Q&A from Using Near-Infrared Analysis in Breeding Programs, Harvest, Post-Harvest & Packaging Innovation

October 19, 2019 at 12:04 am | Updated October 19, 2019 at 12:04 am | 16 min read

“What is your reference measuring procedure for titratable acidity? Is it a constant pH of 8.1? What is the concentration of NaOH you are adding?”

To answer your question, the titratable acidity studies that were done were before I was with the company, so I am actually not sure what their procedure was for doing the titratable acidity measurement. We have an auto-titrator, I know, in the lab, so I’m assuming that they used one of those. But I would assume that they used a procedure similar to what would be found in AOAC reference methods for food commodities. So, if you’re wondering where to find a good method for doing titratable acidity, I would recommend looking up the AOAC official methods, just to make sure that you’re getting as accurate a result as you can. But personally, I do not know, the study was conducted back in 2017, I believe, so for the blackberries and titratable acidity, I’m not 100% sure what their specific procedure was for that. But I would assume it would be similar to the AOAC reference methods.

“Hi, I see this is a handheld NIR that is held up to the sample. Do you have accessories or other NIR options?”

Subscribe to the Felix instruments Weekly article series.


By submitting this form, you are consenting to receive marketing emails from: . You can revoke your consent to receive emails at any time by using the SafeUnsubscribe® link, found at the bottom of every email. Emails are serviced by Constant Contact

We have multiple accessories that you can order with the F-750, one of the most important, I think, for people especially that are measuring small commodities, we have these cones that you can put, that essentially minimize any stray light that would be entering lens, so that you can place your sample on it, it will technically cover the entire lens, because the light is being funneled by that cone. And then, oh, I’m mistaken, the cones are actually included with all your F-750 purchases, so you don’t even have to ask for them, we’ll send them anyways. So, that’s an accessory that comes with the F-750. Other options, we do also have a liquids and powders kit, so that essentially utilizes a cuvette with a thick Teflon backing, and the piece fits over the lens and then you’re able to measure powders or liquids, and that’s especially helpful when it’s time to look at things like trade viability. But as far as other formats for NIR, we have our leaf spectrometer is a slightly different format, and we’re currently in development of a more bench top version of a liquid analysis system, an IR liquid analysis, so, for things like wine and olive oil, so that’s in the works, but it won’t be out for quite a while. So for now, the handheld NIR is what we mostly manufacture here.

“Where do you take your measurements in fruits, due to the variability of BRIX in different parts of fruits? How many measurements do you take per fruit, i.e. pears?”

That’s a really good question, and it’s like you said, not just BRIX, but even things like dry matter vary quite wildly. Dry matter in avocado, for instance, varies very wildly throughout the fruit. So, what we recommend, first of all, is doing do diligence as far as research is concerned. So, any research articles, any publications that have been made that kind of illustrate where in the fruit certain BRIX values or dry matters can be found, plotting out if the stem area is going to be higher or lower dry matter, nearest the stem, all the way down to the center of the fruit. For a lot of fruits I know that BRIX can vary as you get closer to, especially like stone fruit, as you get closer to the stone or you get closer to the center of the fruit, and so really the important part is that you get kind of an accurate average. Most fruit, there is going to be a spot where you can scan where you can get the most average or accurate representation of the BRIX or the dry matter throughout the entire fruit. So, what I would recommend, and for avocados, what we recommend for dry matter, is the dorsal side equator, and that’s kind of an area where when you penetrate half way into the pit from the skin, you’re going to get a pretty average value that is kind of a median value between the super low dry matter, and then the super, super high dry matter areas. So, for BRIX, kind of the same concept. You just want to make sure that you’re very consistent in how you do your analytical measurements and where you’re scanning. To increase your accuracy, and if you want to incorporate multiple scans, that is also something I highly encourage. Because all you’re going to do is increase your accuracy. So, with a pear, if you want to take a scan at the equator at a side that was facing the sun, versus the side that isn’t facing the sun, and then do the analytical measurements for both of those sides, and incorporate both of those into your model, then you know that you have an accurate prediction, no matter where you scan it, right? And that scan will actually give you a prediction that’s representative of the part of the fruit where you’re scanning. Or, alternatively, you can do a general, you know, take four scans around the whole equator and then take one dry matter measurement value, or the average of the dry matter from the four scans, and use that as your analytical value for those scans, and then you can be sure that no matter where you scan on the fruit, it’s going to give you that same average measurement. So, really, it’s up to your how you want to approach it. But it is something that needs to be taken into account when building models.

