Quantitative analysis of ingredients of blueberry fruits by near infrared spectroscopy

Near infrared spectroscopy was used to estimate the contents of sugars, organic acids and anthocyanins in blueberry fruit. The contents of the fruit determined by high-performance liquid chromatography were used to construct models to estimate these ingredients by partial least-squares regression. The dependence of the accuracy of estimation on the mode of measurement, i.e. transmission mode vs. diffuse reflectance mode, and on the position of measurement on the fruit, i.e. the calyx or a position not on the calyx, was examined. The fruit contents were found to be estimated with practical accuracy by choosing the mode and the position of measurement appropriately for each target. Sugar and anthocyanin contents could be estimated from spectra measured in either mode at any fruit position. Estimation of the fruit contents by a model constructed from all spectra, regardless of the position of measurements, was also found to be possible. The organic acid content should be estimated from spectra measured in the diffuse reflectance mode by the model constructed for the specific position of measurements, which could be inferred by the principal-component analysis score plots of the observed spectra. The accuracy of the estimation of organic acid content based on spectra measured in transmission mode was found to be very low.