Enhanced chemical classification of Raman images in the presence of strongfluorescence interference

Raman spectra and spectral images containing severe fluorescence interferen ce are analyzed by using a variety of correlation and classification algori thms, both before and after preprocessing with the use of the Savitzky-Gola y second-derivative (SGSD) method (and other related methods). Spectral cor relation coefficient, principal component, and minimum Euclidean distance a nalyses demonstrate superior suppression of background and noise interferen ce in Raman spectra when using SGSD preprocessing. The tested spectra inclu de fluorescence interference that is more intense than the Raman features o f interest and also contains broad background peaks that vary in shape and intensity from sample to sample. The high chemical information content of t he SGSD-processed Raman spectra is demonstrated by using quantitative compa risons of correlation coefficients in a series of synthetic Raman spectra w ith either different or identical large backgrounds. The practical utility of SGSD in chemical image classification is illustrated by using an experim ental Raman image of sugar microcrystals on substrates with large interferi ng background signals. The functional equivalence of SGSD and other windowe d preprocessing algorithms is discussed.