Quantitative Estimation of Component Amplitude in Multiexponential Data: Application to Time-Resolved Fluorescence Spectroscopy

摘要

Quantitative information of individual component contribution from multiexponential data is obtained by a reiterative, linear least-squares algorithm. Uncertainty in the parameter estimates, arising from uncertainty in the data and overlap in the response, are predicted from first principles. The analysis method includes weighting to account for the Poisson error distribution arising from shot-noise limited signals, which increases the accuracy of the amplitude estimates. While the algorithm is applicable to a variety of kinetic methods, it is applied in the present work to the analysis of time-resolved fluorescence decay curves. A fluorescence decay curve, written as a row vector, D, is decomposed into two factors: A, a column vector containing the amplitude contribution of each component, and C, a matrix which contains temporal behavioral of each component in its rows. The analysis uses linear least-squares to obtain estimates of A, which increases the efficiency by reducing the number of parameters which are searched. The theory of error in linear least-squares allows the uncertainty of the component amplitudes to be determined from the C matrix, derived from best estimates of the temporal behavioral of the sample. Keywords: Regression analysis; Reaction kinetics.(AW)