PARAFAC analysis of front-face fluorescence data: Absorption and scattering effects assessed by means of Monte Carlo simulations

Three-way fluorescence data originating from mixtures of fluorophores embedded in turbid media such as biological media get strongly modulated by wavelength dependent absorption and scattering phenomena. Thus the consistent resolution and quantitative determination of the mixture becomes a difficult task. In this study two chemometric methodologies frequently used to deal with this type of data were applied to fluorescence simulated data sets qualitatively similar to those measured in biological samples: Parallel Factor Analysis (PARAFAC) that does require the fulfillment of trilinearity, and multivariate curve resolution-alternating least squares (MCR-ALS) which decomposes the data according to a model lacking this structure. Monte Carlo simulations were used to simulate fluorescence excitation-emission matrices (EEMs) of known fluorescent mixtures under separated and simultaneous variations of the absorption parameter mu(a) and the scattering parameter mu(s). PARAFAC and constrained MCR-ALS models were then fitted to the simulated data. Both algorithms failed the recover the true profiles. The results obtained with PARAFAC and MCR-ALS models are similar and the recovered profiles exhibit severe distortions due to the absorption and scattering effects. Finally, qualitative and quantitative effects of the absorption and scattering on the fluorescence data were assessed and discussed. (C) 2012 Elsevier B.V. All rights reserved.