Food Research International 64 (2014) 363–370

Contents lists available at ScienceDirect

Food Research International

journal homepage: www.elsevier.com/locate/foodres

Wheat flour granulometry determines colour perception

Alyssa Hidalgo a, Lorenzo Fongaro a,1, Andrea Brandolini b,⁎
a
Department of Food, Environmental and Nutritional Sciences (DeFENS), Università degli Studi di Milano, via Celoria 2, 20133 Milano, Italy
b
Consiglio per la Ricerca e la Sperimentazione in Agricoltura - Unità di Ricerca per la Selezione dei Cereali e la Valorizzazione delle Varietà Vegetali (CRA-SCV),
Via Forlani 3, 26866 S. Angelo Lodigiano, LO, Italy

a r t i c l e i n f o a b s t r a c t

Article history: Einkorn, durum wheat and Kamut® are rich of carotenoids, antioxidants with beneficial effects on human health.
Received 24 April 2014 In the present study the effect of flour granule size on carotenoid content and colour was assessed; furthermore,
Received in revised form 11 June 2014 the suitability of two colorimeters (Minolta Chroma meter CR-210 and Minolta Chroma meter II Reflectance), a
Accepted 17 June 2014
spectrophotometer (Jasco V-650 with integrating sphere) and image analysis to define colour and determine ca-
Available online 7 July 2014
rotenoid concentration in wheat flours of different granulometry was tested. Carotenoid content did not vary
Keywords:
across flours of diverse size, except in the finest einkorn fraction (b80 μm), which had lower concentration.
Carotenoids For all instruments colour coordinate L* decreased and b* increased with flour size growth, while a* varied
Durum wheat with the different devices. A Principal Components Analysis (PCA), performed considering carotenoid content
Flour size and all colorimetric indices, distinguished einkorn from durum and Kamut®, and divided samples of different
Einkorn granulometry; a similar result was achieved by a PCA performed on the absorbance spectra from the integrating
Image analysis sphere. The PCA on image texture data classified the samples following flour size. In conclusion, flour colour is
Kamut® determined not only by carotenoid content but also by flour particle size: therefore, direct colour measurement
seems not suitable to predict flour carotenoid content.
© 2014 Elsevier Ltd. All rights reserved.

1. Introduction
Carotenoids are lipid-soluble pigments that play a relevant role in
many essential functions of plants and animals. In humans they are
the precursors of vitamin A and exert a positive function on health,
because their antioxidant activity protects cells and tissues from free
radicals and oxygen ion damages (Adom & Liu, 2002). Carotenoids are
scarce in bread wheat (Triticum aestivum L. subsp. aestivum), but are
more abundant in durum wheat (Triticum turgidum L. subsp. durum
Desf.) and Kamut® (T. turgidum L. ssp. turanicum Jakub.), ranging from
1.5 to 4.8 mg/kg (Hidalgo, Brandolini, Pompei, & Piscozzi, 2006;
Panfili, Fratianni, & Irano, 2004), and in einkorn (Triticum monococcum
L. subsp. monococcum), where they vary between 5.3 and
13.6 mg/kg dm (Abdel-Aal et al., 2002; Brandolini, Hidalgo, &
Moscaritolo, 2008; Hidalgo et al., 2006). The yellow colour
imparted by carotenoids (mainly lutein and its fatty acid esters)
has become a very important quality trait for pasta and other food
products (Blanco et al., 2011).
In wheat flour, carotenoid evaluation is analytically performed by
HPLC (Fratianni, Irano, Panfili, & Acquistucci, 2005; Leenhardt et al.,
⁎ Corresponding author. Tel.: +39 0371 211261.
E-mail addresses: alyssa.hidalgovidal@unimi.it (A. Hidalgo),
Lorenzo.Fongaro@ec.europa.eu (L. Fongaro), andrea.brandolini@entecra.it (A. Brandolini).
1
Present address: European Commission Joint Research Centre, Institute for
Transuranium Elements JRC.02 Hot Cells, Karlsruhe, Germany.
2006), a procedure that is very precise but also time-consuming and ex-
pensive. Easier, quicker and cheaper alternatives commonly adopted
are the use of visible/near-infrared reflectance spectrophotometry on
water-saturated butanol extracts (AACC International, 2012) or the
direct measurement of flour colour by reflectance spectrophotome-
try (Oliver, Blakeney, & Allen, 1992). The direct flour analysis is
based on the CIE colorimetric space (McCaig, 2002; Posner, 2009);
the colour is classified in three dimensions: L*, which measures
brightness (0 = black and 100 = bright), a*, where positive a* indi-
cates redness and negative a* indicates greenness, and b*, where
positive b* indicates yellowness and negative b* indicates blueness
(CIE Commission Internationale de l'Eclairage, 2004). CIE flour col-
our is determined mainly by a combination of brightness and
yellowness: brightness is influenced by bran content, while
yellowness is affected by carotenoid content of the endosperm
(Hidalgo & Brandolini, 2008a; Oliver et al., 1992). Variation in L* af-
fects the measurement of b*, and can lead to errors in estimating ca-
rotenoid content (Mares & Campbell, 2001).
Wheat flour carotenoid content and colour are influenced by inher-
ent genotypic characteristics (Brandolini et al., 2008), environmental
conditions (Hidalgo, Brandolini, & Ratti, 2009), stresses during grain
production (Fratianni, Giuzio, Di Criscio, Zina, & Panfili, 2013; Hidalgo
et al., 2009; Lachman, Hejtmánková, & Kotíková, 2013), milling
(Posner, 2009) and storage conditions (Hidalgo & Brandolini, 2008b).
In fact, milling procedures and extraction rates have a strong bearing
on several components, such as ash, protein, pigments and damaged