“How would you measure a large dimension of plastic pellets or particles that are in a homogenous” mixture I’m assuming is what they’re implying there.

If it has to be a large dimension, I’m kind of having trouble imaging what this is, so I’m just going to kind of imagine you have a large clear bag, like a gallon bag full of pellets or particles that are homogenously distributed throughout water. What you can do is you can just the instrument and put it up to, it is a homogeneous mixture, you can put it up to any point of that bag and take your scan, as long as the scan location is fully covering the lens. Or you can try to take a subsample of that and utilize our powders and liquids kit, and use that.

“How to make a calibration curve for fish?”

I would recommend that you check out our model building webinar, but essentially you’re just going to follow the same steps as you would for any other commodity like earlier in the slides, in the model building slide, it’s the same steps. You’re going to want to collect a sample of fish that you know for certain are within the range of what you’re going to be measuring and as far as something like maturity or age, or something like that, and you essentially want to make sure it’s representative across the entire distribution, and so you want to then take your scan, then take your destructive measurement, match up your destructive measurement to your scan, and then from there you will build your model, then validate it. So, similar to as any other commodity.

“What is the sample set for the applications following that since different methods require certain sample sizes, what techniques are used for predictions?”

Different methods require certain sample sizes, so all of these applications are on our website and I know that for most of them they use at least 70 to 100 samples, and that’s individual fruits, and then the techniques that are used to do the analytical values, if that’s what you’re asking. So, BRIX, we use a traditional refractometer, for dry matter we use typically a dehydrator, and then more recently we’ve been adopting new technologies like microwave moisture analysis, and then for titratable acidity, a classic titration method, but using an auto-titrator. I think that’s most of the parameters we’ve really looked at. For fat analysis and other things, we usually send it to a third party lab, an ISO accredited lab, so they would typically utilize an AOAC method of some sort.

“Has anyone used this for testing seed crop quality testing, like germination, viability, and vigor, using difference in absorbent spectrum?”

That’s a very interesting question. We have not had anyone use the 750 for seed crop stuff, but our leaf spectrometer, we’ve had requests of people wanting to use our CI-710 leaf spectrometer for seed crop quality testing, and I also have, we have a colleague here that also specializes in equipment for seed crop quality testing, and has some really cool technological tools, that I can put you in touch with, if you want to reach out to me via email, I can put you in touch with him. He has some really cool solutions for imaging seeds and stuff like that.

“Can it work in the field under full light at 105 degrees Fahrenheit?”

Yes, it can work in the field under full light, under the assumption that you are ensuring that the sample is completely covering the lens. So, you really want to make sure that when you take a scan, if you’re out in full light, that sample is completely covering that lens. The temperature, 105 is very, very hot, I hope you don’t have to work out in those conditions all the time, but the models can be built to include scans at varying temperatures. So, if you build your model with a high temperature of 105, a middle temperature of let’s say 80, and then a low temperature of 40, then you can be fully confident that your model is going to work across that entire temperature range. So, it’s really up to you what temperature range you’re intending the instrument to be used at. So, if it’s going to be at 105 most of the time, then you’re going to want to make sure you take all your scans at 105 degrees Fahrenheit, in full light, as well.

“How easy is it to enter the plot number in fruit subsample number for research trials? Is there a Bluetooth barcode scanner or wired connectivity option with unit for tracking sample IDs?”

