3686acsomegaACS OmegaACS OmegaAmerican Chemical SocietyPMC1201972112019721120197214029091510.1021/acsomega.4c10753New Analytical
Screening Method for Fast Classification
of Hemp Oil Based on THC ContentE A SouzaThaineh†‡BertolGustavo‡M SantosPoliana†*Universidade
Tecnológica Federal do Paraná, 81280-340 Curitiba, Paraná, BrazilDall
PhytoLab SA, 82540-040 Curitiba, Paraná, Brazil
This study presents a novel analytical approach for classifying
commercial hemp oil samples according to their Δ9-tetrahydrocannabinol
(THC) content, employing mid-infrared (MIR) spectroscopy combined
with machine learning algorithms. A total of 204 commercial hemp oil
samples, with THC concentrations ranging from 0.0% to 16.6% w/w, were
analyzed. Partial least-squares-discriminant analysis (PLS-DA) was
employed for classification purposes. Two classification models were
developed based on international regulatory thresholds: model A, which
classifies samples with THC concentrations exceeding 0.2% w/w, and
model B, designed to classify those with THC levels above 0.3% w/w.
Both models demonstrated good performance, achieving accuracy values
higher than 88.50%. Notably, model B reduced false negatives, improving
sensitivity (STR) values from 93.75% to 98.31% for the training set
and from 77.27% to 95.00% for the test set, compared to model A. This
approach offers a viable alternative to conventional laboratory methods
by eliminating complex sample preparation steps and enabling simple
and rapid THC screening.
Received 2024 Nov 26; Accepted 2025 Apr 1; Revised 2025 Mar 26; Collection date 2025 Apr 22.
Introduction
In recent years, the use
of cannabis-based products for medicinal
purposes has increased significantly due to their therapeutic benefits.
Derived from the well-known Cannabis sativa L. plant, which contains at least 554 identified compounds, including
more than 100 cannabinoids,1 their potential
health benefits are mainly attributed to the pharmacological properties
of the cannabinoids Δ9-tetrahydrocannabinol (THC) and cannabidiol
(CBD).2 THC, known for its psychoactive
effects, has shown potential in stimulating appetite in patients with
cancer and HIV/AIDS and in reducing nausea and vomiting, particularly
in those undergoing chemotherapy.3 In contrast,
the nonpsychoactive CBD has demonstrated anticonvulsive, anti-inflammatory,
analgesic, and neuroprotective properties.3−5Figure 1 illustrates the chemical structures
of the cannabinoids THC and CBD. Both compounds have the same molecular
formula, C21H30O2, and consist of
a cyclohexene ring, a phenolic ring, and a pentyl side chain. Despite
these similarities, THC contains a cyclic ether ring, whereas CBD
features a hydroxyl group. This subtle difference in molecular structure
underlies their distinct pharmacological properties.6
Chemical structures of (a) THC and (b) CBD cannabinoids.
Currently, a wide range of cannabis-based products
are available
on the market. Consequently, many countries have implemented specific
legislation to regulate their production and distribution. These legislations
include limits on THC content. Ensuring that cannabis-derived products
meet these regulatory thresholds is essential, not only for legal
compliance but also for consumer safety, as THC levels directly affect
the psychoactive effects of these products. Additionally, maintaining
appropriate THC levels helps direct the product toward the desired
clinical applications, considering the distinct pharmacological actions
of CBD and THC. In the European Union, the lowest limit is typically
set at 0.2% w/w,7,8 whereas in the USA, it is set
at 0.3% w/w.9 In Brazil, the National Health
Surveillance Agency (ANVISA) has established that THC content must
not exceed 0.2% w/w.10 However, an exception
applies to products intended for palliative care in patients with
irreversible or terminal conditions, where other therapeutic options
are not viable.10
Liquid chromatography,11−13 alongside gas chromatography,14,15 has been the preferred
technique for quantifying THC in cannabis-based
products. However, these methods often require time-consuming sample
preparation procedures and involve the use of large amounts of organic
solvents, expensive instruments, and analytical standards, making
them less feasible for large-scale or in situ applications. As the
demand for cannabis products continues to grow, alternative analytical
methods have been developed to address these limitations, focusing
on speed, environmental sustainability, and cost-effectiveness.16−21
In this context, infrared spectroscopy, particularly in the
mid-infrared
(MIR) and near-infrared (NIR) regions, has gained attention as a promising
alternative for the rapid and nondestructive analysis of cannabis-based
products. These techniques enable direct analysis without extensive
sample preparation and can be easily adapted for in situ analysis
using portable equipment. Previous studies have successfully employed
infrared spectroscopy to quantify THC in fluids,22 hemp oil,23,24 cannabis flowers, and extracts.8,25 However, despite these advancements, there is still a lack of simple
screening methods for the rapid classification of commercial hemp
oil based on THC content. The methods reported in the literature have
concentrated on analyzing the cannabis plant itself, often aiming
to classify it according to growth stage,26 chemovar,27 or fiber type.28 In this sense, a practical and cost-effective
approach for fast screening of THC content has not been proposed yet.
