Food Analytical Methods (2019) 12:2886–2894

https://doi.org/10.1007/s12161-019-01645-x

Simplified Quantification of Representative Bioactives in Food

Through TLC Image Analysis
Lujing Xu 1 & Tong Shu 2 & Songbai Liu 1,2

Received: 11 July 2019 / Accepted: 9 September 2019 / Published online: 15 October 2019
# Springer Science+Business Media, LLC, part of Springer Nature 2019

Abstract
In this work, simplified quantification of representative bioactives through thin-layer chromatography (TLC) image was
established employing state-of-the-art quantitative image analysis technology. A general protocol including developing system,
visualization condition, and image analysis procedure was developed for representative bioactives such as polyphenols, flavo-
noids, anthocyanins, phytosterols, and carotenoids. Applicability of this method including linearity, repeatability, and accuracy
was validated by comparison with the conventional UV-Vis spectrophotometry methods. Application in actual food samples
including carrot and green tea demonstrated this method was accurate and selective. The high-throughput potential of this method
was also demonstrated. This method is free from special instrument, is efficient, and would be a fantastic substitute for the
conventional UV-Vis spectrophotometry methods. This work has unveiled the power of quantitative image analysis in bioactive
analysis and would encourage further application in food industry.

Keywords Thin-layer chromatography . Image analysis . ImageJ . Bioactive . Quantitative analysis

Introduction 2012; Thilakarathna and Rupasinghe 2013; Tsuda 2012;

Demonty et al. 2009; Maiani et al. 2009). Analysis of bioac-
Recent years have witnessed booming of computer technolo- tives is fundamental in food studies and valuable to facilitate
gy such as deep learning for artificial intelligence (Chen and their application in food industry. Conventionally, UV-Vis
Lin 2014). Application of state-of-the-art computer technolo- spectrophotometry and high-performance liquid chromatogra-
gy revolutionalizes research practice in biology and chemistry. phy (HPLC) are employed to quantify the content of bioac-
For instance, computer-aided retrosynthesis equivalent to tives. Although these methods are highly sensitive and accu-
those produced by humans has been successfully achieved rate, special instruments are required for analysis. Especially,
with deep neural networks and symbolic artificial intelligence careful pretreatment of samples is needed for HPLC analysis
(Segler et al. 2018). Therefore, it is essential to employ appro- to eliminate tiny precipitates which is harmful for the instru-
priate computer technology in traditional chemical studies to ment system.
greatly improve efficiency. Thin-layer chromatography (TLC) is generally ap-
Natural bioactives including polyphenols, flavonoids, an- plied for monitoring progress of reaction and character-
thocyanins, phytosterols, and carotenoids in food have re- ization of bioactives owing to operational simplicity and
ceived much attention because of their beneficial health ef- effectiveness. Usually, analysis by TLC is qualitative or
fects (Murray et al. 2015; Yang et al. 2011; Peterson et al. semi-quantitative. Because of outstanding advantages of
TLC analysis including easy sample preparation, broad
sample applicability, and high throughput parallel run-
* Songbai Liu
ning, great efforts have been made to afford quantitative
[email protected]
TLC analysis. Currently, TLC analysis can be carried
1
Department of Food Science and Nutrition, Fuli Institute of Food out quantitatively (Wang et al. 2019); however, a special
Science, Zhejiang Key Laboratory for Agro-Food Processing, scanner is required to obtain UV-Vis spectroscopy of the
Zhejiang R & D Center for Food Technology and Equipment, sample. To fully liberate the power of TLC, we want to
Zhejiang University, 866 Yuhangtang Road, Hangzhou 310058,
employ state-of-the-art computer technology to renovate
China
2
traditional TLC analysis and render it free from special
Qinghai Food Inspection and Testing Institute, 12 Beidajie,
instruments (Fig. 1).
Xining 810000, China
Food Anal. Methods (2019) 12:2886–2894 2887

Fig. 1 Conceptual illustration of quantitative analysis of the bioactives through TLC image analysis

