2.2. Melanin Fluorescence

Dermatofluoroscopy was developed with the aim of becoming a new diagnostic tool for early detection of melanotic malignant melanoma [19,20,23]. It is based on the measurement of the spectral distribution of ultra-weak melanin fluorescence from the melanosomes of cutaneous pigmented cells. With conventional one-photon excitation, this melanin fluorescence is completely concealed by the much more intense fluorescence of other endogenous fluorophores, e.g., of NAD(P)H and flavins (the so-called autofluorescence of skin). To overcome this problem, a special fluorescence excitation mechanism of non-linear spectroscopy is used: a stepwise two-photon excitation of melanin with nanosecond pulses at 800 nm. This is the hallmark of dermatofluoroscopy. This type of excitation is unique for melanin in contrast to all other endogenous fluorophores because only melanin can absorb 800-nm photons. However, the energy of a single 800-nm photon is too low to excite fluorescence in the visible spectral range; this requires the subsequent absorption of a second 800-nm photon in the already excited state (for details, see below and Figure 1). The parameters of the excitation laser are chosen to prevent a simultaneous absorption of two 800-nm photons, as this would be more or less equivalent to a single 400-nm photon absorption and result in (unwanted) autofluorescence.

The principle of dermatofluoroscopy. Different modes of excitation of fluorescence in the visible spectral region: (a) conventional fluorescence excitation by one photon (e.g., 400 nm); (b) excitation by simultaneous absorption of two photons via an only virtual energy level (e.g., 800 nm, preferably from a femtosecond laser); (c) excitation by stepwise absorption of two photons (e.g., 800 nm, preferably from a nanosecond laser) via a real intermediate energy level.

Figure 2 illustrates the energy level scheme of melanin (Jablonski diagram) and the underlying processes which determine the fluorescence spectrum. After absorption of the first photon, a first excited state is reached where there is competition between the (“downhill”) relaxation (k (N) in the nevus, k (MM) in the melanoma) and the strength of the second (“uphill”) absorption (determined by absorption cross-section and photon flux density of the excitation). Absorption cross-section and relaxation rate are quantities which sensitively depend on fluorophore electron structure and the microenvironment.

Simplified energy level scheme of melanin in melanosomes of nevi and of melanoma illustrates the process responsible for fluorescence spectra in the excited state: radiationless relaxation in nevi k (N) and in melanoma k (M). The relaxation includes ultrafast Franck–Condon relaxation (tuning of the core configuration to the changed excited-state electron distribution) and nonradiative vibrational relaxation.

The spectrum of normally pigmented skin (Figure 3a) is a superposition of melanin fluorescence with a maximum at about 500 nm with the NAD(P)H-dominated autofluorescence (maximum at about 470 nm, obtained from oculo-cutaneous skin tissue). The ratio of the two components determines the position of the resulting fluorescence maximum [20]; on low pigmented Fitzpatrick skin type 1, both bands appear separately (D. Leupold et al., submitted). Melanin fluorescence from the nevomelanocytes of benign or dysplastic nevi (Figure 3b,c) shows a distinctly different, flatter, and red-shifted spectral profile than fluorescence from melanocytes. An obvious cause of the difference could be the lacking discharge of melanosoma out of melanocytes (shutdown of the dominant influence of keratinocytes) and the changing structural alignments of the melanin π-systems that determine fluorescence (“π-stacking”). The red shift of the fluorescence spectrum changes into the characteristic curve according to Figure 3d in melanoma: a constant increase in intensity from 440 nm to 650 nm. Mathematically described, this means that the first derivative of the spectral fluorescence course dI_F (lambda)/d (lambda) for melanoma is a horizontal straight line between 440 and 650 nm. The spectral fluorescence course of dysplastic nevi (Figure 3c) differs from that of melanoma by a reduced increase above about 570 nm (this means a drop in the constant value of the first derivative above 570 nm). Benign nevi are characterized by a zero crossing of the first derivative in the range between about 530 and 550 nm.

Dermatofluoroscopy of skin from Caucasian patients (Fitzpatrick type 2, 3) in vivo. The four representative classes of melanin-dominated spectra: (a) normal pigmented skin (class 4), (b) benign nevus (class 3), (c) dysplastic nevus (class 2), and (d) melanoma (class 1).

Such dermatofluoroscopic investigations were carried out on normal pigmented skin, and benign and dysplastic nevi, as well as melanomas of more than 500 patients of Caucasian origin. This resulted in a database of several tens of thousands of spectra, the vast majority of which can be assigned to one of the following four classes of melanin-dominated fluorescence spectra: melanosomes of melanocytes (Figure 3a, class 4), melanosomes of nevomelanocytes of benign nevi (Figure 3b, class 3), melanosomes of nevomelanocytes of dysplastic nevi (Figure 3c, class 2), and melanosomes of melanoma cells (Figure 3d, class 1). It is important to note that these four types of spectra capture all measured cutaneous melanin fluorescence. This implies that class 1 spectra display a fingerprint of melanoma cells, regardless of the melanoma subtype. This is valid for the main melanoma subtypes studied so far: in situ, superficial spreading, nodular, lentigo maligna, and acrolentiginous [24]. It also means that malignant melanocytic degeneration of a nevus toward melanoma is always represented by classes 2 and 1. The latter is also confirmed by our follow-up measurements of nevi over several years [25]. The automatic assignment of measured fluorescence spectra to one of these four classes is based on the minimum of the root-mean-squared difference (RMSD) between the measured curve and each of the model curves. The RMSD has a fixed upper limit, and spectra which exceed this limit elude this classification (e.g., hairs, marker fluorophores, or impurities). Further details of the automated assignment of the spectra measured with derma FC were previously described [23].

The spectral analysis is concentrated on the range between 430 nm and 650 nm. An increase of intensity below 430 nm results from second harmonic generation (SHG) in collagen. The signal in the range above 650 nm stems from a further nonlinear optical effect that it is not considered here.