| Author | Message | ||
     S�verine Gabioud Rebeaud (gab) Junior Member Username: gab Post Number: 6 Registered: 4-2010 |
Dear Paolo, I effectively compared both instruments based on PLS regressions and data sets were not always > 100. I will follow your advice and test MLR. Thanks a lot for your help! | ||
| Paolo Berzaghi (pberzaghi) New member Username: pberzaghi Post Number: 3 Registered: 11-2008 |
S�verine, the detector is important, but you also must look at the other components of your system, like the optical design, light source and reference source, so at the end you have to evaluate the entire system all at the same time. I'm surprise you have similar performance in interanctance and reflectance. With silicon detectors in reflectance I had fair results only for moisture. How many samples do you have in the calibration data set? You must be carefull drawing any conclusion with small calibration data sets (<100 samples), PLS seem to fit almost anything in small calibration sets particularly if the variability of samples is limited. In that case you may consider using multiple linear regression instead of PLS to get more realistic performance. Paolo | ||
| S�verine Gabioud Rebeaud (gab) New member Username: gab Post Number: 5 Registered: 4-2010 |
Both instruments have a silicon detector measuring between 400 and 1100 nm. Paolo: do you think that the type of detector is more important than the type of system regarding stability, reproducibility and transferability of calibration? S�verine | ||
| Paolo Berzaghi (pberzaghi) New member Username: pberzaghi Post Number: 2 Registered: 11-2008 |
Gab, what you describe is an interactance probe, which is in fact a form of transmission as Gabi said. Usually for interanctance you have to measure light in the 400-1100nm range which means you can use low cost silicon detectors. In reflectance, if you have similar and good performance you are probably using the upper range (>1100 nm) which requires more expensive InGaAs detectors. Then if you talk about stability, reproducibility and transferability of calibration a lot depends of the manufactors of those sensors, but in my experience transferability is easier with silicon detectors. Paolo | ||
| Ralf Marbach (ralf) Member Username: ralf Post Number: 14 Registered: 9-2007 |
Hi Severine, There are some applications, and your's may be one of them, where NIR probes can be designed to be insensitive to changes in sample scatter coefficient, at least to some approximation. See: G. Kumar and J.M. Schmitt, Appl. Opt. 36, 2286 (1997). So, if you were bothered by multiplicative errors, i.e., pathlength changes, then this could be one criterium to evaluate probe geometries. Ralf | ||
| S�verine Gabioud Rebeaud (gab) New member Username: gab Post Number: 4 Registered: 4-2010 |
Dear all, Thanks for your responses and your precisions on the different modes of measurements. One of the NIR instrument I'm testing is a system working in a reflectance mode as the light fibers and the receiving fiber are in the same contact sonde. The configuration of the other instrument I'm testing is different: the fruit is placed on a "holder" and illuminated by lamps. The detector is inside and light has to go through the fruit to be detected. However it's not a vertical system as the lamps are below the fruit at about 45�. Is it transmittance? Beside this, I tested these 2 instruments with the same batches of fruits and obtained similar results. As now I have to choose only one instrument, I was wondering if one of these systems could be recommended to measure fruits according to criteria such stability, reproducibility, transfer of calibration, ... Best regards, S�verine | ||
| Gabi Levin (gabiruth) Senior Member Username: gabiruth Post Number: 42 Registered: 5-2009 |
Howard, thanks for the comment on the use of a reflector in cases where the liquid is not clear, and the products you mention are actually where I used it such as in whey protein, juices containing pulp, etc. However, it is not always possible to determine how much of the returned light is coming from diffuse reflectance by the solids and how much is due to reflectance from the reflector. Since this is a combined case, I chose not to include it so as not to complicate things. The third method mentioned by Hopkins is definitely relevant, but in essence it is forward diffuse transmission, and the fact that the light is not collected at straight line from the entry point is useful in reducing the effective "path length" and enabling instruments with lower light intensity to be useful. One of the most critical problems in performing diffuse reflectance in fruits is the limit on depth of penetration, and the fact that in many cases the composition of the fruit mass is not uniform from outside inwardly. In many cases there is a gradient from outside to the inside, and the gradient is not constant around the fruit, leading to the need to average several readings to arrive at a final value. Thanks to all, Gabi Levin | ||
