Vis-NIR spectroscopic and chemometric models for detecting contamination of premium green banana flour with wheat by quantifying resistant starch content

Abstract

This study investigated the effect of nine different wheat adulteration levels (10 %, 20 %, 30 %, 40 %, 50 %, 60 %, 70 %, 80 % and 90 % added wheat) on the concentration of resistant starch (RS) of green banana flour (GBF). The study further evaluated the potential application of visible to near infrared (Vis-NIR) spectroscopy with multivariate analysis to detect changes in the concentration of resistant starch of GBF caused thereof by wheat adulteration. Principal component analysis (PCA) and partial least squares (PLS) regression independently paired with 2nd derivative Savitsky-Golay (with 21 smoothing gaps, 2nd order polynomial); Detrend and the combination of these spectral pre-treatments were applied to compare the distribution of spectral data; model and predict the concentration of RS. A significant reduction trend in the concentration of RS of GBF samples was observed as the advancement of wheat adulteration from 38.65 ± 1.27 g/100 g (pure GBF) to as low as 5.37 ± 0.47 g/100 g (with 90 % added wheat). PCA was able to clearly group samples and gave 93 % accuracy based on RS concentration variation. The optimal PLS models obtained after the combination 2nd derivative Savitsky-Golay (with 21 smoothing gaps, 2nd order polynomial) + Detrend demonstrated high accuracy with the coefficient of determination for prediction (R2p) of 0.97; root mean square error of prediction (RMSEP) of 2.43; residual predictive deviation (RPD) of 6.24 and a range error ratio (RER) of 14.27. Based on the research findings, wheat adulteration is a nutritional threat to the production and marketing of green banana flour. There is a strong potential for the tested Vis-NIR technique to rapidly monitor banana flour nutritional changes or deteriorations caused by wheat on RS concentration. This research could provide the banana flour industry with a novel quality index to determine GBF authenticity.