Limit of detection for X-ray machine
Hi,
X-ray technology is adressing various detection of contaminants however a large number of manufacturers can't advertize the limits of the technology without destroying their brands.
I'm looking out to find examples of applications where the X-Ray technology was not able to respond to the customer requirement in a way that the expect level of alarm rate is still too high to demonstrate that the safety requirements are in a safe path..
Do you have some examples?
Well.. I've never actually used one in a food plant. We bought one at my last factory though they never put it in line while I was there (it was sitting under plastic) but we were told that it not detect anything that floated in water. Apperently the density of an object has some direct relevance on what it will detect. So metal shavings may not be picked up or wood.
I was waiting to see if someone would pick up your topic that has more intimate knowledge of metal detectors but that's all I have on the subject.
Hi, here’s an approximation as to what X-ray inspection systems will detect. It is not 100% accurate just a guide. There may be some situations where better detection is achievable and some where it is not.
1. measure the product height (mm)
2. determine the product specific gravity S.G. e.g. cheese ~ 1.0
3. multiply 1. x 2. e.g. 100mm high cheese block x 1.0 = 100; call this absorption
when a contaminant is in an homogeneous product e.g. block of cheese the absorption needs to increase by approximately 5% before it is detected and for a particulate product e.g. bag of frozen french fries the absorption needs to increase by approximately 10% before detection. These are extreme examples, most food products will lie between 5% and 10%.
Take a stainless steel contaminant or test sphere, S.G. ~ 8. For the above 5% change to occur the thickness or diameter (if a typical test sphere was being used) would need to be > 0.6mm. For a particulate product this would need to be > 1.2mm.
The maths homogeneous: 100 + (mm x 8) = 105, mm = 5/8 = 0.63
The maths particulate: 100 + (mm x 8) = 110, mm = 10/8 = 1.25
It’s a bit more complicate than this as when a contaminant or test sample is in the product it displaces some of the product contents but as an approximation it works.
Take glass S.G. ~ 2.5. For the above 5% change to occur the thickness or diameter (if a typical test sphere was being used) would need to be > 2.0mm. For a particulate product this would need to be > 4.0.
You can carry out the same maths on lower density contaminants, thicker and more dense products and determine a rough approximation as to what will or won’t be detected. By using the above gives also an indication of the level of false rejects. If the change in absorption falls below the 5% or 10% or somewhere between the chances are there will be significant false rejects.
The best approach is to have the product checked out by potential X-ray suppliers and a report produced as to the expected levels of detection for given material types. You can however see the relationship between product density, height and contaminant density and height (diameter).
The above does not work too well for very low density contaminants and/or very high/dense products, but seems to offer a reasonable prediction for test samples (spheres) or contaminants in consumer packs.
Hope it helps.
Hi Adelpy,
X-ray detectors will have issues if the mass density of the food and the contaminant is similar. In case some (good) components of the food have similar mass densities to the expected contaminant (plastic, wood, glass...), they would register as contaminants. I would expect that x-ray will generally have trouble detecting wood in a cheese-cake that contains walnuts, for example.
In case you were asking about limitations specifically in accuracy, I don't know much, apart from the fact that a very homogenous substances will probably allow for the highest accuracy one can get out of x-rays (assuming the difference in density of the contaminant and the food is very high).
Hope this helps you.
Cheers,
-Alex
Can see this is an old topic :)
But from the last post I am assuming that if a food had a nut topping, an X ray would struggle picking up wood?
Wow zombie thread!
But I think the previous answers probably are far more expert than I am! However, x-rays are interesting because all the previous posts are right, a density difference would be needed to "see" the contaminant.
BUT there are advancements being made. So the nuts on the top of your product and the piece of wood will look different. So using either vision or x-ray technology (whichever shows it up better) it should be possible to apply some machine learning to "teach" the system what it's seeing. At a cost of course.
But rather then detecting wood in your product, is it possible to detect the wood before it gets there? If the wood may come in with the nuts, using a visual sorter on those might be better and easier than trying to find it in your product? Again it's all money but it depends on risk and how many eyes you might have instead?
These guys might be worth getting in touch with for visual sorting technology. Sorting machines and post-harvest solutions for food - TOMRA