RSC Advances

View Article Online

PAPER View Journal | View Issue

Towards defining new nano-descriptors: extracting

Published on 05 November 2014. Downloaded by INIST - CNRS (Chemistry) on 10/3/2022 6:55:24 PM.

morphological features from transmission electron

Cite this: RSC Adv., 2014, 4, 60135
microscopy images†
Arafeh Bigdeli,a Mohammad Reza Hormozi-Nezhad,*ab Mehdi Jalali-Heravi,a
Mohammad Reza Abedinic and Farzad Sharif-Bakhtiard

Due to the important role of surface-related properties of NPs in their biological behavior, simple and fast
methods that could precisely demonstrate accurate information about NPs' surface, structure and
morphology are highly desirable. In this study a set of surface morphological nano-descriptors (size,
shape, surface area, agglomeration state, curvature, corner count and aspect ratio) have been defined
and extracted from Transmission Electron Microscopy (TEM) images of nanoparticles (NPs) by Digital
Image Processing methods. The extracted data represent a thorough description of the surface and
morphologies of NPs lying beyond their TEM images and can supply the data required for a nano-QSAR
approach for predicting toxicity profiles of NPs. These nano-descriptors can provide a framework to
further understand the mechanisms which govern the adverse effects of NPs in biological systems.
Received 13th September 2014
Accepted 5th November 2014
Metallic nanostructures (gold, silver, palladium.) with different sizes (10 to 100 nm), shapes (cube,
sphere, rod.) and characteristics were taken into account for which physicochemical indexes were
DOI: 10.1039/c4ra10375k
reported. To the best of our knowledge, this is the first ever study that presents numerical values for
www.rsc.org/advances properties such as shape and agglomeration state which significantly affect NPs behavior.

measured prole of a set of nanostructures and their physico-

1. Introduction chemical properties (called “nano-descriptors”). Thus, through
Today with the developments in nanotechnology which has a nano-QSAR approach, one would be able to quantitatively
signicantly improved the quality of life for human beings, it is predict the potential toxicity of a set of un-tested NPs based on
important to address the possible consequences, as with any experimental toxicological data available for a set of tested ones
emerging technology. Computational approaches play an and therefore, prevent expensive and time-consuming empir-
essential role in this risk assessment procedure due to their fast, ical animal testing procedures for NPs risk assessment.
in-expensive and high throughput methods. However, it must However, since NPs signicantly differ to their bulk counter-
be noted that the negative impacts of NPs should be carefully parts, consequently, nano-QSAR differs to the well-known
considered and evaluated by gathering specialists in both conventional QSAR approach (for which there are several
experimental and theoretical elds. The large number of NPs commercial soware available and large sets of molecular
and the variety of their characteristics including various sizes, descriptors are calculated)3 and there is a need to develop QSAR
shapes and coatings suggest that the only rational approach models with a special insight to nanomaterials. Actually, some
which avoids testing every single NP is to nd a relationship major obstacles impede the nano-QSAR approach such as:
between the physicochemical characteristics of NPs and their structural complexity and diversity of NPs, scarce and/or
toxicity.1,2 This approach, namely called Quantitative Structure– inconsistent empirical data and thus lack of available large
Activity Relationship of nanomaterials (nano-QSAR), statisti- scale datasets of NPs' toxicity, and nally lack of rational
cally establishes a mathematical relationship between a modeling procedures in describing the structural properties of
these substances.4 Therefore, nano-specic descriptors
responsible in determining the toxicity of nanostructures are
a
Department of Chemistry, Sharif University of Technology, Tehran, Iran. E-mail: markedly required. Developing these novel nano-descriptors
[email protected]; Tel: +98 21 6616 5337 could be a great challenge for computational experts. A
b
Institute for Nanoscience and Nanotechnology (INST), Sharif University of number of research groups have already expressed computa-
Technology, Tehran, Iran
c
tional and empirical nano-descriptors for revealing the behavior
Department of Electrical Engineering, Sharif University of Technology, Tehran, Iran
d
of nanomaterials.5 For example, Puzyn et al.6 presented a set of
Department of Computer Engineering, Sharif University of Technology, Tehran, Iran
† Electronic supplementary information (ESI) available. See DOI:
quantum mechanical descriptors for modeling the cytotoxicity
10.1039/c4ra10375k of metal-oxide NPs to bacteria Escherichia coli. Martin et al.7

