Estimation of irradiation doses in chicken samples using a reaction-based fingerprinting method

Abstract

Food irradiation on an industrial scale calls for the development of rapid and inexpensive methods for the dose estimation after irradiation. An emerging solution to this problem is a reaction-based optical sensing strategy that is based on monitoring dose-dependent indicator reactions. In this study, raw chicken breast samples from three producers were irradiated with 1 MeV accelerated electrons, extracted with water for 24 h, and introduced into reactions of carbocyanine dyes with H₂O₂ or hypochlorite. The absorbance and fluorescence of the reaction mixtures in different spectral ranges were measured photographically as a function of time. Supervised machine learning methods allowed to confidently discriminate between the samples irradiated with 250, 1000, and 5000 Gy and non-irradiated samples provided that the samples irradiated with known doses were from the same producers as the unknown ones. Dose estimation for samples from an unknown producer could be implemented by constructing a database using samples from a larger number of producers.