Evaluation of a Fourier-Transform-Based Pattern-Recognition Algorithm for Two-Dimensional Fluorescence Data

Abstract

A pattern-recognition algorithm for two-dimensional fluorescence data previously reported is critically evaluated. The three spectral matching criteria—sum of the absolute value of the imaginary coefficients of the frequency-domain correlation function, sum of the absolute value of the real-negative coefficients of the frequency-domain correlation function, and the intervector distance between the abbreviated Fourier transforms of two spectra—are calculated. Spectra simulated with a computer as well as data acquired with a video fluorometer are examined. Results indicate that all three parameters are sensitive to changes in peak position, peak width, relative peak height, and intensity of background noises.