This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
In this example, our API is made up of a mix of Powder A (active) and Powder B (inert) in some ratio. The Videometer assigns these powders a number to quantify the ratio later
This is a little difficult to see, but the VideometerLab has identified each ‘blob’ of powder and given it a label so when we do statistical analysis we can correlate each result to a physical ‘blob’. A detail on the left demonstrates this
Here we see that the VideometerLab has identified that Powder A (left) is more like the unknown substance than Powder B (right). We know this because the left image is more orangey, which implies a positive match
The VideometerLab can also be used to check distribution of actives on individual pills. In this image, the more orangey a pixel the less the active concentration. This distribution can be quantified easily.
These minitabs should normally have a coating on them to delay their release into the bloodstream. The tabs on the left are coated, on the right are uncoated
The VideometerLab quantifies this by replacing each pixel with a false-colour measurement of ‘degree of coating’ – if there is a good chance the tab is coated it is blue, but if there is not it will be red
Visually, there is not a lot of difference between the genuine Viagra (left) and the counterfeit (right).
Counterfeit Spectral Identification
Back to Index 8
Presented side-by-side like this one might see that the counterfeit is slightly darker, but there would be no way a human operator could make that distinction without a control sample to contrast the counterfeit to
At 700nm there is absolutely no doubt there is a difference between the left and right pill
Counterfeit Spectral Identification
Back to Index 9
If the MHRA had the spectral response of Viagra at 700nm on file, this would be enough to prove the right pill was a fake, and only takes an operator scrolling through the NIR wavelengths to see. Yet even this is impossible with traditional RGB photography
What has occurred in this image is that the VideometerLab was taught what a genuine and fake Viagra looked like, then asked to recolour the image so that each pixel which is more ‘Viagra-like’ than ‘fake-like’ is orangey\red and each pixel which is closer to counterfeit than genuine is blue\white
Statistically, this process is equivalent to a normalised canonical discriminant analysis
The transformation does not look for any difference – only those which are relevant. See how similar the transformation of these two genuine Viagra pills is
This information could be used to produce a single number which you could think of as ‘chance of being genuine’ and present to a jury. In the image below, each of the pills on the right is certain to be genuine, whereas the pill on the left is certain not to be.
If you knew –for example - that even the worst possible deviation of genuine Viagra would never fall below 0.5, you would be able to build an unassailable case that the leftmost pill was counterfeit from this single image
Though the VideometerLab is clearly successful at identifying genuine from counterfeit when the fakes are bad, in a lab setting it is more likely they will be visually similar, as with these Casadex. Even looking at spectra a human operator could make a mistake, since they are quite similar with a moderately high deviation
The VideometerLab could be used to build a case on a number of other criteria. For example, the counterfeit Viagra is ever so slightly less eccentric and bigger than the genuine. All this information can be automatically generated.
The Videometer can tell counterfeit from real even without opening the packaging
Counterfeit Packaging Identification
Back to Index 21
The next slides show a Cialis blister packet. The counterfeit packet is on the left, the genuine on the right.
Just as an aside, the Videometer usually has no trouble imaging through plastic. If – for whatever reason – you do not want to break the seal on a blister pack the Videometer can still return high-quality results through the plastic windows
The simplest way to tell the genuine from the counterfeit would be to use the Videometer to build up a model of what a real tablet looks like, then image through the blister pack
The genuine tablet is on the right, the counterfeit on the left (you can probably confirm this visually on the previous slide). The Videometer can quickly demonstrate this, even to an unskilled audience
An operator might want to confirm this by using either a statistical analysis (left image shows the imaging equivalent of a Principle Component Analysis, a Minimum Noise Frequency distribution) or by reading a spectrum (right, genuine tablet in blue)
However, if we pretend for a moment that this is impossible – maybe the counterfeit blister packet has been used completely before being confiscated – the Videometer is still a powerful tool for detecting differences
Counterfeit Packaging Identification
Back to Index 27
The next slides show a series of differences which an operator was able to pick up between the two packets. The operator was moderately skilled at using the Videometer, but completely untrained at counterfeit pharmaceutical detection – all of these differences were pointed out by the Videometer and it is certain that if it were operated by someone who knew something about counterfeiting many more differences could be discovered
In every image, the counterfeit is always presented on the left, genuine on the right
The spectral response of the ‘sun’ and the ‘swirl’ is different between packets. This is easy to see visually, but the Videometer could prove the fake falls well outside normal manufacturing error bounds
This difference is absolutely stark in profile (blue is genuine). What you are looking at is the average intensity of the pixels the small lines in the previous slide pass over. The huge drop in the middle of the counterfeit (red) profile is the line passing over the weld, which is much less neat than the genuine weld
• The Videometer is a powerful tool in pharmaceutical QC and anti-counterfeiting applications. As well as being cutting edge in its own right as an analytical instrument, one major advantage it has over the competition is the fact that the images it produces are extremely accessible, which means it can be operated and interpreted far faster and by less skilled technicians than conventional techniques
37
• The Videometer used in this presentation was absolutely ‘pure’, in the sense that it was given no prior information about what these drugs should look like, other than the training demonstrated in each short section. If an operator could take the time to train the VideometerLab on a library of drug samples, you could scale up the precision of all of these demonstrations by an order of magnitude