| Near-Infrared (NIR) supported by image analysis and machine learning as a fast screening tool for cannabis flower composition analysis by remote sensing and spectral markers for marker assisted breeding.

| Near-Infrared (NIR) supported by image analysis and machine learning as a fast screening tool for cannabis flower composition analysis by remote sensing and spectral markers for marker assisted breeding.

| Near-Infrared (NIR) supported by image analysis and machine learning as a fast screening tool for cannabis flower composition analysis by remote sensing and spectral markers for marker assisted breeding.

Oded Shoseyov Ph.D.

 

In recent years, the use of Cannabis in the adult population has dramatically increased due to legalization of Cannabis in parts of the United States and the use of Cannabis or Cannabis extracts as palliative medication. One of the main barriers hindering the use of Cannabis flowers as medicinal agents is the lack of uniformity in terms of the content of active ingredients. Moreover, Cannabis in nature is highly pollinated, which is the reason for extremely inhomogeneous plant material, hence obtaining new hybrid seeds with consistent genetic makeup is almost an unreachable goal. Furthermore, to stabilize the active ingredients’ concentration in Cannabis plants is even more challenging since it is affected by various factors: plant genetics, growing and storage conditions, the state of maturity at harvest, etc. This is the major reason for avoiding producing Cannabis seeds worldwide and using vegetative propagated plants instead. All these reasons together cause difficulty in the repeatability of the patient’s dosage. Thus, in order to promote Cannabis flowers as validated medicine, the ability to quantify the exact percentage of active ingredients in the plant is required.

The active ingredients of cannabis plants are the cannabinoids, a class of diverse chemical compounds that are concentrated in specialized glandular structures called trichomes. The primary cannabinoid is phytocannabinoid tetrahydrocannabinol (THC) while cannabidiol (CBD), another cannabinoid, is the primary analgesic compound of Cannabis. In addition, there are at least 85 additional plant cannabinoids having varied physiological effects. [1]. Gas Chromatography (GC) and High Performance Liquid Chromatography (HPLC) are commonly used approaches for quantification of plant cannabinoids.[2] Although those approaches can be used to type Cannabis strains according to THC and CBD quantities, they require processed plants’ material and are time consuming. Moreover, they can only be implemented on mature flowers, which requires growing a diversity of all plants before selection can be made. Therefore, in order to promote cannabis as medicine there is a need to act on two parallel platforms; stabilization of the active ingredients in the cannabis plant while producing the ability to quantify the percentage of active ingredients in each flower. One possible approach for quantifying active ingredients in whole (unprocessed) plants or production of Cannabis seeds and breeding acceleration process might be achieved by NIR technology.[3] NIR spectrometry supported by image analysis and machine learning can be used to detect and quantify cannabinoids in whole plants or plant material by remote sensing[4]. Recently, my laboratory has revealed the ability to accurately analyse THC levels in Cannabis flowers by NIR spectroscopy (Fig.1). NIR spectrometers/cameras record the absorbance/reflectance spectrum of samples irradiated with light at wavelengths between 700 nm and 2500 nm. NIR radiation is highly penetrative and can be applied to a sample without any preparation/destruction. Although the resulting absorbance or reflectance spectra is not highly discriminative, it can be used to quantify active ingredients and other agricultural features such as germination rate, male/female plants, disease resistance etc., by using NIR calibration models. This new system technology based on NIR spectroscopy will be able to determine cannabinoids content in Cannabis flowers as well as genetic features in Cannabis seeds, and most notably the active ingredients ratio. By this novel technology, we will be one-step ahead towards implementation of consistent cannabis flowers for medical use. Moreover, we will achieve a shorter and much cheaper breeding process, which can cause a dramatic change in the entire plant breeding industry.
 

REFERENCES

  1. El-Alfy, Abir T.; Ivey, Kelly; Robinson, Keisha; Ahmed, Safwat; Radwan, Mohamed; Slade, Desmond; Khan, Ikhlas; Elsohly, Mahmoud; Ross, Samir (2010). "Antidepressant-like effect of Δ9-tetrahydrocannabinol and other cannabinoids isolated from Cannabis sativa L". Pharmacology Biochemistry and Behavior 95 (4): 434–42.
  2. F.E. Dussy, C. Hamberg, M. Luginbuhl, T. Schwerzmann, T.A. Briellmann, Forensic Sci. Int. 149 (2005). Isolation of Δ9 -THCA-A from hemp and analytical aspects concerning the determination of Δ9 -THCin cannabis products. Forensic Science International 149(1):3-5.
  3. T. H. Reijmers; C. Maliepaard; H. C. Van Den Broeck; R. W. Kessler; M.A.J. Toonen; H. Van Der Voet (2005). Integrated statistical analysis of cDNA microarray and NIR spectroscopic data applied to a hemp. Journal of Bioinformatics and Computational Biology. 3(4): 891-913.
  4. I. Azaria.; N. Goldshleger.; E. Ben-Dor.; R. Bar-Hamburger (2009). Detection of Cannabis Plants by Hyper-Spectral Remote Sensing Means. Tel Aviv University Publication.

 

 Fig. 1: Total THC content in Cannabis flower by NIR spectroscopy analysis. This figure displays the correlation between the NIR data (y-axis) and the known THC level analyzed by HPLC (x-axis). The dots color represent Cannabis species. The middle dashed line denotes the best-fit model, while the other two dashed lines denotes the 20% error border (err<20% = 86%).