Fluorescence sensor array for identification of commercial milk samples according to their thermal treatments

Abstract

Identification of processed milk is of importance for commercial and legal concerns. The fluorescence response patterns induced by fluorophore/protein interactions allow a possible discrimination of processed milk samples corresponding to their thermal treatment. The fluorescence responses of 4 fluorophores upon addition of commercial milk samples in 96-well plate are measured in the range of 400-600 nm using the excitation wavelength at 375 nm. The pattern recognition of the 53,126 fluorescence responses (4 fluorophores×41 wavelengths×4 thermally processed milks×3 brands×3 lots×3 bottles×3 repeats) are analyzed by multivariate statistical methods. Linear discriminant analysis (LDA) successfully recognizes the milk samples according to their thermal processing, i.e. pasteurized milk, sterilized milk, UHT fresh milk and recombined milk (UHT milk having milk powder), with 100% classification accuracy in a cross validation using a leave-one-out technique.

Keywords: Fluorescence sensor; Linear discriminant analysis (LDA); Milk protein; Multivariate statistical analysis; Principal component analysis (PCA); Protein analysis.