Exploring the potential of NIRS technology for the in situ prediction of amygdalin content and classification by bitterness of in-shell and shelled intact almonds

Abstract

Amygdalin is a cyanogenic compound found in almonds which gives them their bitter taste. For the almond industry, it is important to prevent the presence of bitter almonds in batches of sweet almond that can affect their commercialization and even consumer safety. This study sought to ascertain the viability of near infrared spectroscopy (NIRS), as a fast and reliable candidate for non-destructive and in situ quantification of amygdalin levels and for classification of almonds by bitterness, when analysed in bulk. With that purpose, in-shell and shelled sweet and bitter almonds were analysed in dynamic mode using two new handheld NIRS instruments. As a first step, the amygdalin levels in in-shell and shelled almonds were determined using modified partial least squares (MPLS) and local regression algorithms. Next, classification models for bitterness were made using partial least square discriminant analysis (PLS-DA). For the discrimination between sweet and bitter almonds, two strategies to set up the optimum threshold were studied: the mean value of the discriminant variables and the value calculated using the Receiver Operating Characteristic (ROC) curves. The results for measuring amygdalin in shelled almonds showed that NIRS technology, using both regression algorithms, is a robust technology for inspection purpose at an industrial level. Additionally, excellent performances were obtained for the classification models of the two in-shell and shelled almond groups analysed in bulk with both instruments, with better results when the threshold values obtained from the ROC curves were applied.