Neural network pattern recognition of photoacoustic FTIR spectra and knowledge-based techniques for detection of mycotoxigenic fungi in food grains.

Fourier transform infrared photoacoustic spectroscopy (FTIR-PAS), a highly sensitive probe of the surfaces of solid substrates, is used to detect toxigenic fungal contamination in corn. Kernels of corn infected with mycotoxigenic fungi, such as Aspergillus flavus, display FTIR-PAS spectra that differ significantly form spectra of uninfected kernels. Photoacoustic infrared spectral features were identified, and an artificial neural network was trained to distinguish contaminated form uncontaminated corn by pattern recognition. Work is in progress to integrate epidemiological information about cereal crop fungal disease into the pattern recognition program to produce a more knowledge-based, and hence more reliable and specific, technique. A model of a hierarchically organized expert system is proposed, using epidemiological factors such as corn variety, plant stress and susceptibility to infection, geographic location, weather, insect vectors, and handling and storage conditions, in addition to the analytical data, to predict Al. flavus and other kinds of toxigenic fungal contamination that might be present in food grains.

Identification of Fourier transform infrared photoacoustic spectral features for detection of Aspergillus flavus infection in corn.

Aspergillus flavus and other pathogenic fungi display typical infrared spectra which differ significantly from spectra of substrate materials such as corn. On this basis, specific spectral features have been identified which permit detection of fungal infection on the surface of corn kernels by photoacoustic infrared spectroscopy. In a blind study, ten corn kernels showing bright greenish yellow fluorescence (BGYF) in the germ or endosperm and ten BGYF-negative kernels were correctly classified as infected or not infected by Fourier transform infrared photoacoustic spectroscopy. Earlier studies have shown that BGYF-positive kernels contain the bulk of the aflatoxin contaminating grain at harvest. Ten major spectral features, identified by visual inspection of the photoacoustic spectra of A. flavus mycelium grown in culture versus uninfected corn, were interpreted and assigned by theoretical comparisons of the relative chemical compositions of fungi and corn. The spectral features can be built into either empirical or knowledge-based computer models (expert systems) for automatic infrared detection and segregation of grains or kernels containing aflatoxin from the food and feed supply.