So, that is something that we have been looking into, as it’s been requested by other customers to incorporate some code of barcode scanner. So, be on the lookout, that’s on our list of things to develop, but currently two of the easiest ways is to do enter those things directly on the instrument, or to take out the SD card and then use a notepad, the notepad application, to essentially go in and pre-type out all of the plot numbers and subsample numbers that you intend to measure, and then when you eject your SD card and put it back in the instrument, that will be ready for you to go ahead and start measuring, it will have all of those in there, and you can choose then on the instrument which plot number you’re going to be scanning under, and which fruit you’re going to be scanning under.

“Does it need full coverage of the window by the fruit?”

Yes, and if you’re out in the sunlight or in high lights, intense light situation, yes. Is it okay if you’re in a lower light situation like in a laboratory or something like that, and there is a sliver of the lens being uncovered, then that’s probably not going to influence it too much, but yes, typically the window needs to be fully covered, the lens needs to be fully covered, and those cones that we provide are a good solution for people that use smaller commodities.
Data logging is the next question, and I’m assuming that’s just asking how you go about with data transfer and things like that. So, there are a couple of options for data transfer. We utilize an SD card within the instrument that has wireless capabilities, well, the instrument has wireless capabilities, so you can either do it over WiFi or also it does have a Bluetooth option, I believe, but the most efficient way probably is just remove the SD card and put it into a computer and transfer over all your data. There are other solutions that we have been working on for individualized customers, so that might be something that if you have a larger scale operation that you need to have that integrated with, that could be a conversation that you have with us and our engineers. But otherwise, yeah, I would recommend the SD scar transfer.

“How much reference lab data is needed to build an acceptable calibration curve?”

This is a very common question that I get, and really, we recommend a minimum of 200 data points. The most important thing is that really there is no maximum, I guess, that you could go. You can continuously add data, is what I would recommend, over seasons, just keep on adding more data to the model to make sure that it’s hitting the best it can. But really, you also just really need to make sure that the data is fairly evenly distributed across the entire range of maturity of essentially your commodity.

“I was wondering if this tool can determine shoulder softness or blotching of the fruits?”

We do have some people currently investigating the use of the F-750 for things like internal rot and browning, and things like that, in apples specifically right now. Since we know that we can do internal flesh color, we are pretty confident that that is also a trait that you should be able to do, and since firmness is also a parameter that has been successfully modeled, in apples specifically, then I would think that shoulder softness would also be something that you could do with the F-750.

“Does the unit come with a set of models, or must each client develop their own models for unit to be useful?”

As I mentioned earlier, and I’m glad you asked this, because I just want to definitely reiterate. There are models that come on the F-750. These are demo models and they are not models to be used right out of the box. You have to, have to at least tailor those models, if not, build an entirely new model on your own, before you start getting accurate results.

“Please share data of mango fruit.”

We actually have on our F-751 product page at our website, there is a validation data tab, so you can go to our website, click on the F-751 mango, go to validation, and then you will be able to download our entire validation data set.

“What type of bond measured by the 750?”

So, since it sees in the range from 310 to 1100, it sees a lot of different CH bonds, as well as OH bonds, but generally speaking, every different bond that you could potentially see in fruit should be visible in the way of length range. Things that aren’t visible are things like polyphenols or individual fatty acids, things of that nature aren’t going to be visible within the wavelength range, but more simpler lipids, like total lipids and water and starches and protein, those should all be visible within the range.

“Can this device determine fruit color, as well?”

Yes, Dr. Osuna actually was able to make a well fitting model to determine skin color of mangos, and he used A-Star values for that. So it is definitely a viable trait, yes.

“Does the F-750 work with stone fruit?”

And the answer is yes, it does. The only thing is that we don’t have a model built for stone fruit right now. But we do have customers successfully using it, and as I mentioned earlier, peaches and then also nectarines, things of that nature.

“Can we measure betacarotene and lycopene content of tomato? Have we used it in rice? And for which traits of rice was it used?”

As I mentioned earlier, I know that Manjat was trying to use it for betacarotene in tomato, I’m not sure if he tried lycopene or not. But I think that with a more robust model building process, that betacarotene should be possible, like I said, I don’t know if anyone has tried lycopene, and I don’t know of anyone using it in rice.

“Is it possible to use with fluids?”