In traditional TLC analysis, the compounds on TLC are China). Ethyl acetate, trichloromethane, methanol, ethanol,
usually visualized by certain staining reagent after develop- and glacial acetic acid were obtained from Sinopharm
ment. Thus, the compounds are transformed into visible im- Chemical Reagent (Shanghai, China). Ultrapure water was
ages on TLC plates. If the images can be quantitatively rec- made by laboratory’s Milli-Q purification system (Millipore,
ognized, quantitative TLC analysis will be realized according- USA). UV-Vis spectrophotometer was purchased from
ly. Recently, there is considerable progress in computer image Shanghai Spectrum Instrument Company (SP-756, China).
processing and quantitative image analysis has been possible.
ImageJ is a readily available open image processing tool de- Preparation of Standard Solutions for Applicability
veloped by the National Institute of Health, which has been Validation
widely applied in medical imaging, biological science, engi-
neering, and environmental science (Baviskar 2011). Ethanol solutions of quercetin (15 mg/mL), (+)-catechin hy-
Quantification of cellular and subcellular components and es- drate (5 mg/mL), rutin (1.6 mg/mL), cyanidin chloride (2 mg/
timation of DNA and protein concentrations have been suc- mL), cyanidin-3-O-glucoside (2 mg/mL), and oleanolic acid
cessfully realized by quantitative image analysis with ImageJ (16 mg/mL), hexane/acetone solution of β-carotene (1 mg/
(Müller et al. 2016; Zhang et al. 2013). mL, hexane/acetone 3:2), and ethyl acetate solution of β-si-
Hence, quantitative TLC analysis of representative natural tosterol (2 mg/mL) were prepared for checking linearity of the
bioactives of polyphenols, flavonoids, anthocyanins, phytos- TLC method.
terols, and carotenoids was attempted through quantitative Ethanol solutions of quercetin (4.2, 0.07 mg/mL), (+)-cat-
image analysis with ImageJ in this study. Visualization of echin hydrate (1.8, 0.0045 mg/mL), rutin (0.9, 0.045 mg/mL),
images of the bioactives was optimized and quantification of cyanidin chloride (0.5, 0.015 mg/mL), cyanidin-3-O-gluco-
the images was established by ImageJ. Then, the linearity and side (1.0, 0.008 mg/mL), and oleanolic acid (4, 0.08 mg/
repeatability of this method were checked, and the accuracy mL), hexane/acetone solution of β-carotene (0.6, 0.0024
was also confirmed comparing with conventional UV-Vis mg/mL, hexane/acetone 3:2 v/v), and ethyl acetate solution
spectrophotometry. This developed quantitative analysis of of β-sitosterol (0.9, 4.5 mg/mL) were used as analytes mea-
bioactives through TLC image analysis free from special in- sured by the TLC method and spectrophotometric method
struments was accurate, sensitive, and high-throughput with respectively to validate accuracy and repeatability of our
broad scope of substrate. work. Ethanol solutions of quercetin (8.4 mg/mL), (+)-cate-
chin hydrate (3.6 mg/mL), rutin (1.8 mg/mL), cyanidin chlo-
ride (1.0 mg/mL), cyanidin-3-O-glucoside (2 mg/mL), and
Materials and methods oleanolic acid (8 mg/mL), hexane/acetone solution of β-caro-
tene (1.0 mg/mL, hexane/acetone 3:2), and ethyl acetate solu-
Reagents tion of β-sitosterol (2 mg/mL) were used as internal standards
for the TLC method.
The standard substances, rutin, oleanolic acid, cyanidin chlo-
ride, cyanidin-3-O-glucoside, β-carotene, and β-sitosterol Visualization of TLC and Image Acquisition
were obtained from Yuanye Biotechnology (Shanghai,
China), (+)-catechin hydrate was obtained from Sigma- TLC experiments were performed employing a developing
Aldrich (Shanghai, China), and Quercetin dehydrate and p- chamber (100 × 100 mm). Standard solutions (0.5 μL) were
anisaldehyde were obtained from Aladdin (Shanghai, spotted onto silica gel plates (HSGF254, 200 × 100 mm,
2888 Food Anal. Methods (2019) 12:2886–2894

