Latest / Hydroponics Daily / Predicting Cannabis Potency with Your Smartphone: Hyperspectral Breakthrough

Transcript

Predicting Cannabis Potency with Your Smartphone: Hyperspectral Breakthrough · Hydroponics Daily

• 0:00
  Hello there, and welcome back to
• 0:01
  Hydroponics Daily, your go-to podcast for everything soilless cultivation.
• 0:06
  I'm your host, Dr. Russell Sharp, the founder of Utreema Limited.
• 0:11
  Now, what if I was to tell you, you could take a picture of a legal cannabis
• 0:17
  plant, and before you harvest it,
• 0:20
  and tell the active ingredient concentration
• 0:24
  in there, the cannabinoid concentration in the final product.
• 0:27
  You'd say i'm crazy but there's a couple of bits of research that have just
• 0:31
  been well just i've just been made aware of that potentially means you could
• 0:37
  do this not only could you do this you don't need necessarily need a fancy piece
• 0:41
  of kit you could potentially do this with your.
• 0:44
  Smartphone so i'm going to go into how you
• 0:47
  might be able to do that and maybe a couple of you will want to play around
• 0:51
  with this and try if you are legal cannabis growers so the first piece of research
• 0:58
  I came across was work done by the University of Adelaide in collaboration with
• 1:03
  a German tech company called Compolitics.
• 1:07
  And they've developed a non-destructive leaf scanning method,
• 1:12
  they say can accurately predict the cannabinoid concentration
• 1:15
  in the cannabis plants so what it
• 1:18
  does is you take a picture and it's of the fan leaf
• 1:21
  and it's called hyperspectral reflectance so hyperspectral means across the
• 1:25
  whole light spectrum all the wavelengths you can buy hyperspectral cameras they're
• 1:31
  very expensive but this fan leaf hyperspectral flhr for short involves taking
• 1:36
  measurements across the plant's canopy during early and late flowering,
• 1:40
  and it was found to reliably predict the final cannabinoid concentration.
• 1:46
  Now, that measurement is then overlaid with machine learning models trained
• 1:51
  on the results to achieve a predictive accuracy.
• 1:55
  It's not just a case of taking the picture and then it tells you the cannabinoid concentration.
• 1:58
  You then have to put it into some fancy models that come out with it,
• 2:02
  you know, basically correlations so this
• 2:05
  will be probably be cultivar specific as
• 2:08
  well it uses a handheld non-destructive hyperspectral
• 2:12
  device that enables rapid in situ so
• 2:15
  in the in the grow room in the greenhouse in the grow room yep of
• 2:18
  intact fan leaves so you don't need to do sacrificial sampling
• 2:22
  which is always a problem when you're growing high value crops it
• 2:25
  uses specialized lighting and detectors to capture subtle
• 2:28
  differences in the light that's reflected from
• 2:32
  the plant's leaves and every chemical and structure feature in the leaf such
• 2:36
  as pigments water trichomes sugars reflect at different wavelengths so this
• 2:42
  work has been published in the in a journal called industrial crops and products
• 2:46
  and what you do is you scan across the whole spectrum but then focus on a.
• 2:53
  Which is the chemical signature, the spectral signature for each leaf.
• 2:58
  And then obviously the clever work is knowing the mathematical traits and things like that.
• 3:02
  So you could potentially predict the cannabinoid profile weeks before you harvest it.
• 3:08
  And you could then plan what you're going to do with the crop,
• 3:12
  or you could sell it based on this.
• 3:15
  So you could pre-sell your crop based on its THC or CBD or other cannabinoid concentration.
• 3:23
  And if you're growing things like hemp or cbd where you need to know the thc
• 3:28
  is below a certain limit that could also be important so this is an area i think will rapidly increase.
• 3:34
  And you know make and so on
• 3:38
  area to watch so is it the same for all strains
• 3:41
  of legal cannabis and could you use it on
• 3:44
  seedlings to faster breed
• 3:47
  strains that were extremely potent so
• 3:51
  you could screen 10 000 very very
• 3:54
  small seedlings for a very specific cannabinoid now
• 3:57
  it could be thc or cbd it could be cbg or
• 4:00
  some really obscure one that you struggle to
• 4:04
  identify and characterize in a lab and
• 4:07
  also you take ages to get all these mature plants to
• 4:10
  flowering and fruiting and to curing and extraction you just
• 4:13
  do a hyperspectral image you can determine the concentration
• 4:16
  of this cannabinoid and away you go remove all the
• 4:19
  other all the seedlings that were not high in that and.
