Use of NIR spectroscopy to predic rheological characteristics of common wheat flours

To evaluate flour quality, non-traditional approaches such as the Kieffer and the GlutoPeak tests have been proposed when only a small quantity of sample is available, as in the breeding program (see “Characterization of common wheat flours through GlutoPeak analysis and Kieffer test”). Both procedures require a small amount of sample, 50 g and 10 g respectively. Moreover, GlutoPeak test is very rapid (it takes no more than 10 min) and the manual involvement is very limited. Even more interesting is the possibility of using a non-destructive method such as NIR spectroscopy to evaluate the rheological parameters of flours and consequently their suitability to be used in a specific production (Cocchi et al., 2005). The aim of this research is to verify the possibility of predicting the rheological characteristics of common wheat flours by using NIR spectroscopy, considering in the prediction both traditional indices and parameters obtained from the unconventional methods previously mentioned. In this research, 87 samples of wheat flours, provided by five different industrial mills and characterized by different final uses, were analyzed. NIR spectral data were collected in reflectance mode using a FT-NIR spectrometer equipped with an integrating sphere on the raw samples at the resolution of 8 cm−1. Spectra, collected in the spectral range 12500-4000 cm-1, were pre-processed using SNV and derivative transformations. The spectral data were correlated with rheological parameters by Principal Component Analysis (PCA) and PLS regression algorithm. PLS regression models developed for some rheological parameters were characterized by high R2 values, both in calibration and in cross-validation and by low errors. Best results were obtained for farinographic water absorption (R2 = 0.85 in calibration; R2 = 0.84 in cross-validation) and maximum torque from the Glutopeak test (R2 = 0.85 in calibration; R2 = 0.84 in cross-validation). Acceptable results (R2  0.7) were also achieved for predicting the Alveograph W and the Maximum Resistance to extension, evaluated either with the Extensograph or with the Kieffer test. In conclusion, although the results presented here require further investigations, this study provides a good demonstration that NIR spectroscopy can be considered as a valid and reliable tool to predict many rheological characteristics useful for wheat breeding programs, also taking in consideration its non-destructive nature and the possibility to perform analysis on intact grain samples. BIBLIOGRAFY Cocchi M., Corbellini M., Foca G., Pagani M.A., Lucisano M., Tassi L., Ulrici A., 2005. Classification of bread wheat flours in different quality categories by a wavelet-based feature selection/classification algorithm on NIR spectra. Anal. Chim. Acta, 544, 100-107.