Abstract:

The objective of this work was to establish a rapid method for aflatoxin B1 detection in brown rice by attenuated
total reflectance-Fourier transform infrared spectroscopy (ATR-FTIR). A set of 16 brown rice samples were sterilized with
radiation, incubated after inoculation with toxigenic strains and stored under appropriate conditions (28 ℃ and 85% RH)
for 30 days, and the concentrations of AFB1 were determined. Then the samples contaminated by AFB1 were blended with
blank samples in various weight proportions. A total of 128 samples with concentrations ranging from 0 to 890.20 μg/kg
were obtained. The ATR-FTIR calibration results indicated that a classification accuracy of 83.3% was obtained by K
nearest neighbors (KNN) when the samples were divided into four groups according to AFB1 concentration. Quantitative
analysis model was built by partial least square regression (PLSR), which showed the best performance in predication, and
the regression coefficient of validation (rv), root mean square error of prediction (RMSEP) as well as RPD value, which
were 0.970, 70.8 μg/kg and 4.0, respectively. Overall, ATR-FTIR offers a feasible method for rapid screening of brown rice
samples contaminated by AFB1. However, further study is needed to improve the prediction power for samples with low
AFB1 concentrations.

Key words: attenuated total reflectance-Fourier transform infrared spectroscopy (ATR-FTIR), brown rice, aflatoxin B1 (AFB1), rapid detection