Improved prediction of wheat baking quality by three novel approaches involving spectroscopic, rheological and analytical measurements and an optimized baking test

Abstract

Baking quality, defined as loaf volume, is one of the most important quality attributes of wheat. An accurate and rapid determination is of great interest for the wheat supply chain. However, this remains difficult to date, because reported predictions based on other wheat characteristics (e.g. protein content) or flour spectroscopy are poor. This study investigates three novel approaches to improve the prediction of specific loaf volume determined by an optimized mini-baking test. The predictions are based on a large variety of rheological and analytical data as well as fluorescence, near-infrared (NIR) and Raman spectroscopy of flour and flour fractions. Furthermore, the influence of data fusion on the predictions is investigated. All three approaches presented promising results and showed great potential for practical application with R2CV > 0.90 for various regression models. For example, the combination of farinograph data with solvent retention capacity data or NIR flour spectra yielded R2CV of 0.91 in both cases. Combining Raman spectra of the < 32 μm and 75–100 μm fractions as well as NIR spectra of gluten, flour and starch both also yielded R2CV of 0.91. The results underline that loaf volume is a complex quality characteristic that can be better predicted when different data types are combined. Different rheological and analytical tests and different spectroscopic methods capture specific wheat quality characteristics that have different relations to baking volume and can therefore provide complementary information for improved predictions. Furthermore, the importance of rheological tests (especially farinograph, extensograph, alveograph) and the baking procedure for the prediction of baking quality are emphasized.