Rapid Differentiation of Commercially-Available Soy Sauces using Near-Infrared Spectroscopy

Philippine Journal of Agricultural and Biosystems Engineering; Jeric Rosales; Kevin  Yaptenco; Mae Joanne Aguila; Paul Armstrong

doi:10.48196/015.02.2019.01

Philippine Journal of Agricultural and Biosystems Engineering
Jeric Rosales University of the Philippines Los Baños
Kevin Yaptenco University of the Philippines Los Baños
Mae Joanne Aguila University of the Philippines Los Baños
Paul Armstrong Center for Grain and Animal Health Research Manhattan

DOI: https://doi.org/10.48196/015.02.2019.01

Keywords: NIR spectroscopy, spectral fingerprint, soy sauce, differentiation, chemometrics

Abstract

Near-infrared (NIR) spectroscopy was used to differentiate between four leading brands of soy sauce in the Philippines; the local brands included Datu Puti (DP), Marca Piña (MP), and Silver Swan (SS). A naturally-brewed imported brand, Kikkoman (KK), was used for comparison. Spectral data of soy sauce samples were acquired in absorbance mode from 900 nm to 1700 nm. First– (1D) and second– (2D) derivative transforms of preprocessed spectral data were analysed by a combination of principal component analysis (PCA) – linear discriminant analysis (LDA), and by partial least squares regression (PLSR). Spectral data with no derivative (ND) transformation was analysed for comparison. Preprocessing methods included mean scatter correction (MSC), Savitzky-Golay (SG) smoothing, and standard normal variate (SNV); these were used in different combinations to determine the best treatment. PCA-LDA results showed that the MSC-SG-ND treatment gave the best result overall with 97.6% of samples classified correctly, and 96.9% of variability explained by two principal components. For PLSR, data preprocessed using SNV-SG-2D transformation gave the best results after cross-validation, with a 9-factor model with root mean square error (RMSE) and coefficient of determination (R2) of 0.114 and 0.989, respectively. Using an independent set of soy sauce samples, the PLSR model was able to correctly identify 100% of DP, KK and MP; however, all SS samples were misclassified as DP or MP. This underscores the need for extensive sampling to build up a spectral library of soy sauce from different brands and sources; this will enable rapid testing of products for authenticity, quality and safety.

Citation:

ROSALES, J., YAPTENCO, K., AGUILA, M. J., & ARMSTRONG, P. (2019). Rapid Differentiation of Commercially-Available Soy Sauces using Near-Infrared Spectroscopy. Philippine Journal of Agricultural and Biosystems Engineering, 15(2), 3–12. https://doi.org/10.48196/015.02.2019.01