Institute of Chemical Industry of Yantai, Yantai, China) with calibration curve, which was measured at 520 nm (Sripakdee
quantitative capillaries (0.5 μL). For the group of nonpolar et al. 2017). Ethyl acetate solution of β-carotene (0.0015–
bioactives, the developing solvent system is petroleum ether: 0.0035 mg/mL) was read at 450 nm against a blank
ethyl acetate 5:1 v/v, with total development length of 40 mm, (Waramboi et al. 2013). Ethanol solutions of quercetin diluted
while for the group of polar bioactives, development of the with 70% ethanol (0.02–0.12 mg/mL) were measured at 374
plates to a distance of 20 mm was firstly performed with ethyl nm.
acetate: ethanol absolute: water: acetic acid 5:2.85:0.15:0.12 Ethanol solution of oleanolic acid (0.12–1.56 mg/mL) were
v/v/v/v, then dried, and re-developed to the length of 40 mm prepared and 0.5 mL of each concentration was added into a
with trichloromethane: ethyl acetate: formic acid 1:1:0.03 v/v/ 10-mL centrifuge tube, followed by addition of 0.5 mL of 8%
v. The polarity range of the polar bioactives is broader than vanillin in ethanol and 5 mL of 72% H2SO4 in water. The
that of the nonpolar bioactives. As a result, one mobile pro- mixture of oleanolic acid was set in a thermostat at 60 °C for
vided satisfactory separation for the non-polar bioactives with 20 min and at 0 °C for 5 min and then measured at a wave-
a single 40-mm development. However, acceptable separation length of 544 nm (Helaly et al. 2001; Dini et al. 2009).
of the polar bioactives was not achieved with only one mobile Ethyl acetate solution of β-sitosterol (10 mg/mL) was pre-
phase. Application of the two mobile phases and double de- pared and 2–6 mL standard solutions were transferred into
velopment for polar samples was required to afford satisfac- 10-mL volumetric flasks with addition of 2 mL LB reagent
tory separation. (LB, Liebermann-Burchard reagent: a volume of 50 mL of
The spots were visualized using a solution of p- acetic anhydride was transferred to an amber glass vial and
anisaldehyde/acetic acid/95% ethanol/sulfuric acid 9.2 : 3.75 kept in an ice bath. After 30 min, 5 mL of sulfuric acid was
: 338 : 12.5 v/v/v/v followed by heating on a plate heater for added carefully to acetic anhydride). The volume was adjusted
10~20 s at 250 °C. A common computer scanner (LaserJet with ethyl acetate. The absorptions were measured at 625 nm
M1005, Hewlett-Packard, Idaho, USA) was applied to scan in 5 min.
the obverse and reverse of the developed plates after visuali- To (+)-catechin solution (0.2–0.6 mL, 0.1 mg/mL) in 10
zation. The resolution was 300 dpi and all the acquired images mL of volumetric flasks was added 0.5 mL of Folin-Ciocalteu
were saved as TIFF format. reagent and 1.7 mL of 20% Na2CO3 aqueous solution. The
volumetric flasks were then topped up to the volume with
Quantitative Recognition of TLC Images by ImageJ distilled water. After a 60-min reaction, absorption was mea-
sured at 760 nm (Araújo et al. 2013).
Quantitative analysis of TLC images was performed by A total of 200.0 mg/L of the rutin standard solution was
ImageJ program (1.52a edition). The image was loaded in prepared by dissolving rutin in 70% ethanol. Four milliliters of
ImageJ and analyzed with grayscale value. Optimization of each sample was pipetted into a 10-mL volumetric flask. The
the processing procedure was described as follows. Contrast solution was treated with 0.40 mL of the 5% NaNO2 solution for
of the area to be analyzed was adjusted applying the default 6 min and evenly mixed, followed by addition of 0.4 mL of the
method after changed into 8-bit type. Then, the salt and pepper 10% Al(NO3)3 solution. After 6 min, 4 mL of the 4% NaOH
noise was removed by a median filter, replacing each pixel solution was added. The mixture was diluted to the volume with
with the median value in its 3*3 neighborhood. Outliers in the double distilled water, and allowed to stand for 15 min before
image was removed by setting appropriate parameters in the analyzing against the blank solution at 500 nm (Marinova et al.
dialog box. Edges of the image was sharpened and then the 2005). The standard curve was linear over a concentration range
whole image was smoothed. The background of the image of 0.002–0.007 mg/mL.
was subtracted using an appropriate rolling ball radius and
sliding paraboloid. Finally, the image was selected in a rect- Analysis of Food Samples by the Quantitative TLC
angular zone to generate its profile plots and the area of each Image Analysis Method
peak was measured by the Wand tool.
This method was applied to actual food samples. The contents of
Applicability Validation of the Quantitative TLC Image carotenoids in carrot (Xuanma Guoshu Gangwan Store,
Analysis Method by Conventional Hangzhou, China) and catechins in green tea (West Lake
Spectrophotometric Method Longjing Tea, Hangzhou Shifeng Tea Co. Ltd., Hangzhou,
China) were determined by this TLC method and compared with
The determinations by the conventional spectrophotometric the conventional method. The carrot was sliced into thin layers
method were performed in triplicate and calculated by the and vacuum freeze-dried. Three grams of the dried carrot was
calibration curves. Solutions of cyanidin chloride (0.008– crushed and extracted with 15 mL of hexane at 60 °C for 2 h.
0.016 mg/mL) and cyanidin-3-O-glucoside (0.0025–0.015 Then, the content of carotenoids in the extract was determined
mg/mL) diluted with 1% HCl in methanol were used for each employing the TLC image analysis method and also determined
Food Anal. Methods (2019) 12:2886–2894 2889

