TWO-DIMENSIONAL FLUORESCENCE SPECTROSCOPY: IMAGE ANALYSIS TECHNIQUES AND APPLICATIONS IN BACTERIAL IDENTIFICATION (DATA REDUCTION, PATTERN RECOGNITION, FOURIER TRANSFORM).

摘要

Two-dimensional fluorescence spectroscopy is a technique whereby fluorescence spectra are recorded in matrix format as a function of excitation and emission wavelengths. This data format is very useful for the examination of mixtures of fluorophores, especially when advanced methods of data reduction are applied to the interpretation of the emission excitation matrices (EEMs). Although conventional fluorescence instrumentation can be used to acquire EEM formatted spectra, the use of novel instruments incorporating various types of image detectors can result in a significant reduction in the time required to record an EEM. Therefore, throughout most of this dissertation, two novel instruments have been used, i.e., a video fluorometer (VF) and a portable multichannel fluorometer (PMF).;In addition to developing new methods for the interpretation of EEMs, two-dimensional spectroscopy has been applied to the identification of bacterial samples. The first method involves a study of the interaction of bacteria with mixtures of fluorescent dyes. This is known as mixed dye fluorometry (MDF). The primary goal in this portion of the dissertation work is to develop an MDF method capable of reproducibly differentiating between specific test organisms. The gram negative organisms Escherichia coli and Legionella pneumophila were used to demonstrate this goal.;The second approach to bacterial identification explored in the present dissertation is the extraction of naturally occurring fluorescent pigments from bacteria. The VF has been used to record the EEMs of the extracted pigments, and the pattern recognition methods developed in this dissertation are applied to the identification of the micro-organisms which produced the fluorescent pigments. This methodology has been applied to an investigation of the taxonomical relationships between species of Flavobacterium and Sphingobacterium. (Abstract shortened with permission of author.).;A major portion of this dissertation concerns the development of image analysis techniques based on the two-dimensional Fourier transform. Novel mathematical methods and computer algorithms have been developed which allow the application of the Fourier transform for: (1) smoothing and edge enhancement in EEMs; (2) quantitative evaluation of noisy single component and multicomponent EEMs; (3) pattern recognition for the automatic identification of unknown samples based on their EEMs; and (4) determination of the number of components contributing to unknown EEMs.