http://dx.doi.org/10.1016/j.foodres.2014.06.050
0963-9969/© 2014 Elsevier Ltd. All rights reserved.
364 A. Hidalgo et al. / Food Research International 64 (2014) 363–370
starch content (Posner, 2009) which, in turn, influence colour. Addition-
ally, flour particle size, related to differences in wheat species, grain
hardness and moisture content of the kernels at milling (Posner,
2009), is another possible source of variation (Symons & Dexter, 1991).
Several authors have tested different reflectance spectropho-
tometers and colorimeters for colour determination in wheat
flour (e.g. Black & Panozzo, 2004; Dowell et al., 2006; Mares &
Campbell, 2001; McCaig, 2002; Oliver et al., 1992), concluding
that the instruments with visible wavelength sensors had good po-
tential for predicting flour colour values with accuracy.
The aims of this research were a) to assess the influence of wheat
flour granule size on yellow pigment content and colour and b) to
evaluate the suitability of different instruments (two Minolta color-
imeters, Chroma meter CR-210 and Chroma meter II Reflectance, and
one Jasco V-650 spectrophotometer with integrating sphere) and of
image analysis to define colour and to measure the carotenoid
concentration in different wheat flours.
2. Materials and methods
2.1. Materials
The evaluation of protein, ash, pigment content and colour as a
function of flour granulometry was performed on seven samples: two
T. monococcum ssp. monococcum, the free-threshing line SAL98-8-3-2
(2011 harvest) and the hulled cultivar Monlis (three samples from
plots fertilized with different nitrogen levels: 0, 40, and 80 kg/ha of N;
2012 harvest), one T. turgidum ssp. turanicum (Kamut®; 2012 harvest)
and one T. turgidum ssp. durum (durum wheat) cv Saragolla (2011
and 2012 harvests). The accessions chosen are characterized by
medium-to-high carotenoid content (Hidalgo et al., 2006). All acces-
sions were cropped in Sant'Angelo Lodigiano (Italy), in the centre of
the Po plain agricultural belt, following standard cultural practices
(Castagna, Borghi, Di Fonzo, Heun, & Salamini, 1995). After harvesting,
the kernels were stored at 5 °C. Just before milling, the seeds of einkorn
cv Monlis were de-hulled with an Otake FC4S thresher (Satake, Japan);
dehulling was not required for the other accessions, all free-threshing.
Refined flours were obtained using a Bona-GBR lab mill (Bona,
Monza, Italy), that separates refined flour from bran and shorts. Flours
of different granulometry were graded with an automatic lab sifter
(Buhler Plain sifter, Buhler, Switzerland): about 800 g of flour was
screened for 5 min through a set of different sifters with increasingly
smaller mesh sizes (method 66–20, AACC American Association of
Cereal Chemists, 1995). The flour samples were split into five different
fractions: ≥ 200 μm, b 200 and ≥ 160 μm (160–200), b160 and
≥120 μm (120–160), b 120 and ≥80 μm (80–120), b 80 μm and stored
under vacuum at −20 °C until analysis.
2.2. Experimental
The dry matter of the different samples was measured as described
in method 44-15A (AACC American Association of Cereal Chemists,
1995), ash content was assessed with AACC method 08-03 (AACC
American Association of Cereal Chemists, 1995), and protein content
(N × 5.7) was determined following method 46-10 (AACC American As-
sociation of Cereal Chemists, 1995). The extraction of total carotenoids
was performed with saturated butanol, following AACC method 14-
60.01 (AACC International, 2012); the extracted pigments were mea-
sured at 450 nm with a double-beam V650 spectrophotometre (Jasco,
Japan). Total carotenoid content was then computed considering a cali-
bration curve prepared with six different concentrations (from 0.25 to
5.00 mg/L) of lutein standard stock solution (Sigma-Aldrich, St. Louis,
MO, USA). All measurements were performed twice; the results are pre-
sented as means, expressed as mg/kg on a dry matter basis (DM).
The flour colour was measured in triplicate using two Tristimu-
lus colorimeters, Chroma meter CR-210 (Minolta Italia SpA, Italy)
and Chroma meter II Reflectance (Minolta Camera Co., LTD, Japan).
The analysis led to the determination of the values of coordinates
L* (luminosity), a* (red–green) and b* (yellow–blue). Additionally,
the UV–VIS reflectance spectrum was recorded by a V-650
spectrophotometre (Jasco, Japan) with integrating sphere (PIN
757, Jasco, Japan). The flour samples were fitted into a container
(diametre: 20 mm; height: 2 mm); the spectra were measured at
wavelengths between 220 and 800 nm, with a 400 nm min− 1 scan-
sion speed, a 5 nm UV–VIS band broadness, and D2/WI lamp change
at 340 nm. The spectra of each sample were recorded three times;
the colour coordinates L*, a*, and b* were determined with the
spectrophotometre Spectra Manager software, using the parame-
ters: light source C, standard observer 2°, data interval 5 nm and
colour matching JISZ871-1999.
To perform image analysis, for each accession the images of
three flour-filled Petri dishes were acquired with an Epson Perfec-
tion 3170 Photo flatbed scanner (Seiko Epson Corporation, Nagano,
Japan), previously calibrated with IT8 Colour Targets system and
SylverFast software v.6.1 (LaserSoft Imaging Inc., Kiel, D). During
the acquisition process the samples were covered with a black
box to prevent loss of light; the images, acquired at a resolution of
600 dpi (dots per inch) and a colour depth of 24 bits, were saved
in uncompressed TIFF format. To create the final data set, two dif-
ferent ROIs were tested: 400 × 400 pixels and 250 × 250 pixels.
The final results were similar, so a region of interest (ROI) of 250
× 250 pixels, representative of the whole sample surface, was ex-
tracted from each single Petri dish image using the Image-Pro
Plus 7.0.1 software (Media Cybernetics, Inc., USA). This ROI was
chosen because of the inferior analysis time required by the algo-
rithm to process all the image data sets at the same time. On the
whole data set the following evaluations were performed:
- Colour and homogeneity: the colour evaluation was performed in
the RGB space colour, considering the red (R), green (G), blue
(B) and intensity mean (I) parameters. The homogeneity of each
surface was evaluated from the heterogeneity (HTG) parameter pro-
vided by the software Image-Pro Plus. This parameter, ranging from
0 (homogeneous surface) to 1 (heterogeneous surface), is used to
characterize the surface of different types of substrates (Fongaro &
Kvaal, 2013; Marti, Fongaro, Rossi, Lucisano, & Pagani, 2011). For
each parameter, the values are reported as mean of the three images
analysed.
- Image texture: the image texture of each region of interest was
evaluated by means of the Gray Level Co-occurrence Matrix
(GLCM) algorithm (Haralick, 1979). Fourteen different features
(angular second moment, contrast, sum of squares, correlation,
inverse different moment, entropy, sum average, sum variance,
sum entropy, difference variance, different entropy, two informa-
tion measures of correlation and maximum correlation coeffi-
cient) can be obtained by this method to describe the surface
characteristic of the image analysed although generally, as re-
ported by Zheng, Sun, and Zheng (2006), in the food field six of
them (angular second moment, contrast, sum of squares, correla-
tion, inverse different moment, entropy) are enough to find a
good relation between image texture and food properties. In
this work the following five Haralic Descriptors were calculated:
Angular Second Moment (ASM), showing the uniformity of an
image; Contrast (CON), showing the amount of local variation
present in a image; Correlation (COR), showing the pixel linear
dependencies; Entropy (ENT), a measure of statistical random-
ness related to image disorder and the Inverse Different Moment
(IDM), illustrating the homogeneity of an image (Fongaro &
Kvaal, 2013; Zheng et al., 2006). The images were processed
with a revised plug in to run on image stacks, named
GLCM_Texture (Cabrera, 2005), for the ImageJ v. 1.44c software
(Schneider, Rasband, & Eliceiri, 2012). The GLCM algorithm was
 A. Hidalgo et al. / Food Research International 64 (2014) 363–370 365

