Smart screening of durum wheat: a comparative study of NIR portable sensors for protein and moisture content

Durum wheat (Triticum durum Desf.) is a strategic crop for the Italian agro-industrial sector, where rapid, non-destructive assessment of protein and moisture content is crucial for quality evaluation and processing performance. Portable Near-Infrared (NIR) spectrometers provide a promising approach for in-field and in-line analyses, overcoming limitations of conventional laboratory methods. This study evaluates two portable devices, NeoSpectra Scanner and PoliSPEC-NIR, for predicting protein and moisture in intact durum wheat kernels. Over 100 samples from various Italian regions were analyzed using different acquisition modes, and Partial Least Squares (PLS) regression models were developed and externally validated to predict protein content and moisture in samples. Both sensors demonstrated satisfactory predictive capabilities, with higher accuracy for protein and best results achieved by PoliSPEC-NIR in scanning mode (RMSEP 0.35 g/100 g for protein content prediction and 0.21 g/100 g for moisture prediction). Acquisition mode and surface coverage significantly influenced model robustness, highlighting the need for standardized measurement protocols. These findings support the operational feasibility of miniaturized NIR spectrometers for rapid, non-destructive quality monitoring along the durum wheat value chain.