Recently, Monari et al. (2025) investigated the use of an electrochemical
sensor to recognize extracts of Cannabis sativa L. based on THC content.29 Classification
models were developed using the partial least-squares-discriminant
analysis (PLS-DA) method, which successfully classified the samples
into legal and illegal categories, with threshold limits set at 0.3%
and 0.6% w/w. However, it is important to highlight that the study
focused on cannabis extracts. Therefore, developing new strategies
to facilitate and improve the control of commercial cannabis-based
products is crucial, considering the potential impact of THC on health.
In this context, the present study pioneeringly proposes the use
of MIR spectroscopy combined with the PLS-DA method as an alternative
approach for the rapid qualitative discrimination of commercial hemp
oil samples based on their THC content. The models were validated
by estimating the sensitivity rate (STR), specificity rate (SPR),
and accuracy figures of merit. The advantages of the proposed method
include speed, simplicity, cost-effectiveness, and the elimination
of sample preparation requirements.
Materials and MethodsSamples
This study analyzed a total
of 204 commercial hemp oil samples with THC concentrations spanning
from 0.0% to 16.6% w/w. The overall sample set comprised full-spectrum
and CBD/THC isolated products from a wide variety of manufacturers
from different countries (Brazil, United States, Switzerland, and
Colombia). THC concentrations were determined using high-performance
liquid chromatography with diode-array detection (HPLC–DAD).
Analyses were conducted on an Agilent 1260 Infinity HPLC system (Agilent
Technologies Inc.), with the DAD set to 228 nm. Separation was achieved
using a C18 chromatographic column (150 × 4.6 mm, 2.7 μm).
The mobile phase consists of trifluoroacetic acid in acetonitrile,
mixed in a 41:59 (v/v) ratio. The trifluoroacetic acid solution was
prepared at 1% (v/v) by diluting the stock solution in purified water
(resistivity >18 MΩ·cm).
Mid-Infrared Spectroscopy
The spectral
measurements were performed using a Cary 630 Fourier transform infrared
(FTIR) spectrometer (Agilent Technologies Inc.), equipped with an
attenuated total reflection (ATR) module. The spectra were collected
in the range of 4000–650 cm–1, with a resolution
of 4 cm–1, averaging a total of 16 scans.
Data Analysis
PLS-DA was performed
using MATLAB 2021a (MathWorks, Natick, USA), supported by the PLS_Toolbox
8.9.2 (Eigenvectors Research Inc., Manson, USA). Prior to the analysis,
the data were preprocessed using the first derivative (Savitzky–Golay
with a second-order polynomial and a 15-point window), followed by
mean centering.
Considering global regulatory standards, PLS-DA
models were developed with two different classification purposes.
The first approach focused on distinguishing hemp oil samples with
THC concentrations greater than 0.2% w/w (model A), whereas the second
approach aimed to distinguish hemp oil samples with THC concentrations
exceeding 0.3% w/w (model B). The Kennard–Stone algorithm was
used to split the data into training and test sets, with 75% of the
samples allocated to the training set and the remaining 25% to the
test set. The number of latent variables (LV) was chosen by random
subsets cross-validation (6 splits and 10 iterations), based on the
smallest number of classification errors.