applied on the whole region of interest selected, considering dis-
tances δ of 1, 2, 3, 5, 7 and 10 pixels, and angles θ of 0°, 90°, 180°
and 270°; 120 textural features for each samples were thus
obtained.
2.3. Statistical analysis
Mean, standard error (s.e.) and coefficient of variation (CV) were
computed with the software Excel (Microsoft® Office Excel 2003).
Pearson's correlations of the means were calculated using the SYSTAT
for Windows (version 5) software. Standard Normal Variate (Barnes,
Dhanoa, & Lister, 1989), alone or coupled with first derivative, was test-
ed as a pre-treatment but did not improve the results of the analysis and
was not further implemented. Principal Components Analysis (PCA)
was applied to explore data structure (Vandeginste et al., 1997). PCA
defines new variables, consisting of linear combinations of the original
ones: the first axis is in the direction containing most variation; every
subsequent new variable (principal component) is orthogonal to the
previous one, but again in the direction containing most of the remain-
ing variance. The PCAs on the different colour spectra and on the image
analysis data were performed with the software The Unscrambler X
10.2 (CAMO software AS, Oslo, Norway).
3. Results and discussion
Fig. 1 reports the carotenoid, protein and ash content of einkorn
(four samples: Monlis N0, Monlis N40, Monlis N80 and Sal98-8-2-3),
Kamut® and durum wheat (two samples: Saragolla 2011 and Saragolla
2012). A relevant difference in carotenoid content between samples
was observed, and the highest concentration was found in Monlis (up
to 12.5 mg/kg DM). Interestingly, the flours from the three different ni-
trogen fertilisation levels (0, 40 and 80 kg/ha N) did not differ in carot-
enoid content. For all einkorn samples, the flours with smaller
granulometry (b80 μm) presented a lesser concentration of carotenoids
than other sizes. This effect could be due to the extra-soft texture of ein-
korn kernels that, upon milling, leads to the presence of significant
amounts of small size flour (b 80), mainly constituted by little starch
granules (Stoddard, 1999). This possibility is suggested also by the pro-
tein content of the b 80 μm fraction, which in all four einkorn samples
was about 20% lower than that of the other granulometry flours.
Saragolla and Kamut®, instead, showed fairly steady carotenoid and
protein contents across all different flour sizes.
In all samples ash content presented a steady increase, parallel to
flour grain size augment. The ash content gradient in the wheat kernel
increases from the centre to the outer layers (Posner, 2009). Aleurone
cells, in particular, are very rich in micro- and macro-elements
(Posner, 2009), and can lead to higher ash content in coarser flour
fractions.
Fig. 2 depicts the colour coordinates (L*, a* and b*) of einkorn (four
samples: Monlis N0, Monlis N40, Monlis N80 and Sal98-8-2-3),
Kamut® and durum wheat (two samples: Saragolla 2011 and Saragolla
2012). Only the data from the integrating sphere are presented; the re-
sults of the Minolta CR 210 and Minolta Chroma meter II colorimetres
(not shown) were almost identical, with the exception of the a* colour
coordinate, which had a variable behaviour. For all samples, the increase
in flour granulometry led to a decrease in luminosity (L*); on the con-
trary, b* augmented, although in flour grains larger than 200 μm Monlis
displayed a certain decrease. A similar behaviour was noticed by
Hrušková, Švec, and Sekerová (2011) in pasta samples remilled with
two different mills. The a* parameter showed a different trend for

