Detection and identification of fungal growth on freeze-dried Agaricus bisporus using spectra and olfactory sensors
Liuqing Wang,Qiuhui Hu,Fei Pei,Mariga Alfred Mugambi,Wenjian Yang
DOI: https://doi.org/10.1002/jsfa.10348
Abstract:Background: Fungal contamination in food products leads to mustiness, biochemical changes, and undesirable odors, which result in lower food quality and lower market value. To develop a rapid method for detecting fungi, hyperspectral imaging (HSI) was applied to identify five fungi inoculated on plates (Aspergillus niger, Aspergillus flavus, Penicillium chrysogenum, Aspergillus fumigatus, and Aspergillus ochraceus). Near-infrared (NIR) spectroscopy, mid-infrared (MIR) spectroscopy, and an electronic nose (E-nose) were applied to detect and identify freeze-dried Agaricus bisporus infected with the five fungi. Results: Partial least squares regression (PLSR) models were used to distinguish the HSI spectra of the five fungi on the plates. The A. ochraceus group had the highest calibration performance: coefficient of calibration (Rc 2 ) = 0.786, root mean-square error of calibration (RMSEC) = 0.125 log CFU g-1 . The A. flavus group had the highest prediction performance: coefficient of prediction (Rp 2 ) = 0.821, root mean-square error of prediction (RMSEP) = 0.083 log CFU g-1 . The ratio of performance deviation (RPD) values of all of the models was higher than 2.0 for the NIR, MIR, and E-nose results for freeze-dried A. bisporus infected with different fungi. The fungal species and degree of infection can be distinguished by principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) using NIR, MIR, and E-nose, as the discrimination accuracy was more than 90%. The NIR methods had a higher recognition rate than the MIR and E-nose methods. Conclusion: Principal component analysis (PCA) and PLSR models based on full spectra of HSI can achieve good discrimination results for these five fungi on plates. Moreover, NIR, MIR, and the E-nose were proven to be effective in monitoring fungal contamination on freeze-dried A. bisporus. However, NIR could be a more accurate method. © 2020 Society of Chemical Industry.