2.1.

Melanin Solution Preparation

Melanin solutions were prepared by dissolving the melanin (M8631, Sigma-Aldrich) in tris(hydroxymethyl)aminomethane hydrochloride (TrisHCl) and adjusting the pH to 8.0. The solutions were further mixed in an orbital shaker for 8 h. Aliquots of 1 mL were centrifuged at 13,000 rpm for 1 min prior to imaging.

2.2.

Cell Culture

Human melanoma cells MNT1 and HaCaT keratinocytes were maintained in a 5% ${\mathrm{CO}}_2$ humidified incubator at 37°C in cell culture medium composed of RPMI 1640 medium (Sigma) supplemented with 10% heat-inactivated fetal calf serum (Sigma) and penicillin-streptomycin (1.25 mL; Sigma). The cells were collected by centrifugation and transferred in a flask containing 10 mL medium. The cells were harvested by treatment with 0.25% Trypsin/EDTA and collected after centrifugation (700 rpm, 3 min) in cell culture medium. The cells were seeded in an OptiCell (Nunc) at a density of $6\times {10}^5\text{cells}/\mathrm{mL}$ and MPT and FLIM imaging was performed 24 h postseeding.

2.3.

In Vivo Skin Color Measurements

All studies involving volunteers were undertaken with approval from the Princess Alexandra Hospital/University of Queensland Research Ethics committee and written consent from the volunteers. Two male and three female volunteers, aged 19 to 42, from African, Indian, Caucasian, and Asian backgrounds were recruited. A $1\text{-}{\mathrm{cm}}^2$ area on the right dorsal (photoexposed) and volar (photoprotected) forearm of each volunteer was selected. A Konica-Minolta CM-2600d reflectance spectrophotometer was used to measure the tristimulus $L*$, $a*$, and $b*$ values at three spots within the selected area. The $L*$ value represents brightness from black ($L*=0$) to white ($L*=100$). The $a*$ scale is the green to red scale, with $a*<0$ indicating green and an increase in $a*$ ($a*>0$) indicating an increase in redness. Similarly, $b*$ represents blue ($b*<0$) to yellow ($b*>0$) (Ref. 63). Each set of $L*$, $a*$, $b*$ values was measured using the 3 mm aperture plate provided with the spectrophotometer upon calibration with a white tile.

2.4.

MPT and Fluorescence Lifetime Imaging

Multiphoton tomography was performed with a DermaInspect (Jenlab, Jena, Germany) equipped with a tunable (710 to 920 nm) Titanium:Sapphire ultrashort-pulse laser (Mai Tai, Spectra Physics). The laser pulse width and repetition rate were 85 fs and 80 MHz, respectively. A broad bandpass filter (BG39, Schott glass color filter) with a 700-nm cutoff was used to block the excitation laser light. In vitro and in vivo imaging was performed with a high-NA oil-immersion objective ($40\times$, NA 1.3). The FLIM color map distribution was realized by employing a time-correlate single-photon counting (TCSPC) module with time resolution of 200 ps. Images sized $128\times 128\text{pixels}$ were recorded in four different spectral channels: channels 1, 2, 3, 4 corresponding to 350 to 450 nm; 450 to 515 nm; 515 to 620 nm; and 620 to 670 nm spectral bands.

2.5.

Data Analysis

Fluorescence lifetime data analysis was performed using SPCImage 2.9 (Becker & Hickl). Fluorescence was detected within the emission range of 515 to 620 nm, which includes the peak emission wavelengths of melanin, 550 to 575 nm (Refs. 17 and 54). For the images of melanin in solution and of the cell cultures, the bin number was set to 3; for the in vivo images, it was set to 4. This number defines the number of adjacent pixels whose photon decay data is used to obtain a photon decay profile and calculate lifetimes in a given pixel. The bin number $n$ is calculated from

After setting the bin number, regions of interest (ROI) for FLIM analysis were defined such that skin folds, areas of evident photobleaching, and cells out of the focal plane would be avoided in the calculation of lifetimes. The area of the ROI was calculated after calibration using ImageJ®. The time‐dependent photon data were fitted to an exponential function, yielding one or more lifetime distributions ${\tau }_i$ (herein given in ps) and the corresponding weighting coefficient $a_i$ (%) of the overall fluorescence decay signal¹³

3.1.