14 20
18
Total carotenoids (mg/kg DM)

12
16
Protein (g/100 g DM)

10 14
12
8
10
6 8

4 6
4
2
2
0 0
<80 80-120 120-160 160-200 ≥200 <80 80-120 120-160 160-200 ≥200
Granulometry ( µm) Granulometry ( µm)

1.8
1.6
Monlis N0 2012
1.4
Monlis N40 2012
Ash (g/100 g DM)

1.2
Monlis N80 2012
1.0 Kamut 2012

0.8 Saragolla 2012

0.6 Saragolla 2011

0.4 Sal98-8-2-3 2011

0.2
0.0
<80 80-120 120-160 160-200 ≥200
Granulometry (µ m)

Fig. 1. Total carotenoid, protein and ash contents in flours of different granulometry from four einkorn (Monlis N0, Monlis N40, Monlis N80 and Sal98-8-2-3), one Kamut® and two durum
wheat (Saragolla 2011 and Saragolla 2012) samples.
366 A. Hidalgo et al. / Food Research International 64 (2014) 363–370

100 1.0
98
0.0
96
94 -1.0
92
-2.0

L*

a*
90
88 -3.0

86 -4.0
84
-5.0
82
80 -6.0
<80 80-120 120-160 160-200 ≥200 <80 80-120 120-160 160-200 ≥200
Granulometry ( µm) Granulometry ( µm)

25

Monlis N0 2012
20
Monlis N40 2012

Monlis N80 2012

15
Kamut 2012
b*

Saragolla 2012
10
Saragolla 2011

Sal98-8-2-3 2011
5

0
<80 80-120 120-160 160-200 ≥200
Granulometry ( µm)

Fig. 2. Colour coordinates L*, a*, b* assessed by integrating sphere (light source C) in flours of different granulometry from four einkorn (Monlis N0, Monlis N40, Monlis N80 and Sal98-8-2-
3), one Kamut® and two durum wheat (Saragolla 2011 and Saragolla 2012) samples.

1.0 1.0
Monlis N40 Sal98-8-2-3 2011
0.8 0.8

0.6 0.6
Absorbance
Absorbance

0.4 0.4

0.2 0.2

0.0 0.0

-0.2 -0.2
0 200 400 600 800 1000 0 200 400 600 800 1000
Wavelength (nm) Wavelength (nm)

1.0 1.0
Saragolla Kamut®
0.8 0.8

0.6 0.6
Absorbance

Absorbance

0.4 0.4

0.2 0.2

0.0 0.0

-0.2 -0.2
0 200 400 600 800 1000 0 200 400 600 800 1000

Wavelength (nm) Wavelength (nm)

D<80 80≤D<120 120≤D<160 160≤D<200 D≥200

Fig. 3. Absorbance spectra from integrating sphere in flours of different granulometry of einkorns Monlis N40 and SAL98-8-2-3, durum wheat Saragolla 2012 and Kamut®.
 A. Hidalgo et al. / Food Research International 64 (2014) 363–370 367