• 4:22
  Then go again with those plants and take them to maturity use
• 4:26
  them for breeding so potentially could massively speed
• 4:29
  up the breeding of legal cannabis crops tremendously now
• 4:33
  this obviously requires a hyperspectral
• 4:36
  camera but could you get away with using your
• 4:40
  smartphone and that brings me on to the second piece of research that i
• 4:43
  have uncovered this week and that was brought
• 4:47
  to light by a slug network i'm involved with because
• 4:50
  i'm involved in a research project where
• 4:53
  they scan fields for slugs using
• 4:56
  hyperspectral cameras but one of
• 4:59
  them forwarded a interesting paper this week which
• 5:02
  i'm going to go into now and this is an article in ieee
• 5:07
  spectrum and it says the purdue university
• 5:10
  researcher young kim that's his name he's not just that's not like his rapper
• 5:15
  name young kim and his team have combined computer vision color science and
• 5:20
  optical spectroscopy to create an algorithm that recovers detailed spectral
• 5:26
  information from conventional photographs.
• 5:29
  And he says that, in contrast...
• 5:33
  To the human eye which picks up red blue and green in
• 5:37
  the visible range off the shelf smartphone camera sensors
• 5:40
  are potentially hyperspectral in nature meaning that
• 5:43
  each pixel is sensitive to far more spectral
• 5:47
  bands than we can actually see on the on the
• 5:49
  screen and scientists have now found a simple
• 5:52
  way of for any conventional smartphone camera
• 5:55
  to serve as a hyperspectral sensor and what
• 5:58
  they do is they place a card with a chart on
• 6:01
  it within its view so you
• 6:04
  have the picture of say the plant with or the
• 6:07
  slug in this case with with this
• 6:11
  chart and it potentially turns it into a hyperspectral camera so it's a patent
• 6:15
  pending technique so you can't just copy this it is well you could copy it but
• 6:20
  they'd have to find out and sue you and know that you were copying them that's
• 6:24
  the problem with patents so you could potentially get around it that way but
• 6:27
  yeah then you couldn't commercialize it.
• 6:29
  The new pages of Pretendant St. Meek may find applications in defense.
• 6:32
  Security, medicine, forensics, agriculture, environmental monitoring,
• 6:37
  industrial quality control, and food beverage analysis.
• 6:40
  Basically, anywhere we use hyperspectral analysis.
• 6:43
  It's been used in agriculture for a while, hyperspectral imaging.
• 6:46
  I used it back in like 2016 for looking at pest and disease outbreaks in potato crops.
• 6:55
  We flew a drone with a hyperspectral camera and very expensive hyperspectral
• 6:59
  camera from finland up in the james hutton institute up in dundee bit of a wash
• 7:06
  out that that study but yeah so potentially your phone can turn into hyperspectral
• 7:11
  camera you can buy phones or used to be able to buy.
• 7:14
  Phone add-ons like jcb phones with infrared
• 7:18
  cameras on them i don't know you can still buy them but if you're
• 7:21
  looking for like if you're looking to look at how hot
• 7:24
  your plants how stressed they were then you can see how hot
• 7:27
  they are the leaf temperature compared to the surroundings that's another good
• 7:30
  use of smartphones and spectral
• 7:33
  imaging on smartphones get a infrared imaging attachment
• 7:36
  to your smartphone and you can see how stressed your plants are how
• 7:39
  water stressed they are in your environment but back
• 7:42
  to hyperspectral imaging one of the examples they
• 7:45
  show is a iphone with a
• 7:48
  tiny little little it almost looks like
• 7:51
  a piece of modern art it's like a matchbox sized
• 7:54
  image trying to think of the artist who does it damien hirst it's like a damien
• 7:59
  hirst picture with all little pixels and picture colors and then a glass of
• 8:03
  whiskey and obviously whiskey is massive in japan and the research can image
• 8:08
  the transmission spectrum of high-end whiskeys thus determining its authenticity.
