Predicting the physicochemical properties and geographical ORIGIN of lentils using near infrared spectroscopy

Calibration statistical descriptors for both whole and ground lentils using Near Infrared Spectroscopy (NIRS), combined with fiber-optic probe, are presented and discussed. The models were developed for estimating the weight, size, total raw protein, moisture, total fat, total fiber, and ash. Standard methods were used to determine compositional parameters of 42 samples of different varieties of lentils. The calibration curves show a wide range of validity for all parameters. The results showed excellent predictability for the determination of weight, fiber, and ash in whole lentils. However, size, moisture, and total fat were predicted satisfactorily in ground lentils. The total protein content could be predicted for both whole and ground lentils. Moreover, NIRS and Direct Partial Least Squares (DPLS) were used to determine whether a sample of lentils belonged to the Protected Geographical Indication (PGI) Lenteja de La Armuna or not. The results showed that 95% of the samples were correctly classified as belonging to a PGI. This result demonstrates that this technique allows the differentiation of samples from nearby regions.