Monlis, reaching a minimum at flour size 120–160 μm and then increas- measures are strongly dependent on the intensity of each pixel and
ing, compared to SAL 98-8-2-3, which did not vary substantially, and to the position in the image, while R, G, B and I are mean values of total
the other wheats, which displayed a small, gradual decrease. pixels in the image.
Fig. 3 reports the absorbance spectra, obtained with an integrating Table 2 reports correlation coefficients among carotenoid, ash and
sphere, of different-granulometry flour of einkorns Monlis N40 and protein content, L*, a* and b* coordinates of the integrating sphere, co-
SAL98-8-2-3, durum wheat Saragolla and Kamut®, all from the 2012 ordinates from image analysis (R, G, B, I, HTG) and texture indices
harvest. For each spectrum, the absorbance changes as a function of (ASM, CON, COR, ENT, IDM). Only the texture indices computed from
granulometry: higher absorbance corresponds to bigger flour granules. the matrix obtained with the GLCM algorithm with 1 pixel step and 0°
This effect was particularly clear in einkorn, where the absorbance spec- direction were considered: this choice is justified because these settings
tra at different flour sizes were more spaced than in the tetraploid allow to obtain results that describe spot variations in the surface
wheats. These results are in accordance with the statement of Oliver analysed (Fongaro & Kvaal, 2013). The results indicate that the caroten-
et al. (1992) that an increase in particle dimension augments the length oid content is mildly related to protein content, a* and HTG. However no
of the spectral path in the samples, thus resulting in higher absorbance significant correlations were detected between carotenoid content and
and lower reflectance. In fact, smaller granules give a more compact a* with the two Minolta colorimeters. Thus, carotenoid content in
structure (characterized by minimal empty spaces) than the one com- general is not correlated to the colour indices studied. Protein content
posed by bigger particles. is linked only to a*; ash content, instead, shows good positive correla-
Table 1 reports mean, standard error and coefficient of variation tions to b*, HTG and ASM, and negative correlations to most of the
(%) of several colour and texture parameters across all samples. The other parameters assessed.
colour coordinate values obtained with the different instruments The variable L* is positively correlated (r = 0.95) with the I index,
were generally in the same range (L* varied between 88.2 and which describes the mean intensity of the surface; therefore, both
88.9; a* varied between − 1.5 and − 3.8; b* varied between 15.9 parameters express the variation in brightness perception as a function
and 20.5). Nevertheless, a broader difference was observed for the of the granulometry of the samples. Additionally, L* is positively corre-
coordinate a*, followed by the coordinate b*, while the luminosity lated to R (r = 0.91), G (r = 0.97) and B (r = 0.93), and negatively
L* was almost identical. A wider variation of the a* coordinate com- correlated to HTG (r = − 0.69), a parameter that describes surface
pared to L* and b* was observed also by Oliver et al. (1992). The cor- homogeneity. In fact, L* is high in the samples with flour size b 80 μm,
relation coefficients among instruments were significant and high characterized by a smooth and homogeneous surface, with low HTG
for L* and b* (on average, r = 0.94 and r = 0.99, respectively), but values (linked to a low variation of local intensity); as the granulometry
medium for a* (r = 0.45). McCaig (2002) reports a strict correlation size increases, the surface becomes more and more heterogeneous,
(r = 0.99) among the CIE Lab coordinates measured with three leading to a decrease in luminosity. L* shows good correlations also
different Minolta colorimeter models. Nevertheless, the mean with the ASM, CON, ENT and IDM values, other descriptors of the surface
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

2 texture. While a* does not present many significant correlations with
levels of ΔE ( ðΔL
Þ þ ðΔa
Þ2 þ ðΔb Þ ) between Minolta CR-210
2
the other parameters (and when present they are contrasting among
and Minolta Chroma meter II (4.1 ± 0.10), between Minolta CR- the different instruments), b* shows positive correlations to HTG and
210 and integrating sphere with C light source (4.9 ± 0.13), and be- ASM, as well as negative correlations with CON, ENT and IDM, R, G, I
tween Minolta Chroma meter II and integrating sphere with C light and especially B (r = −0.89). This last result is not unexpected, because
source (3.6 ± 0.12) suggest minimal colour difference among b* goes from negative (blue) to positive (yellow), while B refers to the
methods. BLUE chromatic channel and ranges from 0 to 255: thus, when b*
Concerning the colour index obtained by image analysis (Table 1), decreases B increases. Finally, the R, G, B and I parameters present
the variation of each colour channel was in the same range even if the negative correlations with HTG and ASM, and positive with CON, ENT,
B channel had the highest CV%. On the other hand, considering HTG and IDM.
and the texture indices (ASM, CON, COR, IDM), the higher variations The Principal Components Analysis, performed on the spectra cap-
with respect to the colour values were expected, because these tured from 800 to 220 nm with integrating sphere (not shown), led to
a good separation of the different samples, based on flour granulometry,
Table 1 along the PC1 (91% of total variation). An improved separation (Fig. 4)
Mean values, across samples, of carotenoid, protein and ash contents, chromatic was achieved considering only the spectra captured between 500 and
coordinates and texture indices obtained by different analytical methods. 400 nm (i.e. the maximum absorbance range of carotenoids), especially
Average ± s.e. CV (%) for the bigger granulometry classes. Zandomeneghi, Festa, and
Carotenoids (mg/kg DM) 8.7 ± 0.31 29.36
Carbonaro (2000), studying the front-surface absorbance spectra of
Protein (g/100 g DM) 13.7 ± 0.26 16.12 flour from five durum and four bread wheats in the 650–250 nm
Ash (g/100 g) 0.94 ± 0.035 31.51 wavelength range, found that the best region for carotenoid content
L* Minolta CR-210 88.9 ± 0.37 4.26 analysis was between 450 and 500 nm.
L* Minolta Chroma meter II 89.9 ± 0.31 3.58
Fig. 5 depicts the score plot of the PCA performed considering ca-
L* sphere (light source C) 88.2 ± 0.35 4.10
a* Minolta CR-210 −1.5 ± 0.05 37.53 rotenoid, protein and ash contents, colorimetric indices recorded
a* Minolta Chroma meter II −3.8 ± 0.09 25.01 with integrating sphere (L*, a*, b*) and image analysis (R, G, B, I),
a* sphere (light source C) −1.6 ± 0.06 40.67 and heterogeneity. The total variance explained was 82%. The sam-
b* Minolta CR-210 20.5 ± 0.52 25.93 ples were separated along the PC1 for their granulometry, as the par-
b* Minolta Chroma meter II 17.7 ± 0.50 28.72
b* sphere (light source C) 15.9 ± 0.44 28.14
ticle size increased going from positive to negative PC1 values. The
R 227.7 ± 0.83 3.72 separation along the PC2, instead, was a function of the species: the
G 219.1 ± 1.17 5.47 einkorn samples were in the positive part, while the durum and
B 178.3 ± 2.43 13.98 Kamut® samples were in the negative one. The separation along
I 208.4 ± 1.45 7.14
the PC1 is linked mainly to colour, while the separation along the
Heterogeneity 0.0209 ± 0.0023 113.61
ASM 0.005 ± 0.0006 132.61 PC2 is related to protein and carotenoid content. A PCA performed
CON 271.3 ± 7.04 26.58 including also the texture indices (ASM, CON, COR, ENT and IDM)
COR 0.001 ± 0.0000 42.29 gave a similar separation.
ENT 6.4 ± 0.11 18.19 The score plot of the PCA performed on the texture data matrix from
IDM 0.2 ± 0.01 64.62
the GLCM image elaboration is presented in Fig. 6. PC1 and PC2
368 A. Hidalgo et al. / Food Research International 64 (2014) 363–370

