly lower in the active group compared with the sedentary group. Therefore the differences in ${\mathrm{AI}}_s$, ${\mathrm{VR}}_s$, and ${\mathrm{VC}}_s$ among individuals demonstrated in Figs. 7, 8, and 9 may reflect the variations in the level of lifestyle activity. It might be possible to separate the active and sedentary groups based on the measurements of ${\mathrm{AI}}_s$, ${\mathrm{VR}}_s$, and ${\mathrm{VC}}_s$ by doing discriminant analysis such as leave-one-out method. This will be useful for clinical diagnosis of various vascular dysfunctions related to the lifestyle and should be investigated in the future.

In the present study, all experiments were performed in a dark room to prevent interference from the ambient light. If the main light source is used under the ambient artificial light, the skin surface will be illuminated by the mixture of two types of lighting. In such a case, the ambient artificial light may be a source of misestimation in $C_{\mathrm{ob}}$, $C_{\mathrm{db}}$, and $C_m$. To estimate $C_{\mathrm{ob}}$, $C_{\mathrm{db}}$, and $C_m$ accurately, the measurements of color standard for adjusting the RGB responses to $XYZ$ values should be performed under the mixture of main light source and ambient artificial light. The ambient natural light should be avoided because it is often unreliable and variable. The RGB values of skin with darker color will be very small at very low resolution, and the conversion to $XYZ$ color space could compound likely artifacts in measurement. In this case, the conversion from RGB color space to $XYZ$ color space may cause misestimation of total blood $C_{\mathrm{tb}}$ in the dermis. Therefore, the measurements of ${\mathrm{AI}}_s$, ${\mathrm{VR}}_s$, and ${\mathrm{VC}}_s$ could be affected by variations in skin color. Experiments involving individuals of African or Indian descent should be performed in the future.