The assessment of
the classification models was performed using
sensitivity rate (STR), specificity rate (SPR), and accuracy. STR
is defined as the ratio between true positives (TP) and the sum of
TP and false negatives (FN), while SPR is the ratio between true negatives
(TN) and the sum of TN and false positives (FP). Accuracy is the ratio
of the sum of TP and TN to the total sum of TP, TN, FP, and FN. TP
represents the number of correct predictions where the model accurately
identified the positive class, i.e., samples with THC concentrations
above the threshold (0.2% or 0.3% w/w) classified as positive. Similarly,
TN denotes the number of correct predictions for the negative class,
i.e., samples with THC concentrations below the threshold classified
as negative. FP refers to samples that do not belong to the positive
class but were incorrectly classified as such, i.e., samples with
THC concentrations below the threshold misclassified as positive.
Conversely, FN indicates samples that belong to the positive class
but were misclassified as negative, i.e., samples with THC concentrations
above the threshold incorrectly identified as negative.
Results and DiscussionMIR Spectra
Figure 2 shows the average MIR spectrum of the 204
hemp oil samples and the carrier oil. Hemp oil is a common product
derived from Cannabis sativa L. plants,
whereas carrier oil is a vegetable oil used to dilute the Cannabis sativa L. extract. Many different oils can
be used as carrier oils, with medium-chain triglyceride (MCT) being
one of the most used. The general structure of MCT (Figure S1, Supporting Information) consists of three saturated
fatty acids attached to a glycerol backbone. The R groups represent
fatty acid molecules with chain lengths ranging from six to 12 carbon
atoms.30
Average MIR spectrum of the 204 hemp oil
samples and the carrier
oil.
Analysis of the results indicates that the dominant
signals observed
in the MIR spectrum of hemp oil samples are primarily attributed to
the components of the carrier oil. The strong absorptions at 2924
and 2855 cm–1 are associated with asymmetrical and
symmetrical stretching modes of CH2 and CH3 of
methylene groups. The band at 1740 cm–1 is attributed
to the C=O stretching mode of ester carbonyls, whereas the
bands at 1460 and 1377 cm–1 are associated with
C–H bending vibrations of CH2 and CH3 in aliphatic groups. The bands in the 1151–1030 cm–1 region are attributed to the C–O stretching vibration of
acyl ester groups. The weak absorption bands at 983 and 887 cm–1 have been associated with C–H bending vibrations
and =CH2 wagging vibrations, respectively. In addition,
the band at 725 cm–1 is related to the overlapping
of CH2 rocking vibrations and out-of-plane vibration.
Although the spectra of hemp oil samples and the carrier oil are
very similar, some important differences were identified, such as
the small bands around 1621 and 1579 cm–1, which
correspond to the C=C stretching vibration of aromatic rings.
This signal also appears in the MIR spectrum of the THC cannabinoid
standard shown in Figure S2 (Supporting Information). The THC cannabinoid standard spectrum was measured by placing
5 μL of a 1000 mg L–1 standard solution on
the ATR crystal surface and allowing the solvent to evaporate in airflow.
The MIR spectrum of the THC standard (Figure S2, Supporting Information) displays additional prominent signals
in the fingerprint region (1600–800 cm–1),
attributed to C=O stretching, C=C stretching vibrations,
C–H bending, C–O–H in-plane bending, and C–O
stretching vibrations. Additionally, the band between 3700 and 3100
cm–1 corresponds to the O–H stretching modes
of the solvent.
PLS-DA Model AnalysisModel A
Considering the regulatory
standards in Brazil and some European countries, PLS-DA models were
developed to target samples with THC concentrations exceeding 0.2%
w/w. In this approach, samples with THC ≤ 0.2% w/w were assigned
a y-value of 0.0, while those with THC > 0.2%
w/w
were assigned a y-value of 1.0. The training set
comprised 152 hemp oil samples (88 samples with THC concentrations
≤ 0.2% w/w and 64 samples with THC concentrations > 0.2%
w/w),
while the test set consisted of 52 samples (30 samples with THC concentrations
≤ 0.2% w/w and 22 samples with THC concentrations > 0.2%
w/w).