Table 2
Pearson's correlation coefficients (n = 35) among carotenoid, protein and ash contents, colour coordinates L*, a*, b* measured by a spectrophotometre with integrating sphere, chromatic
parametres (R, G, B, HTG) and texture indices (ASM, CON, COR, ENT, IDM) obtained by image analysis.

Carotenoids Ash Protein L* a* b* R G B I HTG ASM CON COR ENT

Ash −0.07
Protein 0.52⁎⁎ 0.01
L* −0.11 −0.85⁎⁎ 0.06
a* 0.36⁎ −0.01 0.47⁎⁎ 0.11
b* 0.13 0.73⁎⁎ −0.01 −0.83⁎⁎ −0.46⁎⁎
R −0.23 −0.72⁎⁎ 0.18 0.91⁎⁎ 0.03 −0.68⁎⁎
G −0.17 −0.81⁎⁎ 0.13 0.97⁎⁎ 0.14 −0.82⁎⁎ 0.97⁎⁎
B −0.19 −0.77⁎⁎ 0.18 0.93⁎⁎ 0.32 −0.89⁎⁎ 0.91⁎⁎ 0.97⁎⁎
I −0.20 −0.79⁎⁎ 0.17 0.95⁎⁎ 0.22 −0.85⁎⁎ 0.96⁎⁎ 0.99⁎⁎ 0.99⁎⁎
HTG 0.43⁎⁎ 0.77⁎⁎ 0.24 −0.69⁎⁎ 0.38⁎ 0.52⁎⁎ −0.64⁎⁎ −0.67⁎⁎ −0.61⁎⁎ −0.64⁎⁎
ASM 0.07 0.79⁎⁎ 0.15 −0.63⁎⁎ 0.21 0.49⁎⁎ −0.54⁎⁎ −0.60⁎⁎ −0.55⁎⁎ −0.57⁎⁎ 0.76⁎⁎
CON 0.03 −0.63⁎⁎ 0.10 0.65⁎⁎ 0.20 −0.67⁎⁎ 0.52⁎⁎ 0.62⁎⁎ 0.59⁎⁎ 0.59⁎⁎ −0.42⁎⁎ −0.48⁎⁎
COR −0.29 −0.28 0.16 0.24 0.18 −0.21 0.30 0.28 0.37⁎ 0.34⁎ −0.18 −0.15 0.07
ENT −0.04 −0.85⁎⁎ −0.05 0.76⁎⁎ 0.02 −0.72⁎⁎ 0.63⁎⁎ 0.73⁎⁎ 0.69⁎⁎ 0.70⁎⁎ −0.69⁎⁎ −0.86⁎⁎ 0.82⁎⁎ 0.23
IDM −0.16 −0.77⁎⁎ 0.05 0.92⁎⁎ 0.32 −0.94⁎⁎ 0.79⁎⁎ 0.90⁎⁎ 0.94⁎⁎ 0.92⁎⁎ −0.60⁎⁎ −0.56⁎⁎ 0.70⁎⁎ 0.29 0.73⁎⁎
⁎ p ≤ 0.05.
⁎⁎ p ≤ 0.01.

