Reagent-free detection of multiple allergens in gluten-free flour using NIR spectroscopy and multivariate analysis

Detecting allergenic ingredients in food products is a major health concern for many consumers, regulatory agencies, and the food industry. It is crucial to ensure that consumers with food allergies have accurate information about their food. DNA and protein-based allergen detection techniques are time-consuming, laborious, and require skilled technicians. In this study, a benchtop near-infrared (NIR) system and a filter-based NIR spectrometer associated with multivariate analysis were used as a rapid method to detect multiple allergenic ingredients in gluten-free flour. Partial least squares regression (PLSR) was combined with different spectral preprocessing methods to obtain an accurate predictive model. Only 9 dominant wavelengths were selected, and a better PLSR model was developed (R2p = 0.99, RMSEP=3.25%). This model was then compared with a similar model that utilized filter-based NIR data consisting of only 10 spectral bands. The PLSR model developed with the selected 9 wavelengths better predicted multiple allergenic ingredients in gluten-free flour than the filterbased NIR (R2p = 0.96, RMSEP=6.32%). This study revealed the efficacy of NIR and multivariate analysis for rapidly detecting multiple allergenic ingredients in gluten-free flour. The study also revealed that it is possible to develop a low-cost, miniature sensor using the selected bands to detect multiple allergens simultaneously.

Automated summary

In this study, a benchtop NIR system and a filter-based NIR spectrometer were used with multivariate analysis to rapidly detect multiple allergenic ingredients in gluten-free flour. Only 9 dominant wavelengths were selected and a better predictive model was developed, which outperformed the filter-based NIR method.