This journal is © The Royal Society of Chemistry 2014 RSC Adv., 2014, 4, 60135–60143 | 60135
 View Article Online

RSC Advances Paper

proposed two models predicting solubility of C60 fullerene in n- capable of dening thresholds for different kinds of images
octanol and n-heptane. Fourches et al.8 developed a quantitative utilizing machine learning techniques. Several transitions were
nanostructure-activity relationship (QNAR) model to predict the examined meanwhile to achieve the most appropriate code that
cellular uptake of 109 NPs in PaCa2 cell. But none of them have could accurately nd all the objects in an image and calculate
yet succeeded to present numerical values for morphological the considered descriptors for them. In addition, a number of
properties of nanoparticles such as their shape or surface- new features such as the amount of rigidity or curvature of the
related features. With this in mind, image processing, might corners and some other possible informative parameters where
be a great clue for extracting quantitative information about the proposed and dened throughout the process of calculating the
Published on 05 November 2014. Downloaded by INIST - CNRS (Chemistry) on 10/3/2022 6:55:24 PM.

structure of the investigated objects in NPs images and could be previous mentioned descriptors which are fully introduced and
used to establish a link between this information and various discussed later. Digital image processing algorithms were
non-visual properties of the objects.9–12 coded and compiled by C++ language and GCC 4.6.1 in Linux,
In the present contribution, a set of novel image nano- respectively.23 In addition, some basic ideas on image process-
descriptors were extracted from TEM images of NPs in order ing were implemented from OpenCV and QT 5.2.1 libraries.24,25
to reveal as much as information possible about their struc- For each TEM image, an output folder was saved and exported
tural, morphological and surface properties. The extracted containing the calculated nano-descriptors related to the
nano-descriptors can be used either as a nal result (for considered image. The complete description of image process-
example estimation of the aggregation state) or treated as an ing steps are explained in the following.
input for further analysis (such modeling NPs toxicity prole or
cellular uptake). Furthermore, by combining these morpho-
logical image nano-descriptors with a series of experimentally
assessed parameters of NPs (i.e., zeta potential, ionic strength,
crystal structure, hydrophobicity, .) and even previously sug- 3. Results and discussion
gested QM descriptors, a favorable dataset could be provided for 3.1. Overall algorithm
applying computational methods such as nano-QSAR in order Aer importing a TEM image in .jpeg or .png format as input,
to assess NPs potential risk to human health and the the desired nano-descriptors were generated in an overall
environment. algorithm as follows: (1) initialization: including pre-processing
steps to improve the image resolution, to remove the back-
2. Methods ground and to nd all the objects in the considered image; (2)
calculating nano-descriptors (size, surface area, aspect ratio,
2.1. Dataset corner count, agglomeration state, shape type and curvature);
An initial set of TEM images of metallic NPs were collected from (3) removing outliers; (4) saving the output le.
our previous work13 and literature14–21 to gather a diverse
selection of different types of NPs including various sizes,
shapes and comparable agglomeration states. Six representative
TEM image candidates are shown in Fig. 1 and a larger set of the 3.2. Pre-processing step
images can be seen in S1.† Because of the wide variety of NPs to
be considered, consisting carbon based materials, metallic and The rst step was to improve the image quality and nd all the
metal oxide nanoparticles, quantum dots., and owing to their objects in that image to then calculate nano-descriptors for
structural heterogeneity, we herein focused on one of the above them. Thus, aer reading the input image some image lters
mentioned classes including metallic and metal oxide NPs due were applied to de-noise the image and sharpen the blurred
to their large number of recent toxicity reports. edges of the objects in the image. The pre-processing was fol-
lowed by employing morphological operators (erosion and
dilation) to separate joined objects and eliminate very small
2.2. Descriptor generation by digital image processing ones (useless shapes which lack information). In the next step,
In order to develop morphological image nano-descriptors, the thresholds were dened for background subtraction. For this
rst step was to dene appropriate descriptors from which purpose, the greyscale image was converted to binary image in
maximum possible surface and structural information22 of NPs which a threshold is calculated individually for each pixel based
could be extracted. An initial set of surface descriptors was rst on its neighborhood pixels and reveals an array of zero and one
dened as suggested by Puzyn et al.4 including size distribution, values. Zero corresponds to the background pixels, while one
agglomeration state, shape, porosity and surface area. For each corresponds to the presence of an object upon that pixel. By
suggested descriptor, a number of different image processing applying this mask on the initial image, the algorithm could
codes were tested and applied to a pool of multiple TEM images successfully discriminate between the object and the back-
(i.e., TEM of rods, spheres, cubes.) to achieve the most ground pixels. Fig. 2 illustrates the background subtraction
possible general code. Actually, one of the basic challenges process on a sample TEM image. Now, by applying the nd
herein was to provide general codes, i.e., codes with maximum contour process on the resulted image, the exact location of
accuracy for diverse TEM images and minimum erroneousness each object and its border could be detected. More information
for noisy/blurred images. Moreover, general codes can be on the nd contour algorithm can be found elsewhere.26