explained 63% and 23% of total variation, respectively. The discrimina-
tion of the samples was similar to those achieved by the other methods.
Fig. 6 shows the distribution of samples classified only by surface tex-
ture characteristics; the results are independent from colour because
the images were converted to grey level before analysis. The indices
for surface texture description in fact refer to the relationship existing
among the intensities of the pixels that compose the image in terms of
frequency and position. Nevertheless, because ASM, ENT and IDM
are correlated to the colorimetric indices (Table 2), there is a strict
dependence of colour from granulometry. As already discussed
above, changes in granulometry lead to variations in texture indices
and, as a consequence, in luminosity.
4. Conclusions
The total carotenoid content was stable across flours of different
granulometry, with the exception of the thinnest einkorn fraction
(b80 μm), which had lower concentration. When flour size augmented,
L* decreased and b* increased while a* showed a varying behaviour
between instruments. A PCA carried out considering carotenoid, protein
and ash contents along with colorimetric indices allowed a good dis-
crimination of the samples of different granulometry, as well as the sep-
aration of T. monococcum from T. turgidum; a similar result was achieved
with a PCA performed on the integrating sphere spectra from 400 to
500 nm. Another PCA, run on the matrix obtained from GLMC image
analysis, classified the samples following flour particle size. Our results
demonstrated that flour colour is determined not only by carotenoid
content but also by flour particle size. Thus, considering the parameters
chosen in this work, the direct measurement of flour colour can hardly
predict the concentration of yellow pigments.
Contributors
Alyssa Hidalgo planned the research, performed the total carotenoid
analysis and the flour colour measurements, participated in the analysis
of the results and in the preparation of the article. Lorenzo Fongaro
carried out the image analysis, participated in the analysis of the results
and in the preparation of the article. Andrea Brandolini planned the