| David W. Hopkins (dhopkins) Senior Member Username: dhopkins Post Number: 165 Registered: 10-2002 |
Hi S�verine, I would like to mention that there is a third sample measurement geometry that has been very successful with fruit: interactance measurements. This involves aiming the measurement beam on a spot on the surface, and collecting the light that has penetrated into the sample and diffusely scattered to be picked up and measured by a probe a certain distance from the point of entry. By choosing an appropriate separation between the irradiation and measurement spots, the depth of the penetration is controlled, depending on the optics of the fruit skin and flesh, so that good determinations of the quality of the fruit are made. A number of commercial hand-held or bench-top instruments are available, so I hope you are not planning to reinvent an instrument. Best regards, Dave | ||
| Howard Mark (hlmark) Senior Member Username: hlmark Post Number: 355 Registered: 9-2001 |
Gabi - your description of transflectance is very good, however you erred in limiting its application to clear liquids. As I recall, one of the earliest applications of that measurement method included milk and dairy products, also emulsions such as mayonnaise and other scattering samples. Arguably, whether you're measuring "transflection" or "diffuse reflection" in such circumstances becomes a matter of debate. The changes in the scattering properties of the sample form a continuum, yet we have to classify the measurement as being one or the other. The major difference between the two techniques depends, as you said, on whether the light primarily penetrates through the sample and is diffusely reflected by some sort of backing plate rather than by the sample itself. Therefore I have to conclude that it's more appropriate to define the label as based on the device used to hold the sample, without the sample itself being involved in the definition. Howard \o/ /_\ | ||
| Gabi Levin (gabiruth) Senior Member Username: gabiruth Post Number: 41 Registered: 5-2009 |
Dear all, Transflectance refers to the method used when measuring clear liquids, whereby the light is launched from one fiber, goes through the liquid, and hits a reflective surface, travels back into the recieving fiber. It has no application in fruits, as the existence of light sources powerful enough to go through the fruit, hit a reflector, and travel all the way back to the spectrometer. In fruits you can do diffuse reflectance, or, if you have the kind of instrument that has enough power to go through the whole fruit, exit on the other side and be collected into a grating and from there into a detector, you can do what some refer to as forward diffused transmission. This is called so because in difference to transmission through clear liquids where there is no scattering, in solids the light does not travel in straight lines. I hope this will help a little Gabi Levin, Brimrose | ||
| Francesco Davini (franz) Member Username: franz Post Number: 12 Registered: 2-2009 |
I do not have a enough scientific knowledge to answer your question from the point of view of theory and science. Anyway a lot depends on your matrices, the purpose of your analyses and the number of samples you want to analyze per day. On some instruments reflectance does not even require any cleaning between samples. It is just a matter of putting the sample in a cup, roughly levelling it and that's all. This is therefore a good choice in a very busy lab environment, or where your personnel is not totally affordable with sample preparation. Transflectance, on the other hand, may allow the analysis of a few more parameters, possibly with better accuracy if the sample preparation is well done and thorough cleaning of cells or cups is always guaranteed. This of course without knowing which parameters are going to analyze and where they are located in the fruits. I think we should know more about your needs: a technique can be defined better than another only when the full picture of the analytical requirements and sample appearence is well defined. Regards, Franz | ||
| S�verine Gabioud Rebeaud (gab) New member Username: gab Post Number: 3 Registered: 4-2010 |
Hi everybody, Does anyone have any recommendation for the most suitable system to measure fruit quality: Reflectance or transflectance mode? Thanks S�verine |