Application of near infrared spectroscopy to detect aflatoxigenic fungal contamination in rice

The objective of this research was to apply the near infrared spectroscopy (NIRS), with a wavelength range between 950 and 1650 nm, to determine the percentage of fungal infection found in rice samples. The total fungal infection and yellow-green Aspergillus infection, which is often indicative of aflatoxigenic fungal infection, are the focus of this research. Spectra were obtained on 106 rice samples, by reflection mode, including 90 naturally contaminated samples, and 16 artificially contaminated samples. Calibration models for the total fungal infection were developed using the original and pretreated absorbance spectra in conjunction with partial least square regression (PLSR). The statistical model developed from the untreated spectra provided the greatest accuracy in prediction, with a correlation coefficient (r) of 0.668, a standard error of prediction (SEP) of 28.874%, and a bias of −0.101%. For yellow-green Aspergillus infection, the most accurate predictive statistical model was developed using a pretreated (maximum normalization) NIR spectra, with the following statistical characteristics (r = 0.437, SEP = 18.723% and bias = 4.613%). Therefore, the result showed that the NIRS could be used to detect aflatoxigenic fungal contamination in rice with caution and the technique should be improved to get better prediction model. However, there is an evident from NIR spectra that the moisture and starch content in rice affects the overall extent of fungal infection. ► NIRS was applied for detection of fungi and potentially aflatoxigenic fungi in rice. ► Moisture and starch contents in rice affect the overall extent of fungal infection. ► NIRS could better predict the total fungal infection than yellow green Aspergillus infection.