Fiber optics fluorescence fingerprint measurement for aerobic plate count prediction on sliced beef surface

• Fluorescence Fingerprint (FF) coupled with fiber optics system was developed.
• FF measurement and aerobic plate count (APC) were performed on beef samples.
• Partial least squares regression model was established for APC prediction using FF.
• APC could be accurately predicted with the error of prediction of 0.831 log CFU/cm2.

The ability to predict the aerobic plate count (APC) on a meat surface by rapid, nondestructive, and at-line measurement using fluorescence fingerprint (FF) spectroscopy coupled with fiber optics is reported here. The FF of beef slices stored for 3 days aerobically at 15 °C was measured using fiber optics at excitation wavelengths ranging from 200 nm to 500 nm and emission wavelengths ranging from 200 nm to 900 nm, which was followed by an APC of the same samples. The APC was successfully predicted by partial least squares regression (PLSR) of the measured FF dataset with a coefficient of determination of 0.831. From the variable importance in projection (VIP) score and the sign of the regression coefficient obtained by PLSR, it was indicated that tryptophan, NAD(P)H, Vitamin A, porphyrins, and flavins changed over time and were correlated with predicted APC. Rapid, nondestructive, and at-line analysis of the APC using FF spectroscopy coupled with fiber optics was shown to be applicable.