Usefulness of near infrared spectroscopy to monitor the extent of heat treatment in fish meal

The objective of this study was to explore the use of near infrared (NIR) spectroscopy and chemometrics to monitor the degree of heat treatment of fish meal. Six batches of fish meal (approximately 500 g) were split in sub‐samples of 50 g and heated at constant temperature (60 ± 5 °C) for different periods of time (15 and 30 min, 1, 2, 3, 4, 5, 6, 8, 48 and 72 h) in a force air oven and scanned in the NIR region (1100–2500 nm) in a monochromator instrument in reflectance. Principal component analysis (PCA), stepwise discriminant analysis (SLDA) and partial least square regression (PLS) models were used to interpret, classify and predict the extent to heat treatment in fish meal samples. The SLDA models correctly classified 80% and 100% of fish meal samples belonging to the untreated fish meal and after 4, 5 and 6 h of heat treatment. However, samples heated for 30 min, 1, 2 and 3 h yield poor classification rates (less than 50%). This study demonstrated the potential ability of NIR spectroscopy to predict and classify the extent of heat treatment during the production of fish meal. However, further research must carry out in order to validate the NIR calibrations to predict the degree of heat treatment in fish meal expose to a shorter time.