Oat flour identification and quantification for multigrain bread - The successful combination of hyperspectral...

Ensuring the authenticity of commercial bread remains a critical concern because fraudulent cereal blends and the substitution of cheaper flours in premium bread can mislead consumers and compromise product integrity. This study describes a novel analytical methodology for the evaluation of oat flour content in cereal bread using the synergy between hyperspectral imaging, information theory, and chemometrics. By evaluating spectral data captured from hyperspectral imaging in short-wave infrared and visible-NIR ranges, Shannon's entropy and information index were employed as objective criteria for selecting the most informative spectral range, ensuring optimal classification performance. Multivariate methods such as partial least squares discriminant analysis and support vector machine were introduced to differentiate single cereal breads (wheat, spelt, oat, and rye) with an accuracy of up to 0.96. Moreover, a novel quantification approach based on previous methodologies (quantification based on pixel counting) was applied to indirectly estimate the proportion of oat flour in binary bread blends. Results show that integrating information theory with hyperspectral imaging can increase efficiency by guiding spectral selection and improving classification and quantification outcomes. This methodology offers a robust alternative for bread conformity assessment, addressing the lack of standardized post-production methods and the need to align with current regulatory frameworks.