Detection of melanomas by digital imaging of spectrally resolved ultraviolet light-induced autofluorescence of human skin

Digital images of autofluorescence excited with 366 nm Hg line were recorded in a narrow 475 nm band for 408 pigmented lesions of the skin (90 melanomas, 205 common melanocytic and dysplastic naevi, 113 lesions of different kinds) and analysed photometrically with respect to spatial distribution of intensity to differentiate between melanomas and other melanocytic lesions. Earlier reports describing patterns of intensity distributions characteristic for melanomas have not been confirmed in this study. However, our evaluations showed that an algorithm based on ratios of maximum intensity recorded outside the lesions and minimum intensity found within them, allows melanomas to be detected with a sensitivity of 82.5%, a specificity of 78.6% and a positive predictive value of 58.9% (melanomas versus common and dysplastic naevi) or 76.7% (melanomas versus other pigmented lesions). The method is now being tested in a multicentre study involving three groups in three different cities in Poland, (C) 1998 Elsevier Science Ltd. All rights reserved.