From Benchtop to Handheld Systems: Calibration Transfer for On-Site NIR System Butter Quality Monitoring

Butter quality assessment requires rapid and reliable portable analytical tools that enable on-site analysis beyond specialized food laboratories. This study evaluated the transfer of calibration models developed on a benchtop Near-Infrared (NIR) instrument (DS25) to three handheld NIR devices (Aurora, JDSU and ISC) for the prediction of butter compositional traits and fatty acid profile. Ninety-nine butter samples from different milk production systems were used for spectral acquisition and calibration development. Different calibration transfer (CT) strategies, including spectral bias correction (SBC), direct standardization (DS), piecewise direct standardization (PDS), Shenk-Westerhaus (S-W) standardization and dynamic orthogonal projection (DOP), were combined with Partial Least Squares (PLS) regression and evaluated on a hold-out validation set. Across both hold-out validation and prediction-based consistency analyses, Aurora and JDSU showed the most robust performance relative to the DS25 benchmark and the multi-instrument consensus, particularly for moisture, fat content, and major fatty acids (FA) groups, whereas ISC showed lower and more variable performance. The outcomes suggested that handheld NIR spectroscopy, supported by appropriate calibration transfer and limited target-instrument updating, can be used for rapid and decentralized butter quality monitoring. However, transfer success was instrument-, variable- and method-dependent, confirming the need for device-specific validation before routine application.