60136 | RSC Adv., 2014, 4, 60135–60143 This journal is © The Royal Society of Chemistry 2014
 View Article Online

Paper RSC Advances

Published on 05 November 2014. Downloaded by INIST - CNRS (Chemistry) on 10/3/2022 6:55:24 PM.

Fig. 1 TEM images dataset including six different shape types of nanoparticles: (a) in spherical colloids,14 (b) Pt cuboctahedrons,15 (c) Pd triangular
nanoplates,16 (d) Pd nanocubes,17 (e) Au dog bone-shaped nanocrystals18 and (f) Au nanorods.13

3.3. Nano-descriptors images is limited to information from 2D shapes. The surface

area calculated herein is actually the surface area of a 2D cross
Aer detecting all the objects in an image, seven nano-
section of the particle which can be calculated by counting the
descriptors were calculated for each object in that image
pixels beneath that shape. Thus, measuring the contour areas of
which are briey dened below:
each image would give the surface areas of the objects in that
3.3.1. Size and surface area. Extracting the exact size of the
image individually and thus, exporting an average value as the
NPs from an image processing insight avoids us from reporting
representative surface area value for that image. This descriptor
imprecise and approximate values for NPs' size (when using just
can be further used to detect outliers in an image, i.e., objects
a simple ruler) and therefore, allows to scrutinize the real effect
which don't match others (are rather too small or too big
of particle size on NPs behavior. In this project, the size of each
compared to most of the objects). The threshold for eliminating
object was dened as the maximum distance between two
outliers is that objects with surface areas higher than twice of
points on the border of that object. Therefore, by nding the
the mean surface are or lower than half of the mean surface area
circumscribed rectangle of each object, we exported its length
should be omitted from the image.
on account of the size of the object within it. The circumscribed
3.3.2. Agglomeration state. Another important factor that
rectangle around some objects in a sample TEM image are
can strongly inuence the behavior of nanoparticles is their
shown in Fig. 3. The aspect ratio descriptor could be also
agglomeration state. An initial idea was to describe a standard
calculated by calculating the length/width ratio of the corre-
value between zero and one as the agglomeration state of the
sponding circumscribed rectangle. Since the TEM images reveal
particles in an image which respectively represent totally
a two-dimentional representation of nanoparticles, the whole
dispersed (particles far away from each other) or highly
process on extracting morphological descriptors from TEM

Fig. 2 Pre-processing results in the initialization step: (a) original image, (b) threshold for background subtraction, (c) background subtracted
image and (d) all the objects found in the image.

This journal is © The Royal Society of Chemistry 2014 RSC Adv., 2014, 4, 60135–60143 | 60137
 View Article Online

RSC Advances Paper

agglomeration state. With this weighted agglomeration state we

could take into account images which included different
agglomeration states in different areas within them. For this
purpose, the next step was to depict a circle around the center of
each object with a radius measured from eqn (3), and nd the
number of objects present in this criteria around each object.
pffiffiffiffi
Dþ S
r¼ (3)
Published on 05 November 2014. Downloaded by INIST - CNRS (Chemistry) on 10/3/2022 6:55:24 PM.