The answer to that is yes, we do have the liquids and powders kit and that is actually an instrument that am personally using right now to do some testing on liquids like wine and olive oil.

“Different wavelengths or wavelength ratios used for different parameters?”

They’re just asking, when you’re building the model, are you using different wavelength ranges for the different parameters? And yes, that is the case. But our software automatically will detect for you where the ranges of the best correlation are for your parameter versus your analytical value.

“Is it possible in the device for you display three predictions? Is there a manual for that?”

In the user manual for model builder and data viewer, there is a whole section on customizing your device and it is possible to display three predictions. Technically, if you have good enough eyes and you want to make the text small enough, you can put as many predictions as you want. But yeah, it’s possible, and I would recommend checking out the model builder and data viewer manual on our website.

“Is it a continuous spectrum if you ask them what is range?”

The range is 310 to 1100 nanometers.

“Are the models built under controlled temperatures? Would it work inside a color, or do you have to create a model under predicted conditions?”

So, as I mentioned earlier, if you’re going to be working in some sort of extreme environment, I recommend that when you build the model, you take your scans of the fruit that are held in that environment. And if you know that only going to be taking samples inside of a cooler, then you only need to take scans inside the cooler. But if you think you’re going to be taking scans in a cooler, as well as in a room temperature environment, then you’re going to have to create your model using scans from both of those different temperatures, so that it works at both of those temperatures.

“Is there a possibility to load an Excel list in the instrument so that the operator knows which sample plant number to scan for breeding?”

This is also where the barcode scanner comes into play, because we’ve had specifically it’s in breeding situations where the barcode scanner really can come in handy, but the software doesn’t use an Excel list, all you have to do is pull up the Notepad app, and you can just type in all of your sample and plant numbers that you’re going to be scanning into that, and that’s just on the SD card of the instrument.

“Have we considered widening the range to include more of the NIR region?”

Yes, we have, and actually, I don’t want to talk too much about it quite yet, because it’s very, very early stages of development, but I’m working on a more bench top version of the instrument that will be used for liquid analysis and it will include a higher range up to into the 2500 nanometer range, so we can capture more of the NIR spectrum.

“I mean that the plant number is shown one by one on the display.” [only a partial question was asked here]

I wish I remembered what your earlier question was, oh, for the breeding, okay, so on the display it will show you, you can customize the display to show you what lot, you can create kind of like an organizational tree, where you go, this is the plot that I’m on, this is the tree that I’m on, this is the fruit number that I’m on, and you can customize it so that every time you take a measurements it will increase incrementally by number, and then you will see which plot you’re on, you will see which tree you’re on, you will see which fruit you’re on. So it doesn’t have to be limited by just the plant number. So you can really customize it to view on the screen whatever really you need to view, I guess.

“Any model building video?”

So, if you tune in on November 5th, we will have demonstrations of that and we will obviously send that webinar out after we’re done. So I would recommend looking into that. Oh yeah, I guess we do have an older video on model building, I was just informed of that, so Suzie will put the link in the chat for you guys.

“What reference material is used to validate the spectrometer before use?”

We use Teflon, these Teflon cylinders, I guess is the best way to put it, and we have different calibration sets that have either a varying range of methanol, of concentrations of methanol internally or just solid Teflon, and we use those to ensure that the spectrometer is working the way we want it to before we start putting the models on and calibrating models.

“Can we measure lycopene quantity?”

As I mentioned earlier, I don’t know of anyone that’s tried it yet, but I know that betacarotene has been tried, so that’s something that you would have to do some research on, to see if lycopene can be seen in the spectral region of 310 to 1100 nanometers, see if anyone else has done it before.

“Is it possible to transfer models built on the 750 to another 750, and is there a need to calibrate first?”

Yes, it is definite possible to transfer models built one 750 to another 750. We do have a procedure for that, and it does require, because the internal components of each instrument are going to vary ever so slightly, just do to them being electronics, there is going to be some modification that needs to be one, so we have an actual process on our website, there is a detailed process, and also the Australian mango industry, they are experts at doing this, because every year when they all get together before the season, they do this process where they transfer basically the model from the golden unit to all of the subsequent units, and then they all calibrate it to be reading the same as well as golden unit. So that you can find on our website, and if you need more help you can reach out to us.