MPT Imaging of Synthetic Melanin Solution

The photon decay profiles obtained from the synthetic melanin solutions of varying concentration (0.5 to $2.0\mathrm{mg}/\mathrm{mL}$) displayed bi-exponential photon decay profiles, yielding a fast (${\tau }_1$) and slow (${\tau }_2$) lifetime. Figure 1(a) shows an example of such a profile. Figure 1(b) to 1(d) shows fluorescence pixel intensity versus lifetime values ${\tau }_1$ and ${\tau }_2$, and amplitudes $a_1$ and $a_2$. Over the range of concentrations investigated, the lifetimes [$\text{mean}\pm \text{standard error of the mean}(\mathrm{s.e.m}\mathrm{.}\mathrm{)}$] corresponding to the highest pixel intensity are ${\tau }_1=(264\pm 8)\mathrm{ps}$ and ${\tau }_2=(2091\pm 23)\mathrm{ps}$. The corresponding amplitudes ($\text{mean}\pm \mathrm{s.e.m.}$) are $a_1=(77\pm 0.8)\%$ and $a_2=(23\pm 0.8)\%$, indicating that the ${\tau }_1$ lifetime is the most significant lifetime signature of melanin in solution. Figure 1(e) and 1(f) shows that the maximum pixel intensity associated with lifetime ${\tau }_1$ increases linearly with melanin concentration from 0.5 to $2.0\mathrm{mg}/\mathrm{mL}$. The maximum pixel intensity associated with lifetime ${\tau }_2$ increases linearly for concentrations ranging from 0.5 to $1.75\mathrm{mg}/\mathrm{mL}$. The correlation coefficient is slightly higher with the ${\tau }_1$ and $a_1$ data compared to the ${\tau }_2$ and $a_2$ data.

Fig. 1

Fluorescence lifetime imaging data from synthetic melanin in TrisHCl (pH 8): (a) representative photon decay profile from the $1\mathrm{mg}/\mathrm{mL}$ solution. A biexponential curve fits the data, with the reduced coefficient $X^2=1.19$. Distribution of (b) the fast lifetime ${\tau }_1$; (c) the slow lifetime ${\tau }_2$; (d) the amplitudes $a_1$ and $a_2$, at concentrations ranging from 0.5 to $2.0\mathrm{mg}/\mathrm{mL}$; (e) maximum fluorescence intensity corresponding to ${\tau }_1$ and ${\tau }_2$ lifetimes versus melanin concentration; (f) maximum fluorescence intensity corresponding to amplitudes $a_1$ and $a_2$ lifetimes versus melanin concentration.

3.2.

Selective Wavelength Imaging of Melanocyte-Keratinocyte Cocultures

The fluorescence intensity image and FLIM ${\tau }_1$ color of melanocyte-keratinocyte cocultures excited at 800 nm shows selective excitation of melanin-producing melanocytes [Fig. 2(a) and 2(b)]. The appearance of strong fluorescence signals within melanocytes corresponds with the melanin granules that extend into the dendrites of melanocytes. FLIM parameters of the melanocyte-keratinocyte cocultures (Fig. 3) show that lifetime values corresponding to the peaks in the melanocyte fluorescence are ${\tau }_1\approx 145\mathrm{ps}$ [Fig. 3(a)]; ${\tau }_2\approx 1883\mathrm{ps}$ [Fig. 3(b)]. The amplitude values corresponding to melanin are $a_1\approx 93\%$, $a_2\approx 7\%$ [Fig. 3(c)]. A weak fluorescence stemming from the keratinocytes surrounding the melanocytes is seen in Fig. 2(a) and 2(b). The lifetimes associated with the peaks in the keratinocyte fluorescence are ${\tau }_1\approx 750\mathrm{ps}$ [Fig. 3(a)] and ${\tau }_2\approx 3000\mathrm{ps}$ [Fig. 3(b)]. This weak fluorescence is likely to be mainly from flavine adenine dinucleotide (FAD), which may be partially excited at 800 nm and has a range of emission wavelengths which overlap with that of melanin (Table 1).