by the conventional UV-Vis spectrophotometry as described systems, the bioactives could be properly spread on TLC with
above with 1 mg/mL β-carotene as the standard. Three grams good separation of different bioactives.
of the green tea was crushed and extracted with 12 mL of 95% The rationale behind choosing double sample development
ethanol at 60 °C for 2 h. The content of catechins in the extract distance of 20 mm and 40 mm was described as follows. The
was determined employing the TLC image analysis method and mobile phase for the first development was more polar than
also determined by the conventional UV-Vis spectrophotometry that of the second development mobile phase. The first devel-
as described above with 3 mg/mL (+)-catechin as the standard. opment distance was 20 mm and the second was 40 mm. The
first 20-mm development provided a good separation for the
polar bioactives with glycoside and brought the polar bioac-
tives without glycoside in the middle part of the whole TLC
Result and Discussion plates, and left the rest 20 mm for the second 40-mm devel-
opment. The second 40-mm development by less polar mobile
Establishment of Quantitative Analysis phase afforded a good separation for the polar bioactives with-
of Representative Bioactives Through TLC Image out glycoside in the rest 20 mm of the TLC plates, and also
Analysis had little interference with the separation of the polar bioac-
tives with glycoside provided by the first development since
During TLC image analysis, the information of the com- the mobile phase applied in the second development was less
pounds is eventually transformed into visible images on polar than the first one.
TLC plates. In this study, to demonstrate the usefulness of After development on TLC, visualization of the bio-
TLC analysis, we want to develop a general protocol to ana- actives was exercised. It is well known that
lyze representative bioactives including polyphenols, flavo- anthocyanidin (cyanidin chloride) and anthocyanin
noids, anthocyanins, phytosterols, and carotenoids. (cyanidin-3-O-glucoside) have distinct visible red color
Therefore, unified conditions should be developed for analy- under acidic condition. As a result, they could be visu-
sis. Generally, the compound to test has to be spotted on TLC alized by any acidic reagent. Carotenoid (β–carotene)
plate in the same volume first and then developed in a suitable has distinct visible color and so could be analyzed with-
solvent system, finally visualized by certain staining reagent out further visualization. For other bioactives without
under heating to afford the visible images on TLC plates. obvious visible color, a general staining reagent was
Once the images are obtained, appropriate image analysis desirable to visualize them. Intensive attempts revealed
method has to be established by ImageJ. As a result, every that the anisaldehyde-based staining reagent worked
step has to be carefully designed and optimized to realize very well for all the bioactives in the study, which pro-
quantitative analysis of bioactives through TLC image vided clear visible images (Fig. 2). Examination images
analysis. on both sides of TLC disclosed that the images on the
The readily available β-sitosterol, β-carotene, (+)-cate- reverse side of TLC were clearer than those on the
chin, oleanolic acid, quercetin, rutin, cyanidin chloride, and obverse side.
cyanidin-3-O-glucoside were selected as typical phytosterol, The images of the bioactives on TLC plates were
carotenoid, polyphenol, triterpenoid, flavonoid, flavonoid gly- collected employing a common computer scanner.
coside, anthocyanidin, and anthocyanin. Owing to the broad Preliminary examination of the digitized images of the
range of polarity of the bioactives, it was difficult to apply one bioactives of different concentrations disclosed that the
developing solvent system for all the bioactives. As a result, intensity of the images was roughly proportional to the
they were divided into two groups, nonpolar (with lower po- concentration. Then ImageJ was applied to further quan-
larity) and polar (with higher polarity) groups to treat. For the tify the images. The optimized image analysis protocol
nonpolar group (β-sitosterol, β-carotene), petroleum ether: by ImageJ was demonstrated by employment of β-sitos-
ethyl acetate was applied as the developing solvent system. terol as an example (Fig. 3). Both sides (obverse and
In contrast, ethyl acetate: ethanol: water: acetic acid and reverse) of TLC images of β-sitosterol were analyzed.
trichloromethane: ethyl acetate: formate were chosen as the Before staining, there was no visible color for both
optimized solvent system for the polar group (oleanolic acid, sides. After staining, visible blue color of β-sitosterol
(+)-catechin, quercetin, rutin, cyanidin chloride, cyanidin-3- was observed on both sides. The area with β-sitosterol
O-glucoside) after experimentations. Owing to the broad signals was cropped and transformed to 8-bit grayscale
range of polarity of the bioactives, it was difficult to apply type. Optimization of the processing procedure is de-
only one developing solvent system for all the bioactives. scribed in “Quantitative Recognition of TLC Images
As a result, two different groups of the developing systems by ImageJ.” The contrast of the image was adjusted at
were chosen, one for nonpolar bioactives and the other for first. Usually, there were tiny contaminated spots named
polar bioactives. Employing the optimized developing salt and pepper noise on TLC images, which were
2890 Food Anal. Methods (2019) 12:2886–2894