The PLS-DA model was built with 5 LV, accounting for 99.04%
of
the spectral variance (X block) and 75.64% of the
class variance (y block). The model results are presented
in the form of a confusion matrix (Table 1). In the training set, 87 out of 88 commercial
hemp oil samples were correctly classified as THC concentrations ≤
0.2% w/w, while 60 out of 64 samples were correctly identified as
THC concentrations > 0.2% w/w. However, the model produced one
FP,
misclassifying one commercial hemp oil sample with THC ≤ 0.2%
w/w as THC > 0.2% w/w, and four FN, misclassifying four samples
with
THC > 0.2% w/w as THC ≤ 0.2% w/w. For the test set, all
commercial
hemp oil samples with THC concentrations ≤ 0.2% w/w were correctly
classified, while 17 out of 22 samples with THC concentrations >0.2%
w/w were correctly identified, resulting in five FN. Among the nine
FN (four in the training set and five in the test set), seven samples
had THC levels between 0.21% and 0.29% w/w, which is notably close
to the established class threshold. The remaining two samples exhibited
THC levels of 0.41% and 0.54% w/w.
Confusion Matrix of Training and Test
Sets of Model A
actual
classes
predicted classes
THC ≤ 0.2% w/w
THC > 0.2% w/w
Training
THC ≤ 0.2% w/w
87
4
THC > 0.2% w/w
1
60
Test
THC ≤ 0.2% w/w
30
5
THC > 0.2% w/w
0
17
The qualitative figure of merit for the model is shown
in Table 2. Overall,
the model
demonstrated good performance, achieving a SPR of 98.86% and 100.00%
for the training and test sets, respectively, indicating its effectiveness
in accurately identifying TN samples. The STR, which measures the
model’s ability to correctly identify TP samples, was 93.75%
for the training set (60 out of 64 samples correctly classified).
For the test set, the STR was 77.27% (17 out of 22 samples correctly
classified). Model A achieved an accuracy exceeding 88%.
Performance Parameters for the PLS-DA
Classification Models Developed to Discriminate Hemp Oil Samples with
THC Concentrations Greater Than 0.2% w/w (Model A) and 0.3% w/w (Model
B)<xref rid="t2fn1" ref-type="table-fn"><sup>a</sup></xref>
The variable importance in the projection (VIP) scores
graph (Figure
S3, Supporting Information) indicated that
the most relevant bands for sample discrimination are found in the
fingerprint region (650–1800 cm–1). Another
important region spans from 1700 to 1000 cm–1, associated
with C=O, C=C, and C–O stretching vibrations,
as well as C–H bending vibrations. Although the signal around
2100 cm–1 has a VIP score above 1, it is linked
to the diamond crystal of the ATR accessory and does not provide meaningful
information. The regression vector (Figure S4, Supporting Information) revealed that the most important wavenumbers
for class 0 (hemp oil samples with THC ≤ 0.2% w/w) are 1735,
1608, 1561, 1412, and 1036 cm–1. For class 1 (hemp
oil samples with THC > 0.2% w/w), the most important wavenumbers
are
1714, 1630, 1585, 1436, 1192, and 1056 cm–1.
Model B
In the second approach,
in compliance with USA legislation, the THC concentration threshold
was established at 0.3% w/w. Accordingly, samples with THC levels
of ≤ 0.3% w/w were assigned a y-value of 0.0,
while those with THC levels >0.3% w/w received a y-value of 1.0. The training set comprised 153 hemp oil samples, with
94 samples having THC concentrations of ≤ 0.3% w/w and 59 samples
with concentrations >0.3% w/w. The test set consisted of 51 hemp
oil
samples, with 31 samples having THC concentrations of ≤ 0.3%
w/w and 20 samples with concentrations >0.3% w/w.
The final
model was constructed with 5 LVs, explaining 98.84% and 82.04% of
the variance in the X and y blocks,
respectively. As shown in Table 3, one FP (a commercial hemp oil sample with THC concentration
≤ 0.3% w/w incorrectly classified as THC concentration >0.3%
w/w) and one FN (a commercial hemp oil sample with THC concentration
>0.3% w/w incorrectly classified as THC concentration ≤
0.3%
w/w) were observed during the training phase. These errors resulted
in SPR and STR values of 98.94% and 98.31%, respectively. During the
test phase, only one FN was observed, leading to SPR and STR values
of 100.00% and 95.00%, respectively. The FN samples had a THC concentration
of 0.41% and 0.54%. The qualitative figure of merit of the model is
summarized in Table 2.
Confusion Matrix of Training and Test
Sets of Model A
actual
classes
predicted classes
THC ≤ 0.3% w/w
THC > 0.3% w/w
Training
THC ≤ 0.3% w/w
93
1
THC > 0.3% w/w
1
58
Test
THC ≤ 0.3% w/w
31
1
THC > 0.3% w/w
0
19
The VIP scores graph (Figure S5, Supporting Information) showed a profile very similar to that of model
A, with the most relevant signals located in the fingerprint region
(650–1800 cm–1).