Fig. 4. Plot of the scores of the PCA carried out on absorbance spectra ranging between 500
and 400 nm, detected by integrating sphere in flours of different granulometry from four
einkorn, Monlis N0 (Mon0), Monlis N40 (Mon40), Monlis N80 (Mon80) and Sal98-8-2-3
(SAL), one Kamut® (Kam) and two durum wheat, Saragolla 2011 (Sara11) and Saragolla
2012 (Sara12) samples.
Fig. 5. Plot of the scores of the PCA carried out on protein, ash, total carotenoids, colour co-
ordinates assessed by integrating sphere (L*, a*, b*) and image analysis (R, G, B, I) and het-
erogeneity (HTG) in flours of different granulometry from four einkorn, Monlis N0
(Mon0), Monlis N40 (Mon40), Monlis N80 (Mon80) and Sal98-8-2-3 (SAL), one
Kamut® (Kam) and two durum wheat, Saragolla 2011 (Sara11) and Saragolla 2012
(Sara12) samples.
 A. Hidalgo et al. / Food Research International 64 (2014) 363–370
Fig. 6. Plot of the scores of the PCA carried out on the matrix obtained from GLMC image analysis of flours of different granulometry from four einkorn, Monlis N0 (Mon0), Monlis N40
(Mon40), Monlis N80 (Mon80) and Sal98-8-2-3 (SAL), one Kamut® (Kam) and two durum wheat, Saragolla 2011 (Sara11) and Saragolla 2012 (Sara12) samples.
research, cropped the wheat genotypes, prepared the flour samples,
determined dry matter, ash and protein contents, participated in the
analysis of the results and in the preparation of the article.
References
AACC American Association of Cereal Chemists (1995). AACC Official Methods 08-03, 44-
15A, 46-10, 66–20. In Approved Methods of the American Association of Cereal Chemists
(9th ed.). MN, USA: Minneapolis.
AACC International (2012). Approved methods of analysis Method 14-60.01. (11th Ed ).
Total carotenoid content of cereal grains and flours. MN, U.S.A.: AACC International,
St. Paul, http://dx.doi.org/10.1094/AACCIntMethod-14-60.01.
Abdel-Aal, E. -S. M., Young, J. C., Wood, P. J., Rabalski, I., Hucl, P., Falk, D., et al. (2002).
Einkorn: A potential candidate for developing high lutein wheat. Cereal Chemistry,
79, 455–457.
Adom, K. K., & Liu, R. H. (2002). Antioxidant activity of grains. Journal of Agriculture and
Food Chemistry, 50, 6182–6187.
Barnes, R. J., Dhanoa, M. S., & Lister, S. J. (1989). Standard normal variate transformation
and de-trending of near-infrared diffuse reflectance spectra. Applied Spectroscopy,
43, 772–777.
Black, C. K., & Panozzo, J. F. (2004). Accurate technique for measuring color values of
grain and grain products using a visible-NIR instrument. Cereal Chemistry, 81,
469–474.
Blanco, A., Colasuonno, P., Gadaleta, A., Mangini, G., Schiavulli, A., Simeone, R., et al.
(2011). Quantitative trait loci for yellow pigment concentration and individual
carotenoid compounds in durum wheat. Journal of Cereal Science, 54, 255–264.
Brandolini, A., Hidalgo, A., & Moscaritolo, S. (2008). Chemical composition and pasting
properties of einkorn (Triticum monococcum L. subsp. monococcum) whole meal
flour. Journal of Cereal Science, 47, 599–609.
Cabrera, J. E. (2005). GLCM_Texture plug-in. http://rsb.info.nih.gov/ij/plugins/texture.
html
Castagna, R., Borghi, B., Di Fonzo, N., Heun, M., & Salamini, F. (1995). Yield and related
traits of einkorn (Triticum monococcum ssp. monococcum) in different environments.
European Journal of Agronomy, 4, 371–378.
CIE Commission Internationale de l'Eclairage (2004). Colorimetry Publication CIE 15. (3rd
ed ). Vienna, Austria: Commission Internationale de l'Eclairage.
Dowell, F. E., Maghirang, E. B., Xie, F., Lookhart, G. L., Pierce, R. O., Seabourn, B. W.,
et al. (2006). Predicting wheat quality characteristics and functionality using
near-infrared spectroscopy. Cereal Chemistry, 83, 529–536.
Fongaro, L., & Kvaal, K. (2013). Surface texture characterization of an Italian pasta by
means of univariate and multivariate feature extraction from their texture images.
Food Research International, 51, 693–705.
369
Fratianni, A., Giuzio, L., Di Criscio, T., Zina, F., & Panfili, G. (2013). Response of carotenoids
and tocols of durum wheat in relation to water stress and sulfur fertilization. Journal
of Agricultural and Food Chemistry, 61, 2583–2590.
Fratianni, A., Irano, M., Panfili, G., & Acquistucci, R. (2005). Estimation of color of durum
wheat. Comparison of WSB, HPLC, and reflectance colorimeter measurements.
Journal of Agricultural and Food Chemistry, 53, 2373–2378.
Haralick, R. M. (1979). Statistical and structural approaches to texture. Proceedings of the
IEEE, 67, 786–804.
Hidalgo, A., & Brandolini, A. (2008a). Protein, ash, lutein and tocols distribution in einkorn
(Triticum monococcum L. subsp. monococcum) seed fractions. Food Chemistry, 107,
444–448.
Hidalgo, A., & Brandolini, A. (2008b). Kinetics of carotenoids degradation during the stor-
age of einkorn (Triticum monococcum L. ssp. monococcum) and bread wheat (Triticum
aestivum L. ssp. aestivum) flours. Journal of Agricultural and Food Chemistry, 56,
11300–11305.
Hidalgo, A., Brandolini, A., Pompei, C., & Piscozzi, R. (2006). Carotenoids and tocols of ein-
korn wheat (Triticum monococcum ssp. monococcum L.). Journal of Cereal Science, 44,
182–193.
Hidalgo, A., Brandolini, A., & Ratti, S. (2009). Influence of genetic and environmental fac-
tors on selected nutritional traits of Triticum monococcum. Journal of Agricultural and
Food Chemistry, 57, 6342–6348.
Hrušková, M., Švec, I., & Sekerová, H. (2011). Colour analysis and discrimination of labo-
ratory prepared pasta by means of spectroscopic methods. Czech Journal of Food
Science, 29, 346–353.
Lachman, J., Hejtmánková, K., & Kotíková, Z. (2013). Tocols and carotenoids of einkorn,
emmer and spring wheat varieties: Selection for breeding and production. Journal
of Cereal Science, 57, 207–214.
Leenhardt, F., Lyan, B., Rock, E., Boussard, A., Potus, J., Chanliaud, E., et al. (2006). Genetic
variability of carotenoid concentration, and lipoxygenase and peroxidase activities
among cultivated wheat species and bread wheat varieties. European Journal of
Agronomy, 25, 170–176.
Mares, D. J., & Campbell, A. W. (2001). Mapping components of flour and noodle colour in
Australian wheat. Australian Journal of Agricultural Research, 52, 1297–1309.
Marti, A., Fongaro, L., Rossi, M., Lucisano, M., & Pagani, M.A. (2011). Quality characteristics
of pasta enriched with buckwheat flour. International Journal of Food Science and
Technology, 46, 2393–2400.
McCaig, T. N. (2002). Extending the use of visible/near-infrared reflectance spectropho-
tometers to measure colour of food and agricultural products. Food Research
International, 35, 731–736.
Oliver, J. R., Blakeney, A. B., & Allen, H. M. (1992). Measurement of flour color in color
space parameters. Cereal Chemistry, 69, 546–551.
Panfili, G., Fratianni, A., & Irano, M. (2004). Improved normal-phase high-performance
liquid chromatography procedure for the determination of carotenoids in cereals.
Journal of Agricultural and Food Chemistry, 52, 6373–6377.
370 A. Hidalgo et al. / Food Research International 64 (2014) 363–370
Posner, E. S. (2009). Wheat flour milling. In K. Khan, & P. R. Shewry (Eds.), Wheat chem-
istry and technology (pp. 119–152) (4th Ed.). St Paul, MN, USA: AACC International.
Schneider, C. A., Rasband, W. S., & Eliceiri, K. W. (2012). NIH Image to ImageJ: 25 years of
image analysis. Nature Methods, 9, 671–675.
Stoddard, F. L. (1999). Survey of starch particle-size distribution in wheat and related spe-
cies. Cereal Chemistry, 76, 145–149.
Symons, S. J., & Dexter, J. E. (1991). Computer analysis of fluorescence for the measure-
ment of flour refinement as determined by flour ash content, flour grade color, and
tristimulus color measurements. Cereal Chemistry, 68, 454–460.
Vandeginste, B. G. M., Massart, D. L., Buydens, L. M. C., De Jong, S., Lewi, P. J., & Smeyers-
Verbeke, J. (1997). Handbook of chemometrics and qualimetrics. Part A. Amsterdam:
Elsevier.
Zandomeneghi, M., Festa, C., & Carbonaro, L. (2000). Front-surface absorbance spectra of
wheat flour: Determination of carotenoids. Journal of Agricultural and Food Chemistry,
48, 2216–2221.
Zheng, C., Sun, D. -W., & Zheng, L. (2006). Recent applications of image texture for
evaluation of food qualities — A review. Trends in Food Science & Technology,
17, 113–128.