Fig. 2 Visualization of the representative bioactives on TLC. Visible light (1–8): before staining; visible light (9–10): after staining but no heating;
coloration (1–8): after staining and heating; obverse: obverse side of TLC plate; reverse: reverse side of TLC plate

required to be removed for quantitative analysis. The not stable directly after smooth operation. As a result,
noise and abnormal outliers were then eliminated by the baseline was corrected by the operation of subtract
the tools provided by ImageJ. During the study, it was background in ImageJ. Finally, the amount of the bio-
found that sharpening the edges of the image afforded active was quantitatively transformed into the regular
better recognition results, which was performed by un- spectrum through image analysis. The intensity of each
sharp mask operation in ImageJ. Then the whole image peak in the spectrum corresponded to the amount of the
was smoothed and the image could be transformed to a respective bioactive. The images on the reverse side of
spectrum. As demonstrated in Fig. 3, the baseline was TLC provided better results and accordingly the reverse

Fig. 3 Quantitative image analysis employing ImageJ exemplified by β-sitosterol

Food Anal. Methods (2019) 12:2886–2894 2891

side images were chosen in further analysis. Therefore, Beer law (Butnariu and Coradini 2012; Minnaar et al.
the approaches for quantitative analysis of the bioactives 2018; Bunea et al. 2008). β-carotene (carotenoid) and quer-
through TLC image analysis were established. cetin (flavonoid) were directly measured at certain wave-
lengths. Cyanidin chloride (anthocyanidin) and cyaniding-3-
Applicability Validation of Quantitative TLC Analysis O-glucoside (anthocyanin) were treated with acidic reagents
of Representative Bioactives during spectrophotometry. In contrast, oleanolic acid
(triterpenoid) was treated with vanillin-based chromogenic
Upon establishment of the bioactive quantitative analysis by reagent, β-sitosterol (phytosterol) with Liebermann-Burchard
TLC image analysis, the applicability of this method including reagent, (+)-catechin (polyphenol) with Folin-Ciocalteu re-
linearity, repeatability, and accuracy was investigated. The agent, and rutin (flavonoid glycoside) with NaNO2 and
linearity range is the concentration range during which the Al(NO3)3 chromogenic reagent during spectrophotometry.
result is directly proportional to the amount of the bioactive. As shown in Fig. 4, the results determined by this TLC
To probe linearity range, stock solutions of the eight represen- image analysis method matched well with those by the con-
tative bioactives mentioned above were subsequently diluted ventional methods. Then, this method was applied in actual
into series of appropriate concentration (Table 1). Each series food samples. The contents of carotenoids in carrot extract
of stock solutions was performed triple times. It was unveiled and catechins in green tea extract were determined by the
that R2 (coefficient of determination) values of regression TLC image analysis (Fig. 5a). It was revealed that the con-
analysis for all the bioactives were no less than 0.95, which tent of carotenoids in the carrot extract determined by this
indicated this method of TLC image quantitative recognition TLC method was 3.1 mg/g which agreed very well with the
was generally of good linearity. In addition, the concentration content determined by the conventional method, 3.3 mg/g.