When comparing the
performance of the models, a significant reduction
in false negative samples was achieved in model B. In the training
set, the number of false negatives decreased from four to one, while
in the test set, it decreased from five to one. As a result, the STR
values increased from 93.75% to 98.31% for the training set and from
77.27% to 95.00% for the test set. Accordingly, the overall accuracy
improved from 96.50% to 98.62% for the training set and from 88.50%
to 97.50% for the test set. This improvement is attributed to the
simple strategy of avoiding false negatives by raising the threshold.
Conclusions
This study successfully
demonstrates, for the first time, the feasibility
of combining MIR spectroscopy with PLS-DA for the qualitative discrimination
of commercial hemp oil samples based on their THC content. The proposed
method is simple, fast, and cost-effective, enabling direct analysis
without the need for sample preparation, thereby reducing waste generation.
Additionally, the use of portable equipment facilitates in situ screening,
allowing the identification of potentially noncompliant samples for
further confirmatory analysis.
The classification models showed
good accuracy, although their
performance may be affected when THC levels are very close to the
threshold, particularly at 0.2% (w/w). However, it is important to
emphasize that the majority of regulations worldwide permit THC concentrations
above 0.3%, and in these cases, the model performs exceptionally well,
making it a valuable tool for rapid screening and compliance verification.
Acknowledgments
We would like to thank Agilent Scientific for supplying
the portable spectrometer and providing technical support. We would
also like to thank Dall PhytoLab for granting access to the commercial
hemp samples and their CBD/THC content used in this study.
Supporting Information Available
The Supporting Information is
available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.4c10753.
General structure of medium-chain triacylglycerol, MIR
spectrum of THC standard solution, VIP scores, and regression vector
graphics (PDF)
The Article Processing
Charge for the publication of this research was funded by the Coordenacao
de Aperfeicoamento de Pessoal de Nivel Superior (CAPES), Brazil (ROR
identifier: 00x0ma614).
The authors
declare no competing financial interest.
Supplementary Material
ao4c10753_si_001.pdf
ReferencesAizpurua-Olaizola O.; Soydaner U.; Öztürk E.; Schibano D.; Simsir Y.; Navarro P.; Etxebarria N.; Usobiaga A.
Evolution of the Cannabinoid and Terpene Content during
the Growth of Cannabis sativa Plants from Different Chemotypes. J. Nat. Prod.
2016, 79 (2), 324–331. 10.1021/acs.jnatprod.5b00949.
ElSohly M. A.; Slade D.
Chemical Constituents of Marijuana:
The Complex Mixture of Natural
Cannabinoids. Life Sci.
2005, 78 (5), 539–548. 10.1016/j.lfs.2005.09.011.
Grotenhermen F.; Müller-Vahl K.
Medicinal
Uses of Marijuana and Cannabinoids. Crit. Rev.
Plant Sci.
2016, 35 (5–6), 378–405. 10.1080/07352689.2016.1265360.Mechoulam R.; Peters M.; Murillo-Rodriguez E.; Hanus L. O.
Cannabidiol—Recent
Advances. Chem. Biodivers.
2007, 4 (8), 1678–1692. 10.1002/cbdv.200790147.
Khan R.; Naveed S.; Mian N.; Fida A.; Raafey M. A.; Aedma K. K.
The Therapeutic Role of Cannabidiol
in Mental Health:
A Systematic Review. J. Cannabis Res.
2020, 2 (1), 2–21. 10.1186/s42238-019-0012-y.
Martinez
Naya N.; Kelly J.; Corna G.; Golino M.; Abbate A.; Toldo S.
Molecular and Cellular Mechanisms of Action of Cannabidiol. Molecules
2023, 28 (16), 5980. 10.3390/molecules28165980.
Duchateau C.; Kauffmann J.-M.; Canfyn M.; Stévigny C.; De Braekeleer K.; Deconinck E.
Discrimination of Legal and Illegal
Cannabis spp. According to European Legislation Using Near Infrared
Spectroscopy and Chemometrics. Drug Test. Anal.
2020, 12 (9), 1309–1319. 10.1002/dta.2865.
