MULTIVARIATE STATISTICAL ANALYSES IN OPTICAL SPECTROSCOPIC SENSOR ARRAY FOR DISCRIMINATING ORGANIC AND BIOLOGICAL SAMPLES

Abstract

Due to environmental and public health concerns, detection of organic materials in biological samples, such as proteins contained in dairy products, surfactants extensively used in cosmetic and pharmaceutical products, volatile organic compounds (VOCs) contaminated in soft drinks, and foodborne pathogens, is of importance for economic and legal issues. Multivariate statistical analyses applied to the optical spectroscopic response which is induced by interactions between sensing element and organic materials allow a possible discrimination of biological samples. The pattern recognition of fluorescence responses upon addition of commercial milk samples to fluorophore solutions are analyzed by linear discriminant analysis (LDA). Milk samples according to their thermal processes; i.e. pateurized milk, sterilized milk, UHT fresh milk and recombined milk (UHT milk having milk powder) are discriminated with 100% classification accuracy. LDA is also successfully applied for the discrimination of eight foodborne pathogens. Anionic surfactants are successfully identified by using principal component analysis (PCA) applied to percentage of colorimetric response obtained from the UV-Vis spectra of polymerized diacetylenes in the presence of anionic surfactants. PCA is also applied to the RGB color changes profile plot obtained from images of paper-based polydiacetylene colorimetric sensor array developed for the detection and identification of VOCs.