Researchers have designed an artificial intelligence (AI) program capable of detecting contaminated foods in factories and farms, stopping them from reaching customers and possibly saving more than four million lives every year.
The study, led by the University of South Australia, used advanced hyperspectral imaging (HSI) and machine learning (ML) to target the presence of mycotoxins in food. Mycotoxins are dangerous compounds created by fungi that can contaminate food in the growing, harvesting or storage phase.
The important research was funded by the Federal Government’s Research Training Program.
Food for thought
According to the World Health Organisation, mycotoxins are naturally produced by certain types of moulds, growing on food such as cereals, nuts, fruits, and coffee beans. This typically occurs in warm and humid conditions.
Mycotoxins can cause a range of significant health issues, from acute poisoning to immune deficiency and cancer. Foodborne contamination, including from mycotoxins, is responsible for 600 million illnesses and 4.2 million deaths every year.
Research from the UN-based Food and Agricultural Organisation reveals that around 25 per cent of the world’s crops are contaminated by mycotoxin-producing fungi. This highlights a need for greater identification and prevention to address the ongoing threat of disease around the world.
Lead author of the UniSA study, PhD candidate Ahasan Kabir, reveals that traditional methods of mycotoxin detection are expensive and destructive, meaning they aren’t suitable for large-scale real-time food processing.
“In contrast, hyperspectral imaging – a technique that captures images with detailed spectral information – allows us to quickly detect and quantify contamination across entire food samples without destroying them,” Ahasan says.
Toxin testing
The study, published in the journal Toxins, was a collaboration between researchers in Australia, Canada and India. The team tested the effectiveness of HSI in detecting the presence of toxic compounds in cereal grains and nuts, the world’s most produced food. Both are susceptible to mycotoxin infection at any stage from growth to storage.
“HSI captures an optical footprint of mycotoxins, and when paired with machine learning algorithms, it rapidly classifies contaminated grains and nuts based on subtle spectral variations,” Ahasan says.
The team also reviewed over 80 studies into wheat, corn, barley, oats, almonds, peanuts and pistachios. They found that HSI systems routinely detected key mycotoxins at a better rate than conventional technology.
According to project lead and UniSA Professor Sang-Heon Lee, the technology is most effective when identifying aflatoxin B1, one of the most carcinogenic substances found in food.
“It offers a scalable, non-invasive solution for industrial food safety, from sorting almonds to inspecting wheat and maize shipments,” says Professor Lee.
Unlike traditional methods of detecting mycotoxins, the HSI and ML program can work in real time, meaning it could be deployed on handheld devices or in larger-scale processing lines, to reduce the risk of infected foods making it onto supermarket shelves.
The team is currently working on improving the accuracy and reliability of the software through deep learning and AI.
To learn about the AI model helping doctors to diagnose skin cancer, click here.