ranges of linearity for the bioactives were around 0.5 to sev- The content of catechins in the green tea extract was 62.0
eral mg/mL, which was convenient for analysis of the natu- mg/g obtained by this TLC method and the content deter-
rally derived bioactives because the concentrations of bioac- mined by conventional method was 77.6 mg/g.
tives after extraction and concentration usually fell into this Determination of the conventional Folin-Ciocalteu method
range. Only 0.5 μL of the bioactives was applied, and so, included all the phenolic compounds except for catechins.
around 0.25 μg of the bioactives was sufficient for the quan- As a result, this TLC method could selectively analyze spe-
titative analysis, suggesting good sensitivity of this method. cific compounds. Thus, quantitative TLC analysis of repre-
Then repeatability of this method was explored by deter- sentative bioactives was validated.
mination of an arbitrarily chosen concentration during linear- As mentioned above, the conventional spectropho-
ity range for the eight bioactives in triplicate. The results dem- tometry methods required the spectrophotometer instru-
onstrated that coefficient of variation (CV) of the determina- ment and various chromogenic reagents to determine a
tions for β-sitosterol, β-carotene, (+)-catechin, quercetin, different class of bioactives. Compared with the conven-
oleanolic acid, rutin, cyanidin chloride, and cyanidin-3-O-glu- tional methods, there were obvious advantages for this
coside were 0.99%, 5.41%, 2.18%, 1.83%, 1.58%, 3.16%, quantitative TLC image analysis method. A unified pro-
2.67%, and 4.78%, respectively, illustrating good repeatability tocol including visualization and image analysis was
of TLC image quantitative analysis method. applied and free from special instrument in this method.
To check the accuracy of this method, determination re- The conventional tedious operation was greatly simpli-
sults of the chosen concentrations of the eight bioactives fied. Furthermore, it was difficult for conventional
were compared with measurement by the conventional spec- methods to analyze different class of bioactives simulta-
trophotometric methods. Conventionally, UV-Vis spectropho- neously because of conflicts of these conditions.
tometry is employed to quantify the content of bioactives Nevertheless, owing to the separation ability of TLC
after appropriate chromogenic reaction through Lambert- and unified protocol, different class of bioactives could

Table 1 Linearity range of quantitative TLC image analysis of the representative bioactives

Bioactives β- β- (+)- Quercetin Oleanolic acid Rutin Cyanidin chloride Cyanidin-3-O-

carotene sitosterol catechin glucoside

Concentration (mg/mL) 0.3–1.0 0.4–2.0 0.7–5.0 2.0–15.0 1.0–16.0 0.3-1.6 0.5–2.0 0.5–2.0
R2 (1st) 0.96 0.98 0.99 0.99 0.99 0.99 0.99 0.95
R2 (2nd) 0.98 0.99 0.99 0.99 0.99 0.99 0.99 0.95
R2 (3rd) 0.96 0.98 0.96 0.99 0.97 0.99 0.99 0.98
2892 Food Anal. Methods (2019) 12:2886–2894

Fig. 4 Applicability validation of

the TLC image analysis method
compared with the conventional
UV-Vis spectrophotomery
methods

Fig. 5 a TLC quantitative image

analysis of carotenoids in carrot
extract and catechins in green tea
extract. b Demonstration of high-
throughput potential of the TLC
image analysis method
Food Anal. Methods (2019) 12:2886–2894 2893