In eqn (3), n corresponds to the total number of objects in

the image. r will be calculated for each object separately and
then a mean value will be used as the circle radius covering
around each object in which the number of objects will be
counted, as shown in Fig. 4. Finally, a representative agglom-
eration state value could be calculated from eqn (4):
X
n
Nj  aggj
j¼1
Agglomeration state ¼ X
n (4)
Nj
j¼1

In this equation Nj corresponds to the number of neighbors

Fig. 3 Circumscribed rectangle depicted around each object. in the dened criteria (explained above) around each object.
3.3.3. Corner count. With the aim of extracting further
information from the TEM images, one idea was to reveal the
agglomerated particles (particles adjacent to each other). exact number of corners of the objects. Though we already had
Therefore, in order to dene this descriptor, the rst step was to an approximation of the number of corners throughout the
nd the neighbors around each object in an image and then shape descriptor dened above, we were looking for a more
dene a function related to both the distance and the number of precise feature that could somehow reveal the amount of
neighbors. Thus, three nearest neighbors to each object were homogeneity or isotropy of the shapes. For example, if we
rst found and a corresponding weight was then assigned to consider two squares, we expect they both have four number of
each of the neighbors in a descending order (i.e., a weight of 4 to corners. This would be true just if we had two clean squares,
the rst nearest neighbor, a weight of 2 to the second nearest with no noisy or damaged corners. But actually in the case of
neighbor, and a weight of 1 to the third nearest neighbor). Then real samples, we rarely face such clean squares. Instead, the
a weighted distance (D) was calculated from these three shapes include protrusions or depressions along their borders
neighbors according to eqn (1): that bring about more or less number of expected corners.
Therefore, by nding the exact number of the corners of the
X
3
shapes (which we called it corner count), we will be able to
wi di
i¼1 disclose information about the shapes symmetry and could
D¼ (1)
7

Now, for each object in an image, an agglomeration value

could be calculated from eqn (2):
 
D
Agg ¼ exp pffiffiffiffi (2)
S

In which D is the mean distance calculated by eqn (1) and S is

the surface area of the corresponding object which has been
measured in the surface area descriptor. Before reporting a
mean value of the measured agglomeration values for all the
objects in an image as a representative output, a much more
accurate agglomeration value could be obtained if we would be Fig. 4 Finding the neighbors around each object in the determined
region around it: (a) simple representation by a green shadowed circle
able to count the number of neighbors around each object
around the central square, (b) true calculated region around each
within a specic region and then calculate a weighted object determined by the software.

60138 | RSC Adv., 2014, 4, 60135–60143 This journal is © The Royal Society of Chemistry 2014
 View Article Online

Paper RSC Advances

Published on 05 November 2014. Downloaded by INIST - CNRS (Chemistry) on 10/3/2022 6:55:24 PM.

Fig. 5 A method for discrimination between a square and a triangle using a 1-D functional representation of the shape boundary (the distance
from the centroid is illustrated vs. the angle).33

even discriminate precisely between objects within a same collected TEM image database, we dened six major classes for
shape class. For this purpose the idea was to nd the approxi- shape type including: rod, dog-bone, square, triangle, circle and
mate polygon depicted around each shape and report its exact hexagon. It must be kept in mind that since we are restricted to
number of corners. It was found that the approximated polygon 2D images, we could offer 2D information for shape type; i.e.,
could precisely take into account interrupted edges of the noisy these shape types are in some cases common cross sections of
shapes. different 3D nanostructures. Various algorithms were tested to
3.3.4. Shape type. Shape, as a substantial parameter distinguish between these six shapes. For example to decide
affecting the behavior of nanoparticles,27–32 has been poorly whether a shape is a square or a triangle, one idea was to count
taken into account in nanomaterial risk assessment studies the number of peaks appearing in the histogram of the tangent
suffering from the lack of a method to translate the shape type angle. As illustrated in Fig. 5, a triangle shape would exhibit
into numerical values. In the present contribution, we three peaks due to three corners, while a square shape would
successfully dened shape descriptors that could distinguish expose four peaks.
between different shape types. For this purpose, due to our

Fig. 6 Histograms resulted for the discrimination between (a) triangle15 and (b) square16 shaped TEM images.

This journal is © The Royal Society of Chemistry 2014 RSC Adv., 2014, 4, 60135–60143 | 60139
 View Article Online

RSC Advances Paper

Published on 05 November 2014. Downloaded by INIST - CNRS (Chemistry) on 10/3/2022 6:55:24 PM.

