U.S – Researchers from the University of California Davis have conducted a study that suggests artificial intelligence (AI) combined with optical vision may be a potential method for identifying infections in food.

The “You Only Look Once version 4″ (YOLOv4)” algorithm was evaluated in the study to see how well it could identify and categorize Escherichia coli (E.coli).

The software-enabled technology offers an advantage over conventional E. coli detection techniques, which call for the timely and time-consuming isolation of bacterial macrocolonies for biochemical or genetic analysis.

The unique bacterial identification approach entails two steps: real-time detection using the object detection and classification algorithm YOLOv4, and microcolony incubation and white light imaging. It is based on an investigation of morphological distinctions among bacterial microcolonies.

The researchers first examined the method’s viability for pathogen enumeration and determined that a cultivation duration of 3 hours was adequate for E. coli.

It was demonstrated that the method made it possible to quickly quantify E. coli concentrations.

The study also investigated whether YOLOv4 can distinguish between E. coli microcolony traits and those of other food spoilage and pathogenic bacteria.

The scientists classified E. coli and seven other non-E. coli organisms that were isolated from foods, habitats, animals, and people using YOLOv4.

Listeria innocua, Salmonella enterica, Listeria monocytogenes, Pseudomonas fluorescens, Bacillus coagulans, and Bacillus subtilis were the other seven non-E. coli species.

When combined with phase-contrast microscopic imaging, YOLOv4 had an average precision of 94 percent in differentiating E. coli from the other seven prevalent foodborne bacterial species.

Additionally, researchers tested the software against E. coli in a romaine lettuce solution to see how well the YOLOv4 detector performed in distinguishing E. coli in fresh food.

With a false-negative rate of less than 10% after 3 hours of incubation, YOLOv4, trained with several bacterial species, identified E. coli. A typical plating test was used to validate the results.

The current technique might be useful to the food industry as an easy, affordable, and rapid way to find pathogen contamination in food products.

In addition to being fast, the method doesn’t call for either expensive equipment or highly skilled labor.

Besides achieving high classification accuracy for bacteria, the study’s AI-assisted detection method can detect bacteria automatically.

The presence of Escherichia coli is a crucial sign of fecal contamination when evaluating the microbiological safety of food.

For the protection of consumers and the control of outbreaks, early detection of microbial contamination in food products is essential.

Worldwide, it is estimated that eating contaminated food causes 550 million illnesses (nearly 1 in 10 persons) and 230,000 fatalities, says the study.

For all the latest food safety news from Africa and the World, subscribe to our NEWSLETTER, follow us on Twitter and LinkedIn, like us on Facebook and subscribe to our YouTube channel.