In vivo Raman and autofluorescence study of the pigmented skin neoplasms

Abstract This paper studies classification of the pigmented skin neoplasms on the basis of the spectral Raman and autofluorescence features. The spectra registration was performed by means of the portable spectroscopic system with 785 nm laser radiation. The analysis of the spectral data was carried out using a projection on the latent structure with linear discriminant analysis. Using regression analysis the correlations between Raman and autofluorescence data were estimated. The variable importance in projection analysis allows one to find out important spectral information to differ spectra of the different neoplasm types. The use of combination of the Raman/autofluorescence data and projection method made it possible to classify pigmented skin neoplasms with 81% accuracy.