Fig. 7 A method for discrimination between a circle and a hexagon. For each shape, imagine a red dashed circle crossing three arbitrary points
on the border of the considered shape. (a) In the case of a circle, the red crossing circle will always have a radius near the radius of the original
circle, no matter where these three points are being chosen. Thus, only one peak will appear in the histogram of the red circle's radius. (b) But in
the case of a hexagon, the red crossing circle can either have a small or a big radius, depending on the distance between the three random points:
if the three points were chosen close to each other, the red crossing circle would have a big radius, but if the three points were chosen far from
each other, the red crossing circle would have a small radius. Therefore, in the case of a hexagon, more than one peak will appear in the
histogram of the red crossing circle's radius.

can be seen, the number of peaks in the histogram is notably

inuenced by the tiny noises in the shape border.
Another challenge was to distinguish between a circle and a
hexagon for which several ideas were tested. For example, one
idea was based on a simple geometric theorem which states that
Fig. 8 Simple demonstration of different curvature limits.
only one circle crosses from any three individual points in a
plane. Utilizing this basic geometric rule, our idea was to
severally select three arbitrary points on the border of a circle or
a hexagon and then calculate the radius of the circle crossing
This idea worked out for clean squares/triangles, but in the
the three points each time. In the case of a circle, no matter
case of noisy ones the algorithm failed to classify these shapes.
where the three points are selected only one unique circle will
The results for two TEM image samples are shown in Fig. 6. As it

Table 1 Nano-descriptors calculated for all the nanoparticles in TEM image in Fig. 2

Aspect Agglomeration
Objects Size Surface area Curvature ratio Corner count Circle Rod Dogbone Triangle Square Hexagon state

1 24.0 416.5 0.211 1.143 8 0.905 0.267 0.236 0.700 0.567 0.792 0.006
2 55.0 1009.5 0.268 2.563 6 0.127 0.994 0.901 0.000 0.730 0.000 0.036
3 46.9 593.0 0.458 2.844 8 0.935 0.969 0.749 0.000 0.609 0.000 0.023
4 53.5 775.5 0.732 3.097 6 0.131 1.040 0.979 0.000 0.744 0.000 0.036
5 63.6 1238.5 0.368 2.871 6 0.125 1.003 0.951 0.000 0.682 0.000 0.065
6 68.2 1509.5 0.434 2.743 6 0.123 0.983 0.955 0.000 0.656 0.000 0.165
7 54.6 875.0 0.322 2.976 6 0.122 1.004 0.976 0.000 0.673 0.000 0.153
8 57.9 1026.0 0.425 2.832 7 0.619 1.014 0.807 0.000 0.750 0.000 0.098
9 64.7 1454.0 1.013 2.516 6 0.125 1.039 0.918 0.000 0.711 0.000 0.266
10 54.1 740.5 0.333 3.212 9 1.025 0.877 0.678 0.000 0.659 0.000 0.099
11 57.8 956.5 0.575 3.062 6 0.125 1.008 0.973 0.000 0.689 0.000 0.176
12 60.1 941.5 0.265 3.542 6 0.122 0.953 0.904 0.000 0.669 0.000 0.153
13 63.5 1385.0 0.447 2.538 6 0.131 0.986 0.943 0.000 0.756 0.000 0.170
14 43.1 1433.0 0.578 1.123 7 0.539 0.233 0.216 0.801 0.607 0.800 0.085
15 64.9 1211.0 0.318 3.075 6 0.121 0.967 0.938 0.000 0.654 0.000 0.088
16 50.6 550.0 0.089 3.831 7 0.584 0.343 0.314 0.000 0.682 0.000 0.064
17 45.0 1783.5 0.635 1.000 8 0.880 0.000 0.000 0.682 0.565 0.757 0.286
18 64.5 1272.0 0.781 2.857 6 0.125 0.985 0.943 0.000 0.697 0.000 0.113
19 59.7 1138.0 0.155 2.715 6 0.135 0.992 0.974 0.000 0.789 0.000 0.093
20 62.8 1215.0 0.513 2.759 7 0.589 0.968 0.845 0.000 0.709 0.000 0.033
21 54.1 1158.5 0.141 2.051 6 0.126 1.024 0.906 0.000 0.733 0.000 0.000

60140 | RSC Adv., 2014, 4, 60135–60143 This journal is © The Royal Society of Chemistry 2014
 View Article Online

Paper RSC Advances

Table 2 Output matrix including nano-descriptors extracted from TEM images in Fig. 1

