| Author | Message | ||
     LIU Yang (Liuy) |
Dear all, I am planning to do some capsules analysis by NIR. The sample will be a series medicine capsules. And I want to use some pattern recongnization methods to discriminate them. The problem is about experimental technique. I can access a FTNIR instrument, but the only available sampling accessaries are big sample cups and some tablet holders for diffuse reflectance spectrum. Can I use these accessaries to do the capsules analysis? Are there some details I should pay attention to when I do this work? Another question, will the color and material of capsule shell influence the spetra much? Since some of the shells of samples may be different. Appreciate any suggestions and information! Thanks! | ||
| Gary Ritchie |
I don't know how much time you will have to do this analysis, but I will have an article published in the October issue of the Journal of Near Infrared Spectroscopy entitled:�A general test method for the development, validation and routine use of disposable near infrared spectroscopic libraries.� In answer to your immedisate concerns however, I recommend the following: 1. You asked: the only available sampling accessaries are big sample cups and some tablet holders for diffuse reflectance spectrum. Can I use these accessaries to do the capsules analysis? Are there some details I should pay attention to when I do this work? You could try to scan the capsules by filling a cup with multiple capsules (all of the same composition of course), but I have not tried this. Generally, I perform discrimination analysis on single unit dosages by reflectance using either a fiber optic probe or a Rapid Content Analyzer (FOSS NIRSystems) or a multitab analyzer (or Intact analyzer) for Transmmitance work. The capsule or tablet holders have to be machined from a blank holder supplied by FOSS to fit the specific shape of the dosage unit. Generally 25-50 units are all that is needed to build a library, but the more lots of the same form you use, mfg'd over a period of time gives you more robustness. 2. You asked: Another question, will the color and material of capsule shell influence the spetra much? Since some of the shells of samples may be different. Genrally not, has been my experience since these effects usuually appear in the 600 - 1100 range. And anyhow, you will find that by carefully selecting wavelengths or a range of wavelengths that correlate highest to the major differences, which you can usually visualize by eye by looking for large gaps between the active and placebo dosage forms in the banding pattern of the spectra, (I assume you want to discriminate between an active unit vs. a placebo unit) on tranformed spectra (2nd derivatives usually), you can create models that can easily discriminate between dosage types in either wavelength space, PC space, or even better, residual analysis. The selectitiviy power of the model typically increases in accuracy from wavelength to PC to residual. Good luck and keep us posted! | ||
| LIU Yang (Liuy) |
Gary,thank you very much for your help! Your suggestion really give me clues. I have experience of tobacco analysis and quantitative calibrations. I'm thinking of extending the field of my analysis, so I choose Pharmaceutical analysis to start.I will spend some times to collect samples and start the work. Looking forward to your article and hope I can discuss with you later. | ||
| Gary Ritchie |
Glad to be of some help. Actually upon reflection, I remember a collegue telling me that an identification using NIR on bulk product (I think they were tablets) used a cup filled with a number of tablets, and this worked pretty well. There is no reason not to do it this way, except that most protocols for identification require unit dose samples to be identified. | ||
| Michel Coene (Michel) |
If you have good reflection from your tablets, you could measure them in diffuse reflectance on e.g. a black surface. This allows you to scan a single tablet. It could however be that a diffuse reflectance probe simply does not return enough light. In this case you need a Free-space diffuse reflectance setup (measurement without fiber-optics). I agree with Gary that (in reflectance), you have to search for wavelengths with deep penetration, in order to pass the (gelatine) capsule. | ||
| sanette |
I am currently validating a gelatine capsule (soft gel) library. The data was collected using a course cell on a FOSS Multi-Mode Analyzer in reflectance mode from 400nm to 2500nm. My library contains 6 different capsules and I was able to discriminate between all of them by using 400-2500nm, Maximum distance in Wavelength space, 2nd derivative pre-treatment. I am however concerned about what properties I am measuring. The library was challenged with other capsule (about 94 different capsules) not contained in the library. Of these capsules, one was found to be ambiguous with one of the library products. These two capsules have the same colour geletine shell, but the contents are different. When I changed the region of the ID method to 700-2500nm to exclude the colour differences of the capsules, I could still discriminate between the products in the library, but more overlapping products were found when challenged. And it was even worse using 1100-2500nm, because visually the 2nd derivative spectra were very similar. Would I be correct to assume that I can�t identify the capsules purely on the contents, but need the colour info from 400-700nm regions to discriminate between them? I don�t want to identify capsules on the physical appearance only, but want to be sure that the contents are contributing to the spectra. Empty and filled gelatine capsules were scanned and clear differences could be seen in the spectra. From that I assume the NIR light is actually penetrating the shell. I read on the previous postings (Capsules: 28 Feb 2002) that one should use wavelengths with deep penetration to pass the gelatine shell. How could I determine what wavelengths to use when comparing the empty and filled capsule spectra? I would appreciate some advice. Thanks Sanette | ||
| Tony Davies (Td) |
Dear Sanette, 1) You really ought to be using transmission then you know that (most of) the energy collected has passed through the capsule. 2) I agree that you should NOT rely on colour information. Discrimination should be on the basis of the active ingredients. 3) The wavelengths you need are in the 850 to 1000nm range. 4) The topic of ID by NIR is one of the subjects for discussion at next month's IDRC (Chambersburg). I am hoping that people will be willing to recognise that the current practise is inherently unsafe and that we ought to do something to improve the situation which your experience has highlighted. Best wishes, Tony | ||
| Gabi Levin |
Hi Yang, Please allow me to add the following: Capsules present a very special problem for reflectance measurements, because when placed horizontally, the powder settles to the bottom, thus leaving an air space between itself and the wall of the capsule. Thus, to do reflectance you will be better of by having them stand up, and illuminate and collect from the side of the capsule. This will assure that you see a fixed amount of powder per a fixed amount of wall. If there is a small probability that the active is not uniformly distributed within the powder, then rotating the capsule while collecting spectra may be essential in order to represent more of the powder. Still, the best way is to do transmittance, and depending on the size of the illuminating beam, and the size of the capsule, you may wish to "scan" the full volume of the capsule to obtain good averaging of the entire powder. Thanks, Gabi Levin Brimrose Copr. | ||
| Gabi Levin |
To Sanette, Since you refer to capsules, if you tested them horizonatlly, you may see significant variations in the "signal" due to variations of "amount" of powder product vs. "amount" of shell. This can cause many errors. To do the capsules in reflectance it may be critical to have them upright, assuring better consistency of the presentation of powdered product. Would like to hear what do you think about it. Thnaks, Gabi Levin |