Deidda R.; Coppey F.; Damergi D.; Schelling C.; Coïc L.; Veuthey J.-L.; Sacré P.-Y.; De Bleye C.; Hubert P.; Esseiva P.; et al. New Perspective
for the In-Field Analysis of Cannabis Samples Using Handheld Near-Infrared
Spectroscopy: A Case Study Focusing on the Determination of Δ9-Tetrahydrocannabinol. J. Pharm. Biomed. Anal.
2021, 202, 114150. 10.1016/j.jpba.2021.114150.
Alharbi Y. N.
Current
Legal Status of Medical Marijuana and Cannabidiol in the United States. Epilepsy Behav
2020, 112, 107452. 10.1016/j.yebeh.2020.107452.
ANVISA . Resolução da Diretoria Colegiada
(RDC) no
327, de 9 de Dezembro de 2019; Ministry of
Health National Health Surveillance Agency, 2024. Disponível em: https://www.in.gov.br/en/web/dou/-/resolucao-da-diretoria-colegiada-rdc-n-327-de-9-de-dezembro-de-2019-232669072. Accessed on: March 1 2024.Christodoulou M. C.; Christou A.; Stavrou I. J.; Kapnissi-Christodoulou C. P.
Evaluation
of Different Extraction Procedures for the Quantification of Seven
Cannabinoids in Cannabis-Based Edibles by the Use of LC-MS. J. Food Compos. Anal.
2023, 115, 104915. 10.1016/j.jfca.2022.104915.Ciolino L. A.; Ranieri T. L.; Taylor A. M.
Commercial
Cannabis Consumer Products
Part 2: HPLC-DAD Quantitative Analysis of Cannabis Cannabinoids. Forensic Sci. Int.
2018, 289, 438–447. 10.1016/j.forsciint.2018.05.033.
Hall D. R.; Sinclair J. S.; Bhuyan D. J.; Khoo C.; Li C. G.; Sarris J.; Low M.
Quality Control
of Cannabis Inflorescence
and Oil Products: Response Factors for the Cost-Efficient Determination
of Ten Cannabinoids by HPLC. Talanta Open
2022, 5, 100112. 10.1016/j.talo.2022.100112.Franzin M.; Ruoso R.; Del Savio R.; Niaki E. A.; Pettinelli A.; Decorti G.; Stocco G.; Addobbati R.
Quantification
of 7 cannabinoids in cannabis oil using GC-MS: Method development,
validation and application to therapeutic preparations in Friuli Venezia
Giulia region, Italy. Heliyon
2023, 9, e15479 10.1016/j.heliyon.2023.e15479.
Ciolino L. A.; Ranieri T. L.; Taylor A. M.
Commercial Cannabis
Consumer Products
Part 1: GC–MS Qualitative Analysis of Cannabis Cannabinoids. Forensic Sci. Int.
2018, 289, 429–437. 10.1016/j.forsciint.2018.05.032.
Birenboim M.; Rinnan Å.; Kengisbuch D.; Shimshoni J. A.
Novel Fluorescence
Spectroscopy Coupled with PARAFAC Modeling for Major Cannabinoids
Quantification and Identification in Cannabis Extracts. Chemom. Intell. Lab. Syst.
2023, 232, 104717. 10.1016/j.chemolab.2022.104717.Valdes N. B.; Gorziza R.; Almirall J. R.
Quantitative
Analysis of Δ9-Tetrahydrocannabinol
(Δ9-THC) in Cannabis Plants Using the Fast Blue BB (FBBB) and
4-Aminophenol (4-AP) Colorimetric Tests. Talanta
Open
2024, 9, 100287. 10.1016/j.talo.2024.100287.Porcu S.; Tuveri E.; Palanca M.; Melis C.; La Franca I. M.; Satta J.; Chiriu D.; Carbonaro C. M.; Cortis P.; De Agostini A.; et al. Rapid In Situ Detection
of THC and CBD in Cannabis sativa L. by 1064 nm Raman Spectroscopy. Anal. Chem.
2022, 94, 10435–10442. 10.1021/acs.analchem.2c01629.
Houtzager D.; Armenta S.; Herrero-Martínez J. M.; Martínez-Pérez-Cejuela H.
Miniaturized
Paper-Based Analytical Device for the Portable Analysis of Phyto-Cannabinoids
in Plant and Oral Fluids. Anal. Bioanal. Chem.
