Application of Fourier-transform near-infrared spectroscopy (FT-NIRS) combined with chemometrics for evaluation of quality attributes of dried pomegranate arils

In this study, we investigated the usage of Fourier-transform near-infrared (FT-NIR) spectroscopy as a fast, non-destructive method. FT-NIR spectroscopy was used over a spectral range of 800-2500 nm to develop multivariate prediction models for physical, chemical, and phytochemical parameters of dried pomegranate arils (‘Wonderful’). Results from two different regression techniques, partial least squares (PLS) and support vector machine (SVM), were compared. Model development results showed varied success with statistics from PLS regression showing reliable prediction for pH (R²=0.86, RMSEP=0.13, RPD=2.38) and TSS/TA (R²=0.74, RMSEP=1.68, RPD=1.68). SVM performed better for the prediction of titratable acidity (R²=0.85, RMSEP=0.04, RPD=2.50) and color attributes for redness (a*) (R²=0.72, RMSEP=1.82, RPD=1.71) and Chroma (C*) (R²=0.70, RMSEP=1.99 RPD=1.77). In summary, SVM performed better than PLS regression in predicting quality attributes for died pomegranate arils. This study demonstrated that FT-NIRs with an SVM regression algorithm can be used as a non-invasive technique to evaluate key visual and sensory attributes of dried pomegranate arils.