Hyperspectral Imaging Is a Cutting Edge Food Safety Technology

Consumers expect freshness, quality, consistency, and—most importantly—safety in their food products. However, delivering on this expectation has become increasingly difficult for food processors. Complicated supply chains and financial pressures to reduce processing costs can unfortunately and inadvertently lead to lapses in food processing and safety procedures, posing a major threat to consumer safety and the food industry at large.

In addition to issues of overall food quality, the food industry is faced with the liability and significant risk to consumers deriving from food pathogens, invisible foes that sicken millions of people each year. Failure to detect the presence of dangerous microorganisms, including Salmonella, Listeria, and E. coli, in foods results in severe outbreaks of foodborne illness. In fact, approximately 48 million episodes of foodborne illness occur in the U.S. every year, with 28,000 hospitalizations and 3,000 deaths. The economic burden of foodborne pathogens is thought to be as large as $36 billion every year. And, the problem seems to be getting worse, as incidents of foodborne infections continue to rise.

Recent technological advances in intelligent hyperspectral imaging (HSI) promise to disrupt the food industry’s present state of detection and response, however, giving processors a new and more effective tool in combating the pathogen breakouts that cause these illnesses.

Challenges in Maintaining Food Safety

The rising risk of infectious foodborne diseases is partly driven by the consolidation and industrialization of food production. As facilities become larger and more automated, the potential for the spread of pathogens increases.

Currently, most companies use nucleic acid-based polymerase chain reaction (PCR) techniques to detect pathogens, and this approach is widely accepted across the industry. However, this assay is costly, involves complicated sample preparation, and, most importantly, it is slow to yield results. The time from test to result ranges from 12 to 36 hours in theory, but in practice may range from three to eight days. This creates bottlenecks in the supply chain that negatively impact operating cycles and increase inventory management costs. The impact is particularly significant with perishables, which have a short shelf life. In these industries, it is commonplace for products to already be in market by the time a pathogen is identified. That means customers could already be sick by the time a problem is discovered, resulting in significant liability and brand damage to processors.

These shortfalls point to the urgent need for better solutions in pathogen detection. New advances in HSI provide such a solution: a faster system that is capable of the early detection of foodborne bacteria at the cellular level before the product is shipped to market.

How Hyperspectral Imaging Technology Works

In broad terms, HSI uses advanced hardware and software to help companies create improved quality assurance indicators. The hardware captures an image, and then the software processes it to provide actionable data by combining the power of conventional spectroscopy with digital imaging.

HSI technology utilizes superior capabilities in two areas: spectral and spatial resolution. Conversely, conventional machine vision systems lack the ability to capture and relay details and nuances to users effectively. That means HSI systems provide a level of detail that far outpaces current industry-standard systems. For example, an RGB camera can only detect three colors (red, green, and blue), while HSI can detect between 300 and 600 real colors, a significant increase of 100 to 200 times.

HSI can also read the ultraviolet or infrared spectrum, providing chemical and structural details of food composition and microorganisms that are not observable within the visible spectrum. HSI cameras do this by generating “data cubes,” which are pixels collected within an image that display subtle reflected color differences not observable by humans or conventional cameras. That information is then processed through a machine-learning algorithm to render a “classified” image, which is labeled and optimized to more efficiently process information in the future.