Where will AI and Hyperspectral Imaging take food safety and quality?

Jordan Gunn

Many industries of today require rigorous quality control to ensure product specifications and avoid defects, contamination, adulteration, omissions, and mislabelling.  These include:

  • Food and Feed
  • Pharmaceutical
  • Dietary Supplements
  • Oil & Gas
  • Mineral Mining
  • Paper and Pulp
  • Recycling

The quality control and safety budgets in these industries varies by company and is a function of their obligation to regulators, dedication to customer satisfaction, and their overall profit structure.  Generally, the quality budgets constitute up to 2% of revenue in the production of food, with a food safety control market value of $2 billion, while in the pharmaceutical industry quality costs can constitute over 20% of production costs.  The FDA and the USDA, which collectively regulate food and feed, pharmaceuticals, and dietary supplements, have an annual budget of $5.9 billion and $12 billion (for food safety and regulatory affairs), respectively.

It might be surprising to some that although their products vary considerably, these industries rely on many of the same methods for quality control.  Typical methods include:

  • Documentation (SOP, HACCP, SDS, etc.)
  • Sensory Tests (Look, smell, taste, feel)
  • Chemical Tests (ELISA, Spot Tests, TLC)
  • Spectrometry (HPLC, GC, MS, IR, NMR)

All of these tests suffer from AT LEAST one of the following issues:

  • Too simplistic
  • Too time-consuming
  • Too costly
  • Too tedious
  • Not objective
  • Not quantitative

There are indications in the market that tides are shifting to more optical-based methods of quality control.  These shifts are most starkly embodied in their compound annual growth rates.

It makes sense that technologies like hyperspectral imaging would be outpacing other methods since optical-based methods hold a few crucial advantages over documentation alone, sensory, chemical, and classic spectrometry methods

  • They can be made comprehensive
  • They can be fast, in some cases extremely fast
  • They are cheap since they don’t require supplies and consumables
  • They can be made simple to use
  • They are objective
  • They can be made quantitative

Notice in the above list there are two kinds of advantages.  Advantages that ARE inherent to the technology, and advantages that CAN BE incorporated into the technology.  And the way that they can be incorporated?  Through software.

  • Software is how you make optical-based methods comprehensive.
  • Software is how you adjust the speed of analysis.
  • Software is how you make these methods simple to use.
  • Software is how you make these methods quantitative.

ImagoAI is a leader in spectral data analysis.  They have worked on a variety of problem statements in all of the previously mentioned industries, helping them to adopt spectral imaging faster, and bringing value to their operations in the form of:-

  1. Decreasing latency/increasing speed.
  2. Creating comprehensive analyses that combine multiple parameters.
  3. Increasing the accuracy of analyses.
  4. Decreasing data points needed to make the same statistical inferences.
  5. Making turn-key, easy to use interfaces.
  6. Producing semi-quantitative and quantitative algorithms.
  7. Creating self-learning algorithms that continually update to produce better results.

In these ways, ImagoAI’s work on the software end is helping to deliver the future of quality control, which is optical-based, to the food and feed, pharmaceutical, dietary supplements, oil & gas, mineral mining, paper and pulp, and recycling industries.