Rapid screening of apple juice quality using ultraviolet, visible, and near infrared spectroscopy and chemometrics: A comparative study

The aim of this study was to evaluate and compare potential of different optical spectroscopic techniques for quality assessment of apple juices. The calibration partial least squares (PLS) regression models were developed and optimized for determination of quality parameters of apple juice non-destructively from the NIR spectra of fruit or directly from the UV-VIS-NIR spectra of juices. The external models validation was performed using the test sample set. Calibration models based on the NIR spectra of fruit were characterized by high predictive ability for soluble solids content, SSC (R2P = 0.93, RPDP = 3.5), moderate for titratable acidity, TA (R2P = 0.81, RPDP = 2.1), pH (R2P = 0.82, RPDP = 2.1) and ratio of SSC to TA, SSC/TA (R2P = 0.72, RPDP = 1.8), and low for total phenolic content, TPC (R2P = 0.44, RPDP = 1.3) and total flavonoid content, TFC (R2P = 0.49, RPDP = 1.2). The performance of PLS models based on the juice spectra was dependent on the spectral region used for analysis. The models with the highest predictive ability for SSC (R2P = 0.95, RPDP = 5.0) and TA (R2P = 0.96, RPDP = 4.6) were obtained using UV-VIS-NIR spectra of juice. The models with the highest predictive ability for SSC/TA (R2P = 0.89, RPDP = 3.0) and pH (R2P = 0.91, RPDP = 3.3) were obtained using VIS-NIR spectra. The optimal model for TPC (R2P = 0.95, RPDP = 4.1) was obtained using UV-VIS spectra, and for TFC using UV spectra of juices (R2P = 0.80, RPDP = 2.5). The results may contribute to the development of method for fast quality control of juices.