• 8:12
  So if you buy a bottle of whiskey that's extremely expensive you can just use
• 8:16
  this really simple method to see whether it is actually what they say is.
• 8:20
  And having mentioned high-end artists like Damien Hurst this
• 8:23
  technique can actually be used to look at pictures in art
• 8:26
  galleries and see if they're actually authentic whether the pigments are
• 8:30
  actually the same paints used by the artist or whether
• 8:32
  they're a fake so another interesting use of hyperspectral imaging
• 8:36
  and the authors acknowledge that there have been previous attempts to to use
• 8:40
  smartphones to get hyperspectral images but often these involve inaccuracies
• 8:46
  or bulky additions to the smartphone extra pieces of kit you have to bolt on
• 8:50
  like i said for that infrared one so potentially this could include infrared and,
• 8:55
  ultraviolet and all the other spectrums and wavelengths obviously it's probably
• 8:59
  gonna you You need the special card to put next to your plant.
• 9:03
  And you will need an app that can have the algorithms in and all the data and
• 9:09
  the data from the previous study on cannabinoids.
• 9:12
  So certainly not something that's coming anytime soon, but it's not beyond the
• 9:16
  wit of man to create a app and a card that you could put in front of a cannabis
• 9:21
  plant and it could tell you the cannabinoid concentration at harvest,
• 9:26
  which is amazing to think.
• 9:28
  It can't be done today but it could
• 9:31
  be done tomorrow if everyone's in the world put
• 9:34
  their minds to it which is probably not going to be a priority probably solve
• 9:38
  the war in ukraine or the problem in
• 9:41
  gaza first before we go on to determining that but it is possible so it's not
• 9:45
  impossible which is really interesting and could be around the corner if some
• 9:50
  clever bods get together and take all these technologies from these different
• 9:55
  companies and researchers and put them together in one smartphone application.
• 10:00
  Would require someone from outside of these teams to do it because the focus
• 10:04
  of the Purdue team will probably be in medical applications to begin with.
• 10:10
  It's not going to be cannabis breeding, certainly not, even if it is legal.
• 10:14
  But there, something to think about. If you've got an interesting take on this.
• 10:18
  If you think I've got something wrong or you'd like an addition or a correction, then let me know.
• 10:23
  If you found this podcast interesting, then please do leave an
• 10:26
  apple podcast review because it helps share the
• 10:30
  show and the topic of hydroponics with
• 10:33
  the wider community tremendously if you're not listening
• 10:36
  on apple podcasts don't bother leaving a review because
• 10:39
  it doesn't help share the podcast but do make sure you click follow
• 10:42
  so you don't miss an episode and if you're
• 10:45
  interested in novel technology check out utreema.co.uk
• 10:49
  because we have a completely unique range of
• 10:53
  fertilizers biostimulants and biopesticides
• 10:56
  that you won't find available from any other manufacturer and
• 11:00
  we sell into arable horticulture sports
• 11:03
  turf so whatever you're doing if you're using crop imports
• 11:06
  and plant health products we've probably got an innovative solution
• 11:09
  to solve one of your key problems of production and
• 11:13
  just like the slug people i talked about we've got a liquid slug
• 11:16
  product a ferric phosphate in us and we're the
• 11:18
  only people to create a soluble liquid
• 11:22
  ferric phosphate product and we do that because slugs don't
• 11:26
  tend to stand stay around long on the ground where
• 11:29
  slug pellets go they tend to go into the soil on the plant eat the plant eat
• 11:33
  the roots and so a liquid you can get anywhere where the slugs are most pellets
• 11:38
  you can't so hence the reason i'm involved in a slug project but that's enough
• 11:42
  on that it's not got much to do with hydroponics because we don't get many slug problems in.
• 11:47
  Hydroponics although we did get one case didn't we previously did an episode
• 11:50
  on a grower that was inundated with slugs in their hydroponic nft system but
• 11:55
  that's enough rambling from me today i hope you've enjoyed this podcast and
• 11:58
  i'll see you again tomorrow goodbye.