Grainswest - Spring 2026

Spring 2026 Grains West 8 THE FARMGATE BY TREVOR BACQUE • PHOTO COURTESY OF SHEILA ANDRADE A NEWER, FASTER AND MORE accurate approach may detect morpho- logical changes associated with DON contamination in Fusarium-damaged wheat kernels. New research from the Crop Development Centre at the Univer- sity of Saskatchewan indicates existing test methods can be improved. In the current testing process, the spikes are threshed to release kernels. A visual assessment then involves a manual count of kernels damaged by Fusarium. Kernels are ground up to extract myco- toxins, a costly process that also involves destructive testing. To develop this stronger, more objec- tive test, PhD candidate Sheila Andrade used the Canadian Light Source synchro- tron, an ultra-bright light that allows scientists to visualize the internal struc- ture of materials with high resolution and precision. She also applied machine learning to use image-derived morpho- logical parameters from Fusarium-in- fected kernels to predict the presence of the common mycotoxin DON, which primarily affects wheat. Trialled in both bread and durum wheat, the test produces a 3D X-ray image of the kernel. This revealed the structure of individual kernels, including area, shape, volume, density and other structural features. The benefits of this technology over current methods are threefold: greater accuracy, non-destructive analysis and the potential to speed up the process. “Breeders usually work with a lot of lines, and they need to select those resistant to Fusarium head blight,” said Andrade. Because current screening methods for Fusarium damage and DON are conduct- ed by humans, they are time consuming and come with the potential for error. Another drawback of visual assess- ments is kernels can be infected without showing symptoms such as shrivelling and a chalky-white appearance. “It’s a tricky task that requires significant expertise and practice to recognize the symptoms and separate the infected kernels,” said Andrade. This new approach allowed the appli- cation of predictive models for mycotoxin contamination. Andrade’s models deliver about 80 per cent accuracy on both durum and bread wheat. “We felt it was fairly high, although there is still room for improvement,” said Randy Kutcher of the technology. Kutcher is Andrade’s PhD supervisor and believes such a test may one day be used at country elevators. “Benchtop units that have similar technology as the synchrotron would be much more practical, so a breeder or a pathologist could have that in their lab and run through hundreds and hundreds of samples a day,” said Kutcher. Research officer and Andrade’s project co-advisor, Lipu Wang has collaborated with Scott Noble’s team from the College of Engi- neering to develop a benchtop machine with an RGB camera that can process about 200 kernels in 90 seconds. The next goal would be to create a similar system using X-ray technology or to integrate this capability within the existing machine. Andrade plans to create a machine learning model for barley testing, as well. Reference images, mycotoxin content and physical parameters for the cereal have been documented. The research group has also compiled images for triticale and oats, which could be used to develop additional models. For barley, said Kutcher, such a fast, accurate test would potentially allow a maltster to run a test prior to purchasing malting barley. Naturally, Andrade’s research has drawn considerable interest from indus- try, which may lead to commercialization or licensing of the technology. “Our first goal was to confirm whether this tech- nology could work. If we can improve the model and then get a portable X-ray machine, that would be a major step forward,” she said. Andrade’s research is sponsored by SaskWheat, Western Grains Research Foundation and the National Science and Engineering Research Council of Canada. Non-destructive contaminationtest New techmakes cost-efficient DON prediction Researcher Sheila Andrade used the Canadian Light Source synchrotron to develop a new, more accurate means to detect Fusarium damage and DON.