Fig. 2

(a) Fluorescence intensity image and (b) FLIM ${\tau }_1$ color map of an MNT1 melanocyte-HaCaT keratinocyte coculture obtained upon excitation at 800 nm.

Fig. 3

Fluorescence lifetime distributions obtained from the MNT1 melanocyte–HaCat keratinocytes coculture ($n=3$ measurements), excitation wavelength 800 nm: (a) fast lifetime ${\tau }_1$; (b) slow lifetime ${\tau }_2$; and (c) corresponding amplitudes $a_1$ and $a_2$.

The ability of the imaging system to delineate melanin based on spectral and morphological differences in a biologically mixed environment provides opportunity to observe the distribution of melanin in human skin in vivo.

3.3.

In Vivo MPT Imaging of Epidermal Melanin in Different Healthy Ethnic Skin

FLIM signatures of differences in human skin melanin content were investigated by comparing dorsal forearm skin of five volunteers of African, Indian, Caucasian, and Asian ethnic backgrounds. Table 4 shows the mean spectrophotometric tristimulus $L*$, $a*$, and $b*$ values acquired from each volunteer forearm. There is a consistent trend where the dorsal side is darker than the volar side, presumably due to increased sun exposure to the former. Interestingly, the contrast ($\mathrm{\Delta }L*$) between dorsal and volar skin color is smaller in darker skin compared to lighter skin (Table 4). Figure 4 shows examples of the MPT signal intensity and FLIM color maps from the strata granulosum and basale of the Asian [Fig. 4(a) and 4(b)] and the African [Fig. 4(c) and 4(d)] volunteer. The densely packed keratinocytes of the stratum granusolum layer and the much smaller, polygonal keratinocytes of the stratum basale^12,65 are clearly visible.

Table 4

Mean spectrophotometric data of volunteers (three measurements within a 1  cm2 area on each side of forearm).

Fig. 4

Fluorescence intensity map after 740 nm excitation and corresponding fast (${\tau }_1$) lifetime color maps obtained upon excitation wavelength at 800 nm. Asian volunteer’s (a) stratum granulosum and (b) stratum basale. African volunteer’s (c) stratum granulosum and (d) stratum basale. The red lines indicate the regions of interest in which the area-normalized fluorescence intensities were calculated.

The corresponding area-normalized fluorescence intensity versus ${\tau }_1$ lifetime profiles is shown in Fig. 5. On the dorsal side of the African volunteer’s skin, the maximum mean fluorescence intensities are $(496\pm 90)\mathrm{a.u.}/{\mu \mathrm{m}}^2$ at ${\tau }_1=140\mathrm{ps}$ and $(549\pm 211)\mathrm{a.u.}/{\mu \mathrm{m}}^2$ at ${\tau }_1=120\mathrm{ps}$ in the stratum granulosum and basale, respectively [Fig. 5(a)]. In volar skin, the peak intensities are $(440\pm 141)\mathrm{a.u.}/{\mu \mathrm{m}}^2$ at ${\tau }_1=130\mathrm{ps}$ and $(521\pm 50)\mathrm{a.u.}/{\mu \mathrm{m}}^2$ at ${\tau }_1=120\mathrm{ps}$ in the stratum granulosum and basale, respectively [Fig. 5(b)]. On the dorsal side of the Asian volunteer’s skin, the peak florescence intensity is $(203\pm 60)\mathrm{a.u.}/{\mu \mathrm{m}}^2$ at ${\tau }_1=220\mathrm{ps}$ and $(195\pm 30)\mathrm{a.u.}/{\mu \mathrm{m}}^2$ at ${\tau }_1=180\mathrm{ps}$ in the stratum granulosum and basale, respectively [Fig. 5(c)]. On the volar side, the peak florescence intensity is $(97\pm 44)\mathrm{a.u.}/{\mu \mathrm{m}}^2$ at ${\tau }_1=280\mathrm{ps}$ and $(137\pm 106)\mathrm{a.u.}/{\mu \mathrm{m}}^2$ at ${\tau }_1=150\mathrm{ps}$ in the stratum granulosum and basale, respectively [Fig. 5(d)].

