Smart technique for accurate monitoring of ATP content in frozen fish fillets using fluorescence fingerprint

Abstract

The aim of the present study was to develop a fast and nondestructive method based on fluorescence fingerprints (FFs) to predict the ATP content in frozen fish meat frozen at early stages after death using fillets of horse mackerel (Trachurus japonicus) as a model. Fifty-six fish were sacrificed instantly, stored in ice for different periods (0–48 h), and then filleted and frozen. The fluorescence fingerprints of the frozen fillet samples were acquired using fluorescence spectrophotometer with fiber probe installed inside a freezer. Subsequently, the ATP-related compounds of the same samples were determined using HPLC. Finally, four different models based on partial least squares (PLS) were developed to predict ATP contents from HPLC and the FFs data. The best PLS model with a correlation coefficient (R2) of 0.88 and root mean square error estimated by cross validation (RMSECV) of 0.97 μmol/g was obtained when the most important combinations of excitation-emission wavelengths were used for prediction. This methodology offers a simple and rapid approach to detect the ATP contents in frozen fish nondestructively without thawing the sample during the assessment that could be applied during any stage of fish marketing, facilitating quality control activities and the determination of fishery market price.