Aspect Corner Agglomeration

Image Size Surface area Curvature ratio count Circle Rod Dogbone Triangle Square Hexagon state

a 11.9 115.9 0.919 0.997 7.263 0.759 0.095 0.088 0.750 0.561 0.764 0.123
b 16.8 196.9 0.896 1.160 7.012 0.560 0.256 0.232 0.769 0.627 0.780 0.275
c 25.9 497.6 0.577 1.117 6.034 0.321 0.194 0.203 0.750 0.637 0.701 0.225
d 13.9 138.1 0.758 1.165 6.465 0.395 0.328 0.319 0.872 0.657 0.831 0.223
Published on 05 November 2014. Downloaded by INIST - CNRS (Chemistry) on 10/3/2022 6:55:24 PM.

e 27.1 204.8 0.182 1.695 8.851 0.352 0.832 0.796 0.352 0.762 0.355 0.217
f 55.6 1080.1 0.431 2.636 6.619 0.367 0.840 0.767 0.104 0.683 0.112 0.105

cross from them that has a radius close to the radius of the demonstrates this idea more clearly. As it could be seen, by
initial circle (see the circle in Fig. 7a). In contrast, in the case of depicting the histogram of the radius of the crossing circle for
hexagon, if these points were selected close to each other, the either a circle or a hexagon, two different patterns would be
crossing circle would have a big radius compared to when the achieved consisting of just one peak in the case of a circle, but
points were selected far from each other (for example each point two or more peaks in the case of a hexagon. Again, this code was
on a separate side of the hexagon) from which a small circle strongly inuenced by noisy shapes and could not be general-
would be resulted (see the hexagon in Fig. 7b). Fig. 7 ized for most cases. Lots of other similar ideas were tested in

Fig. 9 Shape descriptor comparison between some of the objects in a TEM image; C, R, D, T, S and H respectively refer to circle, rod, dog-bone,
triangle, square and hexagon nano-descriptor values.

This journal is © The Royal Society of Chemistry 2014 RSC Adv., 2014, 4, 60135–60143 | 60141
 View Article Online

RSC Advances Paper

order to attain a very precise general code for distinguishing A Gaussian function was used for this purpose with a normal
between the six shape categorizes. distribution around the number of corners of each template.
Finally, we investigated the idea of “template match” and For example, in the case of a square template, a Gaussian
dening six separate descriptors for different shapes, each distribution with a mean value of four will be considered. If the
explaining the amount of similarity of the images to one of the corner count of the objects in the image were close to four, then
well-dened shape classes. In other words, each shape the output of this function was a value next to one. In contrast, if
descriptor is a coefficient demonstrating the amount of simi- the objects differed from a square and revealed a corner count
larity to each shape type. The more the similarity between an far from four, then the output of the Gaussian function
Published on 05 November 2014. Downloaded by INIST - CNRS (Chemistry) on 10/3/2022 6:55:24 PM.

