The Future of Food Safety and Quality Control Is AI-Assisted Hyperspectral Imaging
Updated: Apr 20
Standing as a shining star among incoming technologies being used for food safety and quality issues, AI-Assisted Hyperspectral Imaging provides key benefits over chemical-based methods for the development and execution of food processor’s HACCP Plan that ensure it’s a trend that’s here to stay.
What is AI-Assisted Hyperspectral Imaging?
AI-Hyperspectral Imaging (AI-HSI) is an imaging-based technology that gives information about the chemical composition and 2-D spatial distribution across samples. In other words, it tells you
What's is in a food sample (e.g. moisture)
How much of it is in the food sample (e.g. 40%)
And where it is located (eg. highest in the middle & lowest on the edges )
The process can be scaled up to warehouse information from previous analyses and include multiple parameters that can be used by machine learning systems (the AI in AI-HSI) to provide unprecedented amounts of actionable information about the quality of food products. This allows for a truly robust and reliable hazard analysis that keeps with the spirit of the HACCP Manual and the need for Principles 4&7 of monitoring critical control points and recordkeeping in an effective HACCP plan.
Where did it come from?
Like many promising emerging technologies, AI-HSI was originally developed for various purposes within governmental agencies. This included use by NASA on imaging satellites because it could differentiate not just between water and land, but between dust and sand, oak tree from maple tree forests, and differentiating between different types of crops, all by chemical composition. It has been used in a variety of applications by the United States military as well, where it can be used for hazard analysis of chemical and explosive threats, terrain, as well as facial recognition.
How does it work?
The process works by having a hyperspectral imaging camera (1) emitting broad-spectrum light toward a food sample(2) that is to be analyzed. That sample, which is composed of chemical compounds, interacts with this incoming light, which causes its atoms and bonds to vibrate and a sensor determines the frequencies and amount of light those chemicals absorb/transmit/reflect and scatter as a result of this light/matter interaction in the form of absorbance spectra (3). This data is then used in the AI-analysis(4) where it determines the concentration of that compound in the sample. This process is repeated for each and every pixel of a sample image, and the results are mapped to reveal chemical distribution (5) across the sample.
What does it measure?
The short answer is, it depends. That’s because AI-HSI as a technique can be used with a wide range of spectroscopic sensors. Depending on the chemical composition, one sensor or another may be required, but as a general statement, it can be said that AI-HSI can analyze anything with an appreciable absorbance in the ultraviolet, visible, or infrared range. Because nearly all organic compounds meet this category, it’s appropriate to say that a vast majority of primary (protein, fat, carbohydrates, and nucleic acids) and secondary metabolites (e.g. caffeine, nicotine, capsaicin, curcumin, etc.) can be qualitatively and quantitatively assessed using AI-HSI. Below is a table listing the major parameters that can be assessed in a variety of commodities using an AI-HSI sensor in the 900-1700 nm range.
What are the benefits of AI-HSI?
As an imaging technology AI-HSI has distinct advantages with respect to higher resolution, more detailed spectral data acquisition, and a tremendous amount of firepower within the software’s AI package making associations between datasets.
As a method for the quality control of food AI-HSI has a number of advantages over other spectroscopic techniques such as NIR and chemical-based and ELISA tests:
Direct Cost Savings
(No reagents or supplies and no need for highly-skilled technicians)
(<30sec/sample offline with continuous inline monitoring options available)
Multiple Parameters Simultaneously
(Range of sensors cover all your QC needs with one method)
Reduction in Production Line Hold-Time > 90%
(Faster testing leads to more production)
(Ready-to-use, no learning curve, contactless, hygienic, non-destructive, no harsh chemicals)