A new study led by the University of South Australia (UniSA) offers a promising mycotoxin detection method for the food industry that is based in artificial intelligence (AI) technology, and overcomes some of the limitations of traditional detection methods.

Mycotoxins, a class of compounds produced by fungi, are linked to outcomes like cancer, a comprised immune system, and hormone disruption. The Food and Agricultural Organization of the United Nations (FAO) estimates that about 25 percent of the world’s crops are contaminated by mycotoxin-producing fungi, making these chemicals a critical public health threat.

Unfortunately, traditional mycotoxin detection methods are time-consuming, expensive and destructive, making them unsuitable for large-scale, real-time food processing applications, according to UniSA Ph.D. candidate Ahasan Kabir.

To address the need for an effective and usable mycotoxin detection method, a new study, published in Toxins, demonstrates that hyperspectral imaging (HSI)—a technique that captures images with detailed spectral information—when paired with machine learning (ML), can rapidly and non-invasively detect mycotoxins in food products before they reach consumers. This AI-driven approach identifies subtle spectral variations in contaminated samples, allowing for accurate classification without damaging the product.

When the research team evaluated the effectiveness of HSI in detecting toxic compounds in cereal grains and nuts, they found that it is able to capture an optical footprint of mycotoxins, and when paired with ML algorithms, HSI can rapidly classify contaminated grains and nuts.

The researchers also reviewed more than 80 studies on mycotoxin detection methods across a variety of foods that are highly susceptible to fungal contamination, including wheat, corn, barley, oats, almonds, peanuts, and pistachios. They concluded that ML-integrated HSI consistently outperformed conventional mycotoxin detection methods, especially in identifying aflatoxin B1, a significant carcinogen.

According to the researchers, the technology shows particular promise due to its scalability. From conveyor belt inspections to handheld devices, HSI systems could be deployed across the supply chain to ensure contaminated products are flagged and removed in real time.

Using deep learning and AI, the researchers are now refining the technique to improve its accuracy and reliability. They believe that the technology, once optimized, could become a commonplace food safety intervention.

The study was supported by the Australian Federal Government’s Research Training Program, with top-up funding from SureNut Australia. Led by the University of South Australia, the global research team also included collaborators from Chittagong University of Engineering and Technology, Bangladesh; Lethbridge College, Canada; and the Indian Institute of Technology Kharagpur, India.