“Do you cover UV spectrum 310 nanometer plus, also?”

Our wavelength range is 310 up to 1100 nanometers.

“How many measurements are required for reliable model building?”

I’ve already mentioned this, so we say it’s a minimum of 200, but really you need enough to be representative of your sample population evenly distributed throughout maturity and you want to make sure you’re getting accurate results. So really, the more the better, and also continuously updating your model across harvest seasons is a good idea, as well.

“How deep into the fruit do you actually monitor?”

It depends on the density of the flesh and of the skin of the fruit. So something like kiwi, it penetrates very deep, and then in some fruit with thicker rinds, it penetrates approximately 1 cm. But in general, we say that it penetrates at least 1 cm into the fruit.

“In thick skin citrus fruits, can we differentiate between the peel and the fruit matters?”

If you built a model that way, you could. So I would assume that if you’re looking at something like I guess a dry matter of your citrus fruit, you could peel your fruit, take a scan of the flesh, and then also take a scan of the skin, and then get your analytical dry matter value for both the flesh and the skin, and you could have some sort of differentiation, but typically what would you do is you would just scan it through the skin, like in the mandarins and BRIX, since the skin isn’t going to contribute anything really to BRIX, you scan through the fruit and then you take the BRIX value from that scan site, so the model will associate the scan, which includes the skin and the flesh to be associated with that BRIX value of the flesh itself. So, if you wanted to be able to differentiate, you would have to build a model that way.

“Will the new version of the model builder be able to perform an easier way of validation of independent samples?”

Yes, I think that in general the new model builder is going to be a big hit for everyone because it’s just going to simplify things a lot more and it will make validation easier for clients to do, and it also is going to make my life easier, as I also do validation of our models. So, yes, is the answer to that.

“Has anyone used the instrument for fat and moisture analysis of olives?”

Actually, that is what I am working on currently. I have been developing, well, I’m currently working on developing a tool for fat and moisture analysis in olives, as well as for things like olive oils and olive pumice and wine, but for the olive oil, to determine things like, and olive pumice, for polyphenols, as well. But the fat and moisture tool, yes, is in development and hopefully a prototype will be ready to go before the next harvest season.

“Regarding the question about 105 degrees Fahrenheit, I was referring to the tolerance and accuracy of the hardware components.”

Okay, so our instruments are designed to withstand up to 50 C, so you should be okay, I think, at 105. I wouldn’t recommend leaving it out just to sit and soak in the sun for 12 hours a day, but if you’re going to be actually utilizing it and walking around, then it should be okay.

“Does it work in dark conditions?”

The answer is yes for that.

“In big fruit with thick peel, like watermelon, do you think we can get accurate BRIX data?”

I have actually talked to multiple customers that are wanting to do stuff with watermelon. You will be hard pressed to find a lamp that’s powerful enough to penetrate the rind of a watermelon. The way to get around this is to, as that picture showed earlier in the presentation, is to peel away the rind until you can just barely expose the flesh, and then take a scan from that site, and then once you take your reference values from that scan site, then you should be able to get a pretty accurate model built for BRIX in watermelon. Unfortunately, it’s destructive, slightly, but it will be still an accurate and faster way than the traditional BRIX measurement.

“How much of a problem is it when the surface of the product is wet, e.g. recently washed fruit and vegetables, or cut watermelons, or cut tomatoes?”

Wetness, actually moisture will influence the results and I would recommend, because it’s going to cause some reflectance and the interactance is going to change with the skin. The same kind of goes with dirt, so you want to make sure that your fruit isn’t completely covered in dirt, but water also is the same. I would say that you want to make sure your commodity is dried as much as possible, but if when you created the model, you created with model with scans that were on wet vegetables, then you should be fine to scan wet vegetables, it’s just that if your model was built with dried produce, then when you go to scan wet produce, that moisture on the outside is going to interact with the light from the lamp and it going to cause some issues with the prediction that comes out. Hopefully that helps answer your question.