Fig. 5

Area-normalized fluorescence intensity profiles corresponding to the fast (${\tau }_1$) lifetime from the dorsal and volar skin of (a) and (b) the African volunteer and (c) and (d) the Asian volunteer.

Figure 6 shows the mean fluorescence intensity versus ${\tau }_2$ lifetime profiles obtained from the same FLIM measurements. The profiles obtained from the stratum granulosum of the African volunteer’s skin display maximum fluorescence intensities of $(148\pm 67)\mathrm{a.u.}/{\mu \mathrm{m}}^2$ and ($(128\pm 34)\mathrm{a.u.}/{\mu \mathrm{m}}^2$ on the dorsal and volar skin sides, corresponding to ${\tau }_2$ lifetimes of 1800 and 1720 ps, respectively [Fig. 6(a) and 6(b)]. On the other hand, two distinct peaks in maximum fluorescence intensity are obtained from the stratum basale: ($89\pm 45$) and $(131\pm 33)\mathrm{a.u.}/{\mu \mathrm{m}}^2$ corresponding to ${\tau }_2=520$ and 1760 ps on the dorsal side and ($100\pm 59$) and $(80\pm 22)\mathrm{a.u.}/{\mu \mathrm{m}}^2$ corresponding to ${\tau }_2=640$ and 1680 ps on the volar side [Fig. 6(a) and 6(b)]. The skin of the Asian volunteer yields single maxima in mean fluorescence intensities. In the stratum granulosum, these are $(242\pm 21)\mathrm{a.u.}/{\mu \mathrm{m}}^2$ and $(138\pm 47)\mathrm{a.u.}/{\mu \mathrm{m}}^2$, both at ${\tau }_2=1920\mathrm{ps}$, in dorsal and volar skin [Fig. 6(c) and 6(d)]. In the stratum basale, the peak intensities are $(203\pm 40)\mathrm{a.u.}/{\mu \mathrm{m}}^2$ at ${\tau }_2=1840\mathrm{ps}$ in dorsal skin and $(72\pm 31)\mathrm{a.u.}/{\mu \mathrm{m}}^2$ at ${\tau }_2=1760\mathrm{ps}$ in volar skin.

Fig. 6

Area-normalized fluorescence intensity profiles corresponding to the slow (${\tau }_2$) lifetime from the dorsal and volar skin of (a) and (b) the African volunteer and (c) and (d) the Asian volunteer.

3.4.

Correlation Between the Fluorescence Intensities and Skin Pigmentation

Figure 7(a) through 7(d) shows that the mean maximum fluorescence intensities corresponding to the ${\tau }_1$ lifetime in the stratum granulosum and basale on the dorsal and volar forearm correlates with the measured mean $L*$ value. These data show a general trend in which fluorescence intensities decreases linearly with increasing skin color lightness (increasing mean $L*$ value) when two clusters of skin types are considered: the darkly pigmented African skin and the lighter pigmented skin types (Indian, Caucasian, and Asian skin). The correlation between fluorescence intensities and mean $L*$ values are significantly weaker when only the cluster to lightly pigmented skin types is considered.

Fig. 7

Maximum area-normalized fluorescence intensity corresponding to the lifetime ${\tau }_1$ in stratum granulosum and stratum basale versus mean $L*$ value (Table 4) in (a) dorsal stratum granulosum; (b) dorsal stratum basale; (c) volar stratum granulosum; and (d) volar stratum basale. Data ($\text{mean}\pm \mathrm{s.e.m.}$) from the five volunteers.

