SPECTROPHOTOMETRIC IMAGING OF CUTANEOUS PIGMENTED LESIONS - DISCRIMINANT-ANALYSIS, OPTICAL-PROPERTIES AND HISTOLOGICAL CHARACTERISTICS

Citation
S. Tomatis et al., SPECTROPHOTOMETRIC IMAGING OF CUTANEOUS PIGMENTED LESIONS - DISCRIMINANT-ANALYSIS, OPTICAL-PROPERTIES AND HISTOLOGICAL CHARACTERISTICS, Journal of photochemistry and photobiology.B, Biology, 42(1), 1998, pp. 32-39
Citations number
20
Categorie Soggetti
Biophysics,Biology
ISSN journal
10111344
Volume
42
Issue
1
Year of publication
1998
Pages
32 - 39
Database
ISI
SICI code
1011-1344(1998)42:1<32:SIOCPL>2.0.ZU;2-P
Abstract
Reflectance imaging spectroscopy at 17 selected wavelengths in the ran ge 470 - 1040 nm has been applied as a method to discriminate melanoma from other pigmented cutaneous lesions. Reflectance images are acquir ed for 18 primary melanomas and 33 benign naevi in 44 patients. From e ach spectral image, four parameters related to lesion reflectance (mea n value of reflectance, variegation, area and roundness uf lesion) are derived at the corresponding wavelength. A discriminant function betw een the two groups of lesions is determined by using a stepwise discri minant analysis, resulting in a test with a sensitivity of 89 % and a specificity of 88 % Since analyses are carried out on the same data se t used to obtain the discriminant function, our results should be inte rpreted with caution. Moreover, in an attempt to assign a physical and /or physiological meaning to the lesion image descriptors, some histol ogical features ( i.e., lesion thickness, degree and uniformity of pi mentation) have also been analysed, and their correlations with the im age descriptors investigated. Neither degree nor uniformity of pigment ation can fully explain the variations observed in the lesion image de scriptors. It is hypothesized that the presence of pathologically rela ted biological components such as lymphocytes and highly aggregated ce lls may play an important role. The actual performance of the spectrop hotometric imaging system should be proven in an additional, unselecte d number of cases before being used as a diagnostic adjunct for physic ians in the clinical detection of melanoma. (C) 1998 Elsevier Science S.A.