image and a specic shape class, the higher the value of its expresses this dissimilarity by revealing a value around zero.
corresponding shape descriptor. For example, in the case of a The more the similarity between the corner counts of the
TEM image including triangle shape NPs, the coefficient of the template and the objects, the more the value of the Gaussian
triangle descriptor will be close to one while the other ve shape function.
descriptors will have a value much lower than one (depending 3.3.5. Curvature. The amount of curvature or rigidity of the
on the amount of being similar to/different from a triangle). By corners might be another valuable parameter revealing lots of
developing this idea, we were also able to consider similarities information about the way that particles gather around each
between different shape types. For example the value for the other or how other molecules can attach to their surface. Thus,
circle, the hexagon and the square descriptors will all be high in we dened a descriptor that could approximately report the
the case of a TEM image including hexagon-like NPs, because of amount of curvature of the corners. In order to have a compa-
the similarity between these three shape types. rable value between different images, we scaled this parameter
For performing the template match discussed above, the between zero and one, corresponding to totally keen or totally
algorithm rst builds a template close to the size of each object curved (like circle) corners, respectively. Fig. 8 demonstrates a
in the image but with a specic shape type. For example it starts simple comparison between shapes with different curvatures.
by building circles, then squares, then hexagons and so on till In order to calculate this descriptor, we investigated the idea of
making all six shape templates. The constructed shape subtracting the measured surface area of each object (i.e. the
templates can be seen in S2.† Then it screens through the whole surface area descriptor) from its desired surface area (the
image, rotates the template (if necessary) and measures the mathematical surface area calculated based of the geometry of
amount of overlapping between the templates and the objects. each polygon).
The more the similarity between each shape template and the
objects in an image, means the more matching between them 3.4. Output
and results in a higher coefficient for that specic shape
descriptor. In more details, aer dening a standard template Subsequently, aer running all the above steps and calculating
for each of the six shape classes, a comparison between the the nano-descriptors for each TEM image, the program reveals a
surface areas of all the objects in the image and each template is matrix as an output in which the rows correspond to each TEM
performed separately and the objects are each re-sized in order image and the columns are related to the image nano-
to reach a surface area equal to the templates surface area. By descriptors. The nano-descriptors would be calculated for all
overlaying a specic template on each object, a similarity value the objects in each image (as shown in Table 1); but for
is dened from eqn (5): reporting just one vector containing representative values of
nano-descriptors calculated for each TEM image, a mean value
overlapping area
Similarity ¼ (5) of the nano-descriptors measured for every single object in an
surface area image was then calculated and reported (as shown in Table 2).
As can be seen in Table 1, the rod-shape descriptor column has
The overlapping area is the number of pixels in common greater values than the other shape descriptors, consistent with
between that template and the object (matching area), while the its TEM image which belongs to gold nanorods. For a better
surface area is the number of pixels beneath either the template visual perception of the extracted nano-descriptors, a number of
or the object (which is expected to be the same due to the re- the output shape descriptors are depicted close to their corre-
sizing step performed above). This similarity value lacks from sponding objects in a TEM image shown in Fig. 9. The nal
the distinguishing ability between different shape types and output le in the excel format could be used as an initial dataset
actually needs to be more specied for different shape types for a nano-QSAR study or any other investigation related to
(i.e., two shape templates with same surface areas may result in morphological properties of nanomaterials.
same similarity values, while they belong to different shape
types). Thus, in order to comprise the shape type, a function of
corner count was involved to complete the shape descriptor 4. Conclusion
value as shown in eqn (6): The present contribution introduces a set of surface morpho-
logical nano-descriptors extracted from TEM images of NPs.
Shape descriptor ¼ similarity  f(corner count) (6)
These nano-descriptors including size, surface area, aspect
ratio, curvature, corner count, shape type and agglomeration
state, reveal as much as structural and surface related

60142 | RSC Adv., 2014, 4, 60135–60143 This journal is © The Royal Society of Chemistry 2014
 View Article Online

Paper RSC Advances

information as possible from the TEM images. The represented 8 D. Fourches, D. Pu, C. Tassa, R. Weissleder, S. Y. Shaw,
descriptors can provide informative data required for a thor- R. J. Mumper and A. Tropsha, ACS Nano, 2010, 4, 5703.
ough computational approach on nanomaterials, such as nano- 9 K. Artyushkova, S. Pylypenko, M. Dowlapalli and
QSAR. These descriptors can also be considered in a modeling P. Atanassov, RSC Adv., 2012, 2, 4304.
procedure in order to assign the dominant factors affecting the 10 K. P. Singh and S. Gupta, RSC Adv., 2014, 4, 13215.
toxicity of different types of NPs and consequently, be used to 11 S. Kucheryavski, Chemom. Intell. Lab. Syst., 2011, 108, 2.
further understand the mechanisms which govern the adverse 12 C. Dahl and K. Esbensen, Chemom. Intell. Lab. Syst., 2007, 89,
effects of NPs in biological systems. It must be kept in mind that 9.
Published on 05 November 2014. Downloaded by INIST - CNRS (Chemistry) on 10/3/2022 6:55:24 PM.

