Application of computerized image analysis in pigmentary skin diseases

Melanocyte number and the amount of melanin pigment are related to diagnosis and treatment of pigmentary skin diseases. Various histologic methods are used, such as Fontana-Masson stain for melanin pigment or immunohistochemical stain for melanocytes. Recently, computerized image analysis has been applied to many fields to avoid interobserver bias. In this study, we applied a computerized image analysis to assess the melanin content and melanocyte density of human epidermis.

We evaluated the skin biopsy specimens (paraffin blocks) from normal human skin (33 +/- 6.6, n = 11) and diseased skins; vitiligo (32 +/- 10.0, n = 8), melasma (35 +/- 8.6, n = 11), and lentigo senilis (40 +/- 7.2, n = 11) (mean age +/- SD). Each specimen was stained with Fontana-Masson for melanin pigments and immunohistochemical method for melanocytes. Quantitative analysis of melanin pigment and melanocyte number (density) were investigated through two methods: (1) two dermatologists measured the visual scales; and (2) computerized image analysis was used to measure melanin content indices (MCI). The data were evaluated using one-way ANOVA.

The visual scale of the Fontana-Masson stain was the highest for lentigo senilis (3.8 +/- 0.40), followed by melasma (2.6 +/- 0.67), normal skin (1.8 +/- 0.60) and vitiligo (0) (P < 0.05). These findings were consistent with objective measurements made by computerized image analysis. MCI values were 120.3 +/- 20.74 for lentigo senilis, 81.1 +/- 19.27 for melasma, 45.5 +/- 16.92 for normal skin, and 0.3 +/- 0.30 for vitiligo in decreasing order (P < 0.05). MC/1E (melanocyte number per 1 mm epidermis) was about two fold larger in lentigo senilis (18.1 +/- 8.92) than melasma (9.7 +/- 2.40) or normal skin (9.3 +/- 2.67) (P < 0.05). MC/1B (melanocyte number per 1 mm basal layer) was about 1.5 fold higher in lentigo senilis (13.5 +/- 4.17), compared to normal skin (9.0 +/- 3.55) (P < 0.05). Melasma showed increased melanocyte numbers compared to normal skin, but it was not statistically significant (P > 0.05).

We believe this computerized image analysis could be useful tool for diagnosis and comparison of interval changes in pigmentary diseases like melasma or lentigo senilis by quantifying melanin pigments or melanocytes in skin biopsy specimens.