Digital dewaxing of Raman spectral images of paraffin-embedded human skin biopsies based on ICA and NCLS

A digital automatic dewaxing method of Raman spectral images acquired on paraffin-embedded human skin biopsies is proposed in this paper. This method is based on preprocessing steps specific to Raman spectra, on the estimation of the paraffin spectra by the Principal Component Analysis (PCA) and Independent Component Analysis (ICA), and on the estimation of the paraffin concentrations by the Nonnegatively Constrained Least Squares (NCLS) method. The proposed method allows the automatic extraction of skin signatures which depend on the acquisition points. By this way, further classification of a skin biopsy into tumoral, peritumoral and normal zones should be possible.