Pattern Recognition of Two-Dimensional Fluorescence Data Using Cross-Correlation Analysis

Abstract

An algorithm for spectral matching of excitation-emission matrices (EEMs) is reported. The pattern-recognition method developed operates completely in the frequency domain after Fourier transformation of unknown and standard EEMs. Spectral matches are evaluated by correlation and intervector distances between Fourier spectra of EEMs. Application of this method to single-component EEMs demonstrates the utility of pattern recognition for the identification of spectrally similar anthracene derivatives. In a study of multicomponent EEMs, mixtures of polynuclear aromatics (PNAs) are identified by comparison to a standard set of mixtures, and eigenvector deconvolution is coupled with pattern recognition for the identification of individual components in two-component mixtures. In addition, the feasibility of applying automatic pattern recognition to the identification of bacteria based on their uptake of fluorescent dyes is explored.