Alaluf et al. showed a correlation between total melanin content and $b*$ for lightly pigmented skin types.⁶³ Removing the data points obtained from African skin, we obtain significant correlations between the maximum fluorescence intensities in the dorsal and volar stratum granulosum and $b*$ (Fig. 8). We did not find any significant correlation between the mean maximum fluorescence intensities corresponding to the ${\tau }_2$ lifetime and $L*$, $a*$, or $b*$ values (data not shown).

Fig. 8

Maximum area-normalized fluorescence intensity corresponding to the lifetime ${\tau }_1$ versus mean $b*$ value in dorsal and volar stratum granulosum (Table 4). Data ($\text{mean}\pm \mathrm{s.e.m.}$) from the five volunteers.

Figure 9 juxtaposes the mean maximum fluorescence intensities corresponding with the ${\tau }_1$ lifetime in the dorsal and volar skin at each skin layer. The data from the African volunteer’s skin suggest a higher mean fluorescence intensity in the stratum basale than in the stratum granulosum. Such a pattern is not seen in the data from the other skin types, consistent with results in Fig. 7.

Fig. 9

Maximum area-normalized fluorescence intensity corresponding to the lifetime ${\tau }_1$ versus skin type. Data ($\text{mean}\pm \mathrm{s.e.m.}$) from the five volunteers.

4.1.

Synthetic Melanin Solutions

Our results with melanin solutions of varying concentrations show that the fluorescence intensity measured from a solution increases linearly with melanin concentration [Fig. 1(b) through 1(f)]. The slopes of the maximum pixel intensity versus melanin concentration curves are in good agreement with that of Gallas et al.’s curve of fluorescence intensity versus melanin for an excitation wavelength of 440 nm (Fig. 3 in Ref. 66), which is close to the equivalent two-photon excitation wavelength we used (800 nm). From this result we inferred that we could expect differences in melanin concentration in skin in vivo to be measured based on differences in fluorescence intensity. This is further corroborated by Ait El Madani et al.’s results showing an increase in two-photon excited fluorescence signal intensity with increasing melanin density in human skin.⁵³

The ${\tau }_1$ and ${\tau }_2$ fluorescence lifetime values we obtained from the synthetic melanin solutions are close to the first two lifetimes reported by Teuchner et al.^54,55 and Nighswander-Rempel et al.³² for synthetic melanin dissolved in dimethyl sulfoxide (DMSO) and water, respectively (Table 1). Unlike those authors, however, we did not observe a third lifetime component as all the fluorescence decay curves [e.g., Fig. 1(a)] displayed biexponential decay only. One reason for the difference may be Teuchner et al.’s use of stepwise multiphoton excitation rather than simultaneous excitation. The stepwise multiphoton excitation was capable of achieving improved selectivity of melanin excitation, resulting in the greater suppression of background skin autofluorescence and greater melanin imaging contrast.^32,49

4.2.

Cell Cultures

By selectively exciting the melanin in melanocytes in a melanocyte-keratinocyte coculture at 800 nm [Fig. 2(a) and 2(b)], we confirmed results by Dimitrow et al.,²⁴ which show that the melanocytes could be selectively excited on the basis of the laser excitation wavelength. The fluorescence lifetimes and amplitudes we measured in melanocytes [Fig. 3(a) through 3(c)] are in close agreement with values obtained by Dimitrow et al.²⁴ and Sugata et al.⁶⁰ from cell cultures and in vivo, respectively (Tables 1 and 2). Figure 2(b) shows that melanin was detected in the melanocytes; no transfer to the keratinocytes occurred. The measured fluorescence is strictly from melanin produced in the melanocytes. There is no evidence to date suggesting that the HaCat keratinocytes contribute to any potential eumelanogenesis within a period of 24 h postseeding and without stimulation.

4.3.

