Combining complementary spectral techniques for robust brix prediction in strawberry fruit

Non-destructive techniques for fruit analysis are essential for post-harvest handling and processing of fresh produce. Optical methods based on vibrational spectroscopy enable rapid evaluation of key quality parameters such as °Brix and dry matter. Among these, near-infrared (NIR) spectroscopy is most used, however, Raman spectroscopy also provides valuable information on the chemical composition of fresh fruit. NIR and Raman are complementary techniques where NIR measures light absorption properties, while Raman captures inelastic light scattering and is minimally affected by water. This study, for the first time, investigates the potential of combining complementary information from NIR and Raman spectroscopy to predict °Brix in fresh strawberries. Data fusion was achieved using multiblock partial least squares regression, and the resulting fusion models were compared with models based on individual techniques. Model performance was evaluated using an independent dataset from a cultivar not included in the calibration set and harvested at a different time point. Fusion of NIR and Raman data produced models with the lowest root mean square error of prediction (RMSEP = 0.58 °Brix) compared to single-technique models i.e., RMSEP = 0.72 °Brix and 0.85 °Brix for NIR and Raman, respectively. These findings open new opportunities for multi-sensor fruit quality assessment, demonstrating that combining complementary techniques such as NIR and Raman can significantly improve prediction accuracy and robustness.