be analyzed at the same time and high specificity could Conclusion

be achieved by this method. Therefore, this quantitative
TLC image analysis method would be a fantastic sub- In summary, a simplified quantitative analysis of representa-
stitute for the conventional UV-Vis spectrophotometry tive bioactives through TLC image analysis has been
methods. established. Applicability of this method including linearity,
repeatability, and accuracy was validated. Furthermore, appli-
cability in actual food samples was proved. The high-
High-Throughput Analysis by Quantitative TLC Image throughput potential of this method was also demonstrated.
Analysis This method is free from special instrument and has greatly
liberated the power of TLC. Thus, this method is promising
Apparently, this quantitative TLC image analysis method is for large-scale initial screening of the bioactives to seek out
potentially high-throughput because multiple samples can be qualified candidates for further detailed study. This work
parallel analyzed and different bioactives within the SAME would encourage application of quantitative image analysis
sample can be resolved simultaneously. To demonstrate the in chemical studies.
high-throughput potential, five polar bioactives, (+)-catechin,
quercetin, oleanolic acid, rutin, cyanidin chloride, and a mix- Funding information This study was funded by the National Key R&D
Program of China (2017YFF0207800, 2016YFD0400805), the Zhejiang
ture of the five bioactives were applied in the TLC analysis. It
Public Welfare Technology Research Program (LGN18C200009), the
was exhibited that the five bioactives could be parallel mea- Qinghai Science and Technology Program (2017-ZJ-Y06, 2016-NK-
sured at different lanes of the TLC (Fig. 5b). Especially, the C22, 2015-NK-502), and the Foundation of Fuli Institute of Food
five bioactives in the mixture were nicely resolved and ana- Science at Zhejiang University, Zhejiang Science and Technology
Program (2017C26004).
lyzed in one lane. Although there was some overlap of the
images of cyanidin chloride (anthocyanidin) and rutin (flavo-
noid glycoside), they could be separately recognized at differ- Compliance with Ethical Standards
ent stages of visualization. Directly after treatment by the
Conflict of Interest Lujing Xu declares that she has no conflict of inter-
acidic staining reagent, cyanidin chloride presented distinct est. Tong Shu declares that he has no conflict of interest. Songbai Liu
red image and there was no visible image of rutin, and so declares that he has no conflict of interest.
cyanidin chloride was imaged at this stage. Upon heating,
the color of cyanidin faded and the image of rutin appeared. Ethical Approval This article does not contain any studies with human
As a result, the image of rutin was collected after heating. In participants or animals performed by any of the authors.
the mixture, half amounts of the bioactives were applied cor-
Informed Consent Not applicable.
responding to those of the samples of the single bioactives. As
illustrated in Fig. 5, the obtained areas by quantitative recog-
nition agreed well with each other, suggesting that the quan-
References
titative TLC image analysis method was suitable for high-
throughput analysis. Araújo LBDC, Silva SL, Galvão MAM, Ferreira MRA, Araújo EL,
Compared with high-performance liquid chromatogra- Randau KP, Soares LAL (2013) Total phytosterol content in drug
phy (HPLC) analysis, this method cannot compete with materials and extracts from roots of Acanthospermum hispidum by
its high resolving ability of the bioactives. However, pre- UV-VIS spectrophotometry. Rev Bras 23(5):736–742
Baviskar SN (2011) A Quick & Automated Method for Measuring Cell
liminary screening of the bioactives usually does not re- Area Using ImageJ. Am Biol Teach 73(9):554–556
quire great separation power and the high-throughput Bunea A, Andjelkovic M, Socaciu C, Bobis O, Neacsu M, Verhé R,
method is more desirable to seek out qualified candidates Camp JV (2008) Total and individual carotenoids and phenolic acids
for detailed study. The HPLC analysis is sequentially op- content in fresh, refrigerated and processed spinach (Spinacia
oleracea L.). Food Chem 108(2):649–656
erated and multi-channel parallel analysis is almost impos- Butnariu M, Coradini CZ (2012) Evaluation of Biologically Active
sible. As demonstrated above, multi-channel analysis is Compounds from Calendula officinalis Flowers using
readily available for this method, which is very useful Spectrophotometry. Chem Cent J. https://doi.org/10.1186/-153X-6-
for instant screening. In addition, the instrument and oper- 35
Chen X, Lin X (2014) Big Data Deep Learning: Challenges and
ational costs of HPLC are demanding and laborious pre- Perspectives. 2:514–525 IEEE Access
treatments are generally required. A typical HPLC analysis Demonty I, Ras RT, van der Knaap HCM, Duchateau GS, Meijer L, Zock
takes tens of minutes and the cost for this TLC analysis is PL, Geleijnse JM, Trautwein EA (2009) Continuous Dose-
usually several minutes. This method is free from special Response Relationship of the LDL-Cholesterol-Lowering Effect of
Phytosterol Intake. J Nutr 139(2):271–284
instrument and the power of TLC has been greatly liber- Dini I, Tenore GC, Dini A (2009) Saponins in Ipomoea batatas tubers:
ated. Thus, this method is more suitable than HPLC for Isolation, characterization, quantification and antioxidant properties.
large-scale initial screening of the bioactives. Food Chem 113(2):411–419
2894 Food Anal. Methods (2019) 12:2886–2894