Human Skin In Vivo

Dimitrow et al. showed images of melanocytes in melanoma skin.²⁴ We were not able to identify melanocytes in our in vivo images. This is likely due to the fact that we imaged healthy skin and that the melanocytes, which occur in a density of 1 melanocyte per 16 to 32 keratinocytes in healthy skin,⁶⁷ were not visible at our resolution, making it difficult to distinguish melanocytes from keratinocytes.

The ${\tau }_1$ lifetimes corresponding to the maximum fluorescence intensities on the dorsal and volar side of the African volunteer’s skin (${\tau }_1=120$ to 140 ps) are in very good agreement with the fast lifetimes attributed primarily to melanin fluorescence in Dimitrow et al.²⁴ and Sugata et al.’s⁶⁰ in vivo studies (Table 3). In the Asian volunteer’s skin the ${\tau }_1$ lifetimes corresponding to peak intensities are somewhat higher (${\tau }_1=150$ to 220 ps) but also agree with published values. The ${\tau }_2$ lifetimes corresponding with the maximum fluorescence intensities, ranging from ${\tau }_2=1680$ to 1920 ps, are also close to Dimitrow et al. and Sugata et al.’s published slow lifetime values. We obtained additional peaks in the fluorescence intensity at ${\tau }_2=520$ and 640 ps in the stratum basale of the African volunteer’s dorsal and volar skin. These values are difficult to interpret. They are close to the ${\tau }_2$ values obtained by Sugata et al.⁶⁰ in a hair bulb (Table 3), but we were not imaging at depths at which the hair bulbs are found. These ${\tau }_2$ values might indicate that some melanin is packed in melanosomes of African skin in a similar fashion to melanin found in hair. At this stage though, the ${\tau }_2$ values are not sufficiently informative to draw any conclusions from them.

To measure differences in melanin content in layers of the viable epidermis (stratum granulosum and stratum basale) of a given volunteer and across skin types, we focused on the maximum fluorescence intensity obtained from fluorescence lifetime measurements. The maximum fluorescence intensity associated with the ${\tau }_1$ lifetimes is the most useful in revealing differences in skin pigmentation. The peak intensities obtained from the dorsal skin of the African volunteer’s skin are two to three times greater than from the dorsal skin of the Asian volunteer [Fig. 5(a) versus 5(c)]. The data from the volar sides show four- to fivefold greater peak fluorescence intensities in African versus Asian skin [Fig. 5(b) versus 5(d)]. These results are in agreement with Tadokoro et al.’s data showing the melanin content of unirradiated Black skin being roughly five times greater than unirradiated Asian skin⁴ and consistent with other results showing an increase in melanin concentration with pigmentation.^8,11,68

Whereas there is almost no variation between fluorescence intensities in the dorsal and volar side of the African volunteer’s skin [Fig. 5(a) versus 5(b)], in the Asian skin, the mean intensity measured in the stratum granulosum is about two times greater on the dorsal side than the volar side. The mean intensities measured in the stratum basale differ by a factor of 1.4 [Fig. 5(c) versus 5(d)]. Our results from Asian skin agree with Alaluf et al.’s who have measured a total melanin concentration in photoexposed skin of Chinese volunteers that is about 1.5 times greater than in photoprotected skin, with a difference in mean $L*$ values of 10.7 (Ref. 8), comparable to our $\mathrm{\Delta }L*=8.20$ for the Asian volunteer (Table 4). Their results also yielded a 1.8-fold difference in African skin, but their difference in mean $L*$ values is 11, whereas in our study, $\mathrm{\Delta }L*=2.04$ for the African skin (Table 4).

