Fusion of DSC and FTIR data with physicochemical profiling to distinguish berry seed oils by extraction methods

Growing concerns over the use of n -hexane in oil extraction, mainly due to its neurotoxicity, have led the European Food Safety Authority (EFSA) to recommend re-evaluating its regulatory limits in food. This highlights the need for reliable authentication methods. In this study, a novel data fusion approach combining differential scanning calorimetry (DSC) melting curves and Fourier transform infrared (FTIR) spectra was proposed for the rapid authentication of berry seed oils by extraction methods. The effects of three extraction techniques: cold pressing, n -hexane, and supercritical CO₂, on the physicochemical, thermal, and spectral profiles of blackcurrant, raspberry, and strawberry seed oils were evaluated. Distinctive FTIR features were observed at wavenumbers 2931, 1726, 1408, and 1046 cm⁻¹, while the discriminatory DSC features corresponded to the peak temperatures. SIMCA classification models based on low-level DSC-FTIR data fusion were developed to distinguish ‘oil types and extraction methods’ at one run. The models showed promising performance with 81% overall accuracy in calibration and 77% in validation, which was higher than for single-method models i.e. FTIR (68%, 67%) and DSC (65%, 64%). These results demonstrated the potential of this method for rapid oil screening, supporting authentication protocols for regulatory compliance and label transparency in food industry • Three types of berry seed oils were extracted using three different methods • DSC melting curves, FTIR spectra and compositional data were collected • Different extraction methods affected oils composition except fatty acids content. • Low-level data fusion of DSC curves and FTIR spectra improved the classification results