Glocose Calibration Log Out | Topics | Search
Moderators | Register | Edit Profile

NIR Discussion Forum » Bruce Campbell's List » I need help » Glocose Calibration « Previous Next »

Author Message
Top of pagePrevious messageNext messageBottom of page Link to this message

Avi Gabai (avig)
New member
Username: avig

Post Number: 2
Registered: 12-2007
Posted on Wednesday, November 25, 2009 - 1:56 pm:   

Thank you everybody for your answers. I am sorry that i did not respond to your messages, because i was out of office for long time.
About some points that were raised:
Reference Lab. for sugars analysis is HPLC and it is accurate.
I will look for the articles mentioned in Tony massage.

Thank you all,

Avi
Top of pagePrevious messageNext messageBottom of page Link to this message

Marta Lichtig (marta)
Junior Member
Username: marta

Post Number: 7
Registered: 8-2009
Posted on Tuesday, November 03, 2009 - 11:40 pm:   

Thanks Howard. We'll definitely investigate both options you suggest - the fitting holder and the full rotation option (first to really understand what's happening and than, if needed, recalibration).
Top of pagePrevious messageNext messageBottom of page Link to this message

Tony Davies (td)
Moderator
Username: td

Post Number: 207
Registered: 1-2001
Posted on Tuesday, November 03, 2009 - 8:27 am:   

Hello Martha,

I'm not sure if the RCA uses a circular spot or a rectangular slit. If your tablet is not very large AND it's a slit, you could be loosing some of the radiation at the edges when the tablet is at right angles to the slit and the apparent absorption will be higher. If your tablet is really small I guess you could loose radiation if it was a spot shaped illumination but not sure that that would still result in an orientation difference.

Best wishes,

Tony
PS Hello Don how are you doing?
Top of pagePrevious messageNext messageBottom of page Link to this message

Howard Mark (hlmark)
Senior Member
Username: hlmark

Post Number: 281
Registered: 9-2001
Posted on Tuesday, November 03, 2009 - 8:26 am:   

Marta - I'm surprised that you were able to rotate the sample at all. Usually, tablets are held in closely-fitting holders that are designed to match the tablet shape, and that present the sample to the instrument in a very carefully-controlled and reproducible position, just to avoid problems of the sort you're having. In fact, FOSS is a leader in providing sample holders of that sort, to customers who need them. So my first recommendation is to speak to your sales agent, to see what would be involved in getting that done.

That said, my reactions are two-fold. First, the optics of the instrument are not perfectly symmetrical, so it is likely that changing the position of the sample will indeed change it's relationship with the optics, and thereby cause changes of the type you're seeing, and indeed you say that you notice a difference in the optical readings in the two orientations. These changes may be small, but can be magnified by the calibration model.

My second reaction is that, on the other hand, calibrating using samples in different ("random")orientations should reduce the sensitivity of the model to differences in sample orienation. On the other hand, "reduce" does not necessarily mean "eliminate".

I have an idea for an approach which may help, but it will be time-consuming and is not guaranteed to work: recalibrate the instrument and when you do so, measure EACH sample in many different orientations; I would recommend taking six readings on each sample, rotating the tablet 60 degrees (of angle) each time, or even better, 8 readings of each sample with 45 degrees of rotation each time. This will give you readings at effectively all possible orientations, and hopefully will be more effective in reducing the sensitivity to orientation than you currently have. Again, though, I must repeat that it's not guaranteed.

Getting a properly-fitting sample holder, to keep the tablet oriented the same way for every measurement should remove the effect.

\o/
/_\
Top of pagePrevious messageNext messageBottom of page Link to this message

Marta Lichtig (marta)
Junior Member
Username: marta

Post Number: 6
Registered: 8-2009
Posted on Tuesday, November 03, 2009 - 4:43 am:   

First of all, thanks both Howard and Donald for the prompt replay. Indeed "vertical" and "horizontal" mean that the tablet was rotated by 90� - on the same side of the tablets, so that the engraving, score and everything is the same - except for the position.
The tablets are uncoated, and indeed, there is a difference in the log(1/R) values. The calibration was performed on random readings - we didn't know at the time that such difference exists. We also know that there is a similar difference between the two sides of the tablets, but the explanation is obvious - differences in the physical characteristics due to the build assymetry of the compression machine.
Top of pagePrevious messageNext messageBottom of page Link to this message

Donald J Dahm (djdahm)
Senior Member
Username: djdahm

Post Number: 31
Registered: 2-2007
Posted on Tuesday, November 03, 2009 - 4:28 am:   

My "common sense" does not tell me to expect identical results, but perhaps I have none. I assume, since yours does, that you did two separate calibrations.
I assume that you examined the raw spectra in each orientation and can tell us what happened to the log(1/R) values. (I suspect that the values for the absorption peaks changed in a different proportion than those at low absorption levels.)
I assume you can tell us whether or not there is a coating on the tablet that does not contain the analyte. There�s not much in the problem description for minds, great or otherwise, to work on.
Top of pagePrevious messageNext messageBottom of page Link to this message

Howard Mark (hlmark)
Senior Member
Username: hlmark

Post Number: 280
Registered: 9-2001
Posted on Tuesday, November 03, 2009 - 3:39 am:   

Marta - not sure what you mean by "horizontally" versus "vertically". Presumably both orienations are flat, with respect to the optical beam. Do you mean that one orientation is rotated 90 degrees with respect to the other? If so, then I would suspect that embossing or other variation of the surface or interior of the tablet is interacting with variations of the optical beam, especially if the beam is not completely uniform.