The $L*$ value is the primary tristimulus value, which correlates with total epidermal melanin content.⁶³ While our method is appropriate for quantifying difference in melanin content between highly pigmented African skin and lightly pigmented Asian skin, it is not sufficiently precise to resolve differences in melanin concentration within the cluster of lighter skin types, for which $L*\approx 47$ to 66. This is seen in the fact that the correlation seen in Fig. 7(a) through 7(d) are much weaker when the African skin data are removed, and by the lack of a gradient in mean maximum melanin intensity with skin depth comparable to that obtained for African skin (Fig. 9). A likely explanation is the fact that melanosomes in lightly pigmented skin are smaller and mainly distributed in clusters, whereas in African skin they are mainly dispersed as larger, individual particles.^3,8,63 Clusters of melanosomes in lightly pigmented skin may be dispersed less uniformly than individual particles in African skin, yielding inconsistent results when comparing melanin fluorescence intensity and measurements of visible skin color.

Alaluf et al. have also shown a strong exponential correlation between total epidermal content and $a*$ (redness), owing to the fact that melanin reflects large amounts of red incident light.⁶³ We did not obtain meaningful correlations between $a*$ and the fluorescence intensities associated with ${\tau }_1$ or ${\tau }_2$, probably due to our small number of data points. We obtained a positive correlation between the $b*$ value and the maximum fluorescence intensity skin associated with lifetime ${\tau }_1$ measured in the stratum granulosum, though not the stratum basale, of the lighter pigmented Indian, Caucasian, and Asian volunteers (Fig. 8). This result agrees with Alaluf et al.’s data⁶³ and further confirms that the fluorescence signal associated with ${\tau }_1$ reflects melanin.

We did not obtain any meaningful correlation between maximum fluorescence intensities associated with the ${\tau }_2$ lifetime and $L*$. Making sense of the peak fluorescence intensities corresponding with ${\tau }_2$ lifetimes is difficult. Peak fluorescence intensities obtained from the volar skin overlap, whereas values from the dorsal side corresponding with ${\tau }_2=1760$ to 1920 ps are about 1.6 times higher in Asian skin than in African skin. A correlation between maximum fluorescence intensity and skin pigmentation was not obtained. While the lifetime ${\tau }_1$ is a signature of melanin fluorescence, the fluorescence associated with lifetime ${\tau }_2$ may reflect the interaction of melanin with protein in the cellular environment such as nicotinamide adenine dinucleotide phosphate (NAD(P)H), as its two-photon excitation wavelength overlaps with that of melanin and its range of emission wavelengths is close.¹³ This would explain the nearly overlapping fluorescence intensity profiles associated with ${\tau }_2$ measured African and Asian skin. Due to the protein in our synthetic melanin solution, similar correlations were found between maximum fluorescence intensity and lifetimes ${\tau }_1$ and ${\tau }_2$ [Fig. 1(e) and 1(f)].

In general, a certain percentage of the fluorescence we observed in the skin of volunteers is likely also due to other skin fluorophores such as NADH and FAD. The contribution of NADH is assumed minimal due to its maximum emission wavelength around 450 nm (Refs. 69 and 70) being outside of the range of detected emission used in this study (515 to 620 nm). FAD, however, may be a more significant confounder. The FAD maximum emission wavelength overlaps with that of melanin at about 525 nm (Refs. 69 and 71). The quantum yield of neutral FAD has been reported as 0.03 (Ref. 5), whereas that of synthetic eumelanin is on the order of ${10}^{-4}$ (Ref. 4).

One may be able to improve detection efficiency of melanin fluorescence, in particular as a function of depth into the skin (Fig. 9), by using optical clearing agents. Glycerol and DMSO in solution have been shown to reduce skin optical scattering and increase optical penetration depth.^38,72 However, issues of systemic toxicity of DMSO must be resolved prior to use with volunteers.⁷² We may also achieve enhanced detection of melanin by using stepwise instead of simultaneous two-photon absorption. Eichhorn et al. have shown stepwise multiphoton excitation to be effective in selectively exciting melanin and suppressing FAD excitation.⁷¹ Furthermore, Kerimo et al. have shown that Sepia melanin was efficiently excited by three-photon excitation.⁷³

Finally, a larger study with more than one African, Indian, and Asian volunteer is needed in order to show that the differences in maximum fluorescence intensities (Fig. 9) are statistically significant.