Helaly FM, Soliman HSM, Soheir AD, Ahmed AA (2001) Controlled Segler MHS, Preuss M, Waller MP (2018) Planning chemical syntheses
release of migration of molluscicidal saponin from different types of with deep neural networks and symbolic AI. Nature 555(7698):604–
polymers containing Calendula officinalis. Adv Polym Technol 610
20(4):305–311 Sripakdee T, Mahachai R, Chanthai S (2017) Direct analysis of
Maiani G, Castón MJ, Catasta G, Toti E, Cambrodón IG, Bysted A, anthocyanins-rich Mao fruit juice using sample dilution method
Granado-Lorencio F, Olmedilla-Alonso B, Knuthsen P, Valoti M, based on chromophores/fluorophores of both cyanidin-3-glucoside
Böhm V, Mayer-Miebach E, Behsnilian D, Schlemmer U (2009) and pelargonidin-3-glucoside. Int Food Res J 24(1):215–222
Carotenoids: Actual knowledge on food sources, intakes, stability Thilakarathna SH, Rupasinghe HPV (2013) Flavonoid Bioavailability
and bioavailability and their protective role in humans. Mol Nutr and Attempts for Bioavailability Enhancement. Nutrients 5(9):
Food Res 53(S2):194–218 3367–3387
Marinova D, Ribarova F, Atanassova M (2005) Total phenolics and total Tsuda T (2012) Dietary anthocyanin-rich plants: Biochemical basis and
flavonoids in Bulgarian fruits and vegetables. J Univ Chem Technol recent progress in health benefits studies. Mol Nutr Food Res 56(1):
Metall 40(40):255–260 159–170
Minnaar P, Nyobo L, Jolly N, Ntushelo N, Meiring S (2018) Wang P, Chen Y, Xu X, Hellmann B, Huang C, Bai Y, Jin Z (2019)
Anthocyanins and polyphenols in Cabernet Franc wines produced HPTLC Screening of Folic Acid in Food: In Situ Derivatization with
with Saccharomyces cerevisiae and Torulaspora delbrueckii yeast Ozone-Induced Fluorescence. Food Anal Methods 12(2):431–439
strains: Spectrophotometric analysis and effect on selected sensory Waramboi JG, Gidley MJ, Sopade PA (2013) Carotenoid contents of
attributes. Food Chem 268:287–291 extruded and non-extruded sweetpotato flours from Papua New
Müller M, Öller VM, Hennig S, Hübner W, Huser T (2016) Open-source Guinea and Australia. Food Chem 141(3):1740–1746
image reconstruction of super-resolution structured illumination mi- Yang C, Wang H, Li G, Yang Z, Guan F, Jin H (2011) Cancer prevention
croscopy data in ImageJ. Nat Commun. https://doi.org/10.1038/ by tea: Evidence from laboratory studies. Pharmacol Res 64(2):113–
ncomms10980 122
Zhang Y, Yang P, Sun T, Li D, Xu X, Rui Y, Li C, Chong M, Ibrahim T,
Murray M, Walchuk C, Suh M, Jones PJ (2015) Green tea catechins and
Mercatali L, Amadori D, Lu X, Xie D, Li Q, Wang X (2013) miR-
cardiovascular disease risk factors: Should a health claim be made
126 and miR-126* repress recruitment of mesenchymal stem cells
by the United States Food and Drug Administration. Trends Food
and inflammatory monocytes to inhibit breast cancer metastasis. Nat
Sci Technol 41(2):188–197
Cell Biol 15(3):284–294
Peterson JJ, Dwyer JT, Jacques PF, McCullough ML (2012) Associations
between flavonoids and cardiovascular disease incidence or mortal-
ity in European and US populations. Nutr Rev 70(9):491–508 Publisher’s Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.