Howard

\o/
/_\
Top of pagePrevious messageNext messageBottom of page Link to this message

Marta Lichtig (marta)
New member
Username: marta

Post Number: 5
Registered: 8-2009
Posted on Tuesday, November 03, 2009 - 12:56 am:   

We need the input of "great minds" writing here! When testing content uniformity in oval shaped tablets (reflectance, FOSS RCA)we got consistently values which are higher when the tablet is placed horizontally versus vertically. The difference is quite significant, about 5% of the label claim. The common sense (and the FOSS support) says it shouldn't be any difference.
So please, if anyone can help us with some explanation, we'll be more than happy to receive it!
Top of pagePrevious messageNext messageBottom of page Link to this message

Z. Imbo (zimbo)
New member
Username: zimbo

Post Number: 2
Registered: 5-2008
Posted on Tuesday, October 27, 2009 - 2:43 am:   

Hi Avi,
Mostly one can find correlations of sucrose, glucose and fructose in fruits. They are often present in a certain ratio. This ratio might change over time of ripening, so be sure to take always the same stage, otherwise e.g. the skin"colour" would have a major impact for prediction. Unless you know that there are differences in these ratios, you should be able to predict all sugars with more or less the same efficiency, or maybe even more important, the sum of sugars. That should be at least 5-10% for reasonably easy predictions.
You should also try to work with ground samples, so the skin has less impact and you get a better "view" with your instrument on the inside. And try dried samples, because this reduces the watercontent to zero and you have a higher percentage of sugars in the dryweight.
Cheers.
Top of pagePrevious messageNext messageBottom of page Link to this message

venkatarman (venkynir)
Senior Member
Username: venkynir

Post Number: 88
Registered: 3-2004
Posted on Monday, October 26, 2009 - 7:18 pm:   

Dear Levin;
What about water( moisture )content ?.How do you isolate ?. What is the effect of Temperature and how do you compensate that . Is the parameter various with Temperature it is difficult to calibrate is it.
What is the lowest measure range in NIRS?
Is a Question still now ?.
Top of pagePrevious messageNext messageBottom of page Link to this message

Tony Davies (td)
Moderator
Username: td

Post Number: 206
Registered: 1-2001
Posted on Monday, October 26, 2009 - 6:17 am:   

Hello Avi,

Welcome to the group!

I moved your message to this new thread.

There is lots of published work on glucose in fresh fruit. One of the experts is Sumio Kawano. I suggest you look up his work in the main NIR Publications website (click logo to your left).

One of the problem areas can be your reference analysis. Are you confident that your glucose analysis is good?
Best wishes,

Tony
Top of pagePrevious messageNext messageBottom of page Link to this message

Gabi Levin (gabiruth)
Senior Member
Username: gabiruth

Post Number: 26
Registered: 5-2009
Posted on Monday, October 26, 2009 - 6:16 am:   

Hi guys,

I am assuming some assumptions:

1. They are trying to calibrate for sucrose, or fructose, or both and glucose.
2. The sugars are dissolved in the tissue fluids

I have long time ago done a test on mixture of three sugars - glucose, fructose and sucrose in water. The findings showed a practically zero difference between the spectrum of the dissolved sugars. Unfortunately, sugars that have nice spectral differences in the crystalline form, show no such difference in the dissolved state.
Thus, in calibrating, it is usually possible to nail one of them or two of them, but not all 3 because there is absolute interdependence in their values out of the total sugars.
If the glucose is in minority - it can predict anything - as the loadings will pick the spectral changes that result from variation of the major ones, and will never pick the minor ones correctly.

I hope this explains some of the difficulties.

Gabi Levin
Top of pagePrevious messageNext messageBottom of page Link to this message

Tony Davies (td)
Moderator
Username: td

Post Number: 205
Registered: 1-2001
Posted on Monday, October 26, 2009 - 5:23 am:   

Here is a request from a new member:

Posted by Avi Gabai (avig) on Sunday, October 25, 2009 - 9:46 am:


We are working on sugars content in fresh fruit.
We got some difficulties with glucose calibration. Have someone got
any experience in glucose analysis by NIR?
Top of pagePrevious messageNext messageBottom of page Link to this message

Avi Gabai (avig)
New member
Username: avig

Post Number: 1
Registered: 12-2007
Posted on Sunday, October 25, 2009 - 3:46 am:   

We are working on sugars content in fresh fruit.
We got some difficulties with glucose calibration. Have someone got any experience in glucose analysis by NIR?

Add Your Message Here
Posting is currently disabled in this topic. Contact your discussion moderator for more information.