An international team of researchers demonstrated how AI (AI) detects contaminated food in fields and factories before reaching consumers, potentially saving 4 million deaths per year.
A new paper published in the Journal of Toxin, led by the University of South Australia, describes how advanced hyperspectral imaging (HSI) integrated with machine learning (ML) can identify mycotoxins – dangerous compounds produced by fungi that can contaminate food during growth, harvesting and storage.
Mycotoxins cause a variety of serious health problems, including cancer, immune compromise, and hormone-related disorders. According to the World Health Organization, food-borne contamination, including mycotoxins, results in 600 million illnesses and 4.2 million deaths each year.
The UN-based Food and Agriculture Organization estimates that around 25% of the world's crops are contaminated by mycotoxin-producing bacteria, highlighting economic and health orders to address this threat.
Lead author and UNISA PhD candidate Ahasan Kabir state that traditional mycotoxin detection methods are time-consuming, expensive, disruptive and unsuitable for large-scale real-time food processing.
“In contrast, hyperspectral imaging – a technique that uses detailed spectral information to capture images allows rapid detection and quantification of contamination across food samples without destroying them,” says Kabir.
Kabir and his co-authors from Australia, Canada and India have evaluated the effectiveness of HSI in detecting toxic compounds of grain and nuts, the world's most produced foods, and economic backbones of many countries.
Both are extremely sensitive to fungal and mycotoxin contamination in warm, humid environments, from cultivation to storage.
“HSI captures the optical footprint of mycotoxins and when combined with machine learning algorithms, it quickly classifies contaminated grains and nuts based on subtle spectral variations,” says Kabir.
Researchers reviewed over 80 recent studies via wheat, corn, barley, oats, almonds, peanuts and pistachios. The findings showed that ML-integrated HSI systems consistently outperform conventional techniques in the detection of major mycotoxins.
“The technology is particularly effective in identifying aflatoxin B1, one of the most carcinogens found in foods, according to the project by Professor UNISA Sang-Heon Lee.
“From sorting almonds to inspecting wheat and corn shipments, we provide a scalable, non-invasive solution for industrial food safety,” says Professor Lee.
One of the main advantages of this approach is the ability to work in real time. Researchers say further developments can reduce health risks and trade losses by deploying HSI and ML on processing lines or handheld devices, ensuring only safe and uncontaminated produce reaches consumers.
The team is currently working on improving methods to improve their accuracy and reliability using deep learning and AI.
The study was supported by a federal government research training programme and raised funding from Australian Slenauts.
“Machine Learning Integrated Hyperspectral Imaging: Detection of Grain and Nut Mycotoxins Using Reviews: Featured in Toxins. doi: 10.3390/Toxin 17050219.
This paper was written by Ahasan Kabir, Professor Ivan Lee and Professor San Heon Lee (University of South Australia). Professor Chandra Singh (University of Lethbridge, Canada); Professor Gayatri Mishra and Braj Kumar Panda (Indian Institute of Technology Haragpur).