the interaction of NPs with biological media is dynamic, and 13 M. R. Hormozi-Nezhad, H. Robatjazi and M. Jalali-Heravi,
therefore it is impossible to monitor the real events taking place Anal. Chim. Acta, 2013, 779, 14.
at the moment that NPs meet proteins, or actually reveal what 14 Y. Wang and Y. Xia, Nano Lett., 2004, 4, 2047.
the cell sees. But on the other hand, predicting the possible 15 K. M. Bratlie, H. Lee, K. Komvopoulos, P. Yang and
toxicity caused by NPs from their initial structure in TEM G. A. Somorjai, Nano Lett., 2007, 7, 3097.
images previous to entering them into the biological media is 16 Y. Xiong, J. P. McLellan, J. Chen, Y. Yin, Z. Y. Li and Y. Xia, J.
noteworthy. Am. Chem. Soc., 2005, 127, 17118.
17 Y. Xia, Y. Xiong, B. Lim and S. E. Skrabalak, Angew. Chem.,
Int. Ed., 2009, 48, 60.
Acknowledgements 18 L. Gou and C. J. Murphy, Chem. Mater., 2005, 17, 3668.
19 Y. Sun and Y. Zia, Science, 2002, 298, 2176.
This work was supported by Sharif University of Technology
20 X. Ye, L. Jin, H. Caglayan, J. Chen, G. Xing, C. Zheng,
Research Council. The authors wish to express their gratitude
V. Doan-Nguyen, Y. Kang, N. Engheta, C. R. Kagan and
for the support of this project. In addition, the authors would
C. B. Murray, ACS Nano, 2012, 6, 2804.
like to thank Ali Foroughi-Pour for his special assistance on
21 Y. Xia, B. Gates, Y. Yin and Y. Lu, Adv. Mater., 2000, 12, 693.
generating the image processing codes.
22 J. K. Beddow and T. P. Meloy, Advanced Particulate
Morphology, CRC Press, Boca Raton, FL, 1977.
References 23 http://gcc.gnu.org, accessed December, accessed July 2013.
24 Willow Garage and Itseez, Intel Russia research center in
1 A. Gajewicz, B. Rasulev, T. C. Dinadayalane, P. Urbaszek, Nizhny Novgorod, http://opencv.org, accessed January 2014.
T. Puzyn, D. Leszczynska and J. Leszczynski, Adv. Drug 25 http://qt-project.org, accessed July 2013.
Delivery Rev., 2012, 64, 1663. 26 S. Suzuki and K. Abe, CVGIP, 1985, 30, 32.
2 B. Rasulev, A. Gajewicz, T. Puzyn, D. Leszczynska and 27 M. Muñoz-Mármol, J. Crespo, M. J. Fritts and V. Maojo,
J. Leszczynski, Towards Efficient Designing of Safe Nanomedicine: NBM, 2014, DOI: 10.1016/j.nano.2014.07.006.
Nanomaterials: Innovative Merge of Computational 28 S. Wang, W. Lu, O. Tovmachenko, U. S. Rai, H. Yu and
Approaches and Experimental Techniques, RSC, 2012, ch. 10, P. C. Ray, Chem. Phys. Lett., 2008, 463, 145.
pp. 220–256. 29 B. D. Chithrani, A. A. Ghazani and W. C. W. Chan, Nano Lett.,
3 R. Todeschini and V. Consonni, Molecular Descriptors for 2006, 6, 662.
Chemoinformatics, Wiley-VCH, 2nd edn, 2009, vol. 41. 30 A. M. Schrand, M. F. Rahman, S. B. Hussain, J. J. Schlager,
4 T. Puzyn, D. Leszczynska and J. Leszczynski, Small, 2009, 5, D. A. Smith and A. F. Syed, Wiley Interdiscip. Rev.:
2494. Nanomed. Nanobiotechnol., 2010, 2, 544.
5 P. R. Gil, G. Oberdörster, A. Elder, V. Puntes and W. J. Parak, 31 G. Bystrzejewska-Piotrowska, J. Golimowski and P. L. Urban,
ACS Nano, 2010, 4, 5527. Waste Manage., 2009, 29, 2587.
6 T. Puzyn, B. Rasulev, A. Gajewicz, X. Hu, T. P. Dasari, 32 S. Pal, Y. K. Tak and J. M. Song, Appl. Environ. Microbiol.,
A. Michalkova, H. M. Hwang, A. Toropov, D. Leszczynska 2007, 73, 1712.
and J. Leszczynski, Nat. Nanotechnol., 2011, 6, 175. 33 R. C. Gonzalez and R. E. Woods, Digital Image Processing,
7 D. Martin, U. Maran, S. Sild and M. Karelson, J. Phys. Chem. Prentice Hall, 3rd edn, 2007.
B, 2007, 111, 9853.

This journal is © The Royal Society of Chemistry 2014 RSC Adv., 2014, 4, 60135–60143 | 60143