Corn Spills the Beans

“Interrogate” corn kernels under strobe lights and they may admit aloud that they’re harboring a toxin-producing fungus. Nowadays, Agricultural Research Service scientists with specially programmed computers find such confessions ring true with 96 percent accuracy.

At grain elevators, inspectors routinely check corn for the fungus Aspergillusflavus. It produces aflatoxin, a hazardous substance that poses health risks if it gets into food or livestock feed.

To check for the fungus, inspectors use a bright greenish yellow fluorescence (BGYF) test. Samples that glow under ultraviolet light are suspect and must undergo lab analysis. As another check, the latest cross-examination idea may fit into a system that would monitor corn on a conveyor belt and divert infected grain.

At the National Center for Agricultural Utilization Research, Peoria, Ill., the ARS scientists “interrogated” corn by using Fourier transform infrared photoacoustic spectroscopy (FTIR-PAS). In this process, pulses of infrared light bombard kernels inside a chamber. The resulting heat waves radiate from the corn, sending sound waves to a microphone. Sound, representing different infrared wavelengths, is recorded in a computer database. Infected corn sends out different levels of sound than non-infected corn.

To enable computers to recognize differences in infrared patterns, the researchers chose software written by University of Illinois computer scientists. Called an artificial neural network, the software distinguishes infected from uninfected corn, using conditioned reflexes--somewhat like those in a nervous system.

To apply the same principles to moving corn, the ARS scientists are collaborating with colleagues at Iowa State University, Ames, who research a related technology, Transient Infrared Emission Spectroscopy.

ARS, the chief research agency of the U.S. Department of Agriculture, is currently seeking an industrial partner to help develop portable infrared sensors. The sensors, along with a knowledge-based computer program or expert system, would enhance reliability of neural networks at elevators and corn processing plants.

An article about the research appears in the April issue of Agricultural Research magazine and online at:

http://www.ars.usda.gov/is/AR/archive/apr99/sound0499.htm

Scientific contact: Richard V. Greene and Sherald H. Gordon, ARS, National Center for Agricultural Utilization Research, Peoria, Ill., 61604; phone (309) 681-6591, fax (309) 681-6689, [email protected](Greene) and [email protected] (Gordon).