2024, 416, 255–264. 10.1007/s00216-023-05013-x.
Deenin W.; Wenninger N.; Schmid M. G.; Kalcher K.; Ortner A.; Chaiyo S.
Rapid Electrochemical
Lateral Flow Device for the Detection
of Δ9-Tetrahydrocannabinol. Anal. Chim.
Acta
2023, 1279, 341768. 10.1016/j.aca.2023.341768.
Pholsiri T.; Khamcharoen W.; Vimolmangkang S.; Siangproh W.; Chailapakul O.
Paper-Based
Electrochemical Sensor for Simultaneous
Detection of Salivary Δ9-Tetrahydrocannabinol and
Thiocyanate to Differentiate Illegal Cannabis Smokers. Sens. Actuators, B
2023, 383, 133571. 10.1016/j.snb.2023.133571.Risoluti R.; Gullifa G.; Battistini A.; Materazzi S.
MicroNIR/Chemometrics:
A New Analytical Platform for Fast and Accurate Detection of Δ9-Tetrahydrocannabinol
(THC) in Oral Fluids. Drug Alcohol Depend.
2019, 205, 107578. 10.1016/j.drugalcdep.2019.107578.
Chen Z.; de Boves Harrington P.; Griffin V.; Griffin T.
In Situ Determination
of Cannabidiol in Hemp Oil by Near-Infrared Spectroscopy. J. Nat. Prod.
2021, 84, 2851–2857. 10.1021/acs.jnatprod.1c00557.
Siudem P.; Wawer I.; Paradowska K.
Rapid Evaluation of Edible Hemp Oil
Quality Using NMR and FT-IR Spectroscopy. J.
Mol. Struct.
2019, 1177, 204–208. 10.1016/j.molstruc.2018.09.057.Geskovski N.; Stefkov G.; Gigopulu O.; Stefov S.; Huck C. W.; Makreski P.
Mid-Infrared Spectroscopy
as a Process Analytical Technology
Tool for Estimation of THC and CBD Content in Cannabis Flowers and
Extracts. Spectrochim. Acta, Part A
2021, 251, 119422. 10.1016/j.saa.2020.119422.Borille B. T.; Marcelo M. C. A.; Ortiz R. S.; Mariotti K. C.; Ferrão M. F.; Limberger R. P.
Near-Infrared Spectroscopy Combined
with Chemometrics
for Growth Stage Classification of Cannabis Cultivated in a Greenhouse
from Seized Seeds. Spectrochim. Acta, Part A
2017, 173, 318–323. 10.1016/j.saa.2016.09.040.Birenboim M.; Kengisbuch D.; Chalupowicz D.; Maurer D.; Barel S.; Chen Y.; Fallik E.; Paz-Kagan T.; Shimshoni J. A.
Use of
Near-Infrared Spectroscopy for the Classification
of Medicinal Cannabis Cultivars and the Prediction of Their Cannabinoid
and Terpene Contents. Phytochemistry
2022, 204, 113445. 10.1016/j.phytochem.2022.113445.
Cirrincione M.; Saladini B.; Brighenti V.; Salamone S.; Mandrioli R.; Pollastro F.; Pellati F.; Protti M.; Mercolini L.
Discriminating
Different Cannabis sativa L. Chemotypes Using Attenuated Total Reflectance-Infrared
(ATR-FTIR) Spectroscopy: A Proof of Concept. J. Pharm. Biomed. Anal.
2021, 204, 114270. 10.1016/j.jpba.2021.114270.
Monari A.; Foca G.; Ulrici A.; Zanfrognini B.; Brighenti V.; Verri P.; Pellati F.; Zanardi C.; Pigani L.
Electrochemical Sensors for Fast
Classification of
Different Cannabis sativa L. Samples According to Total Δ9-Tetrahydrocannabinol Content. Talanta
2025, 282, 126958. 10.1016/j.talanta.2024.126958.
Nimbkar S.; Leena M. M.; Moses J. A.; Anandharamakrishnan C.
Medium Chain
Triglycerides (MCT): State-of-the-Art on Chemistry, Synthesis, Health
Benefits, and Applications in Food Industry. Compr. Rev. Food Sci. Food Saf.
2022, 21, 843–867. 10.1111/1541-4337.12926.
Associated DataSupplementary Materials