See discussions, st ats, and author pr ofiles f or this public ation at : https:www .researchgate.ne tpublic ation322499164 [620538]

See discussions, st ats, and author pr ofiles f or this public ation at : https://www .researchgate.ne t/public ation/322499164
ABCD Rules for Melanoma Melanoma Segmentation Skin cancer ABCD rule
Article · Januar y 2018
CITATIONS
0READS
1,005
2 author s, including:
Some o f the author s of this public ation ar e also w orking on these r elat ed pr ojects:
mycomple te thesis View pr oject
a part of my ieee tr ansaction p aper View pr oject
Satheesha T y
C M R Univ ersity
37 PUBLICA TIONS    65 CITATIONS    
SEE PROFILE
All c ontent f ollo wing this p age was uplo aded b y Satheesha T y on 15 Januar y 2018.
The user has r equest ed enhanc ement of the do wnlo aded file.

300 World Journal of Techn
ology, Engineering and Resea
r
ch, Volume 2, Issue 1 (2017)
300
–
308

Contents available at

WJTER

World Journal of Technology, Engineering
and Research

Journal Homepage:

www.wjter.com

ABCD Rules
f
or Melanoma

T Y Satheesha
a*
,

Sushmita N S
b

,

Gururaj Murtugudde
c

a
Associate Professor,Dept. of ECE, NCET, Bangalore
, India

b
Assistant Professor
, Dept. of ISE, NCET, Bangalore,India

c
Professor
and Head, Dept. of ISE, SVCE, Bangalore
, India

Keywords

A B S T R A C T

M
elanoma

Segmentation

S
kin cancer

ABCD rule

Early Detection of melanoma remains the key factor lowering
mortality from the cancer. Melanoma is a difficult challenge
diagnostically and therapeutically.

Diagnosis of the putative
primary lesion is difficult. This difficulty increases when the
primary lesion has undergone complete regression.

Melanoma
is a skin cancer which is early for effective treatment. During
the las
t decade, new computer
–
based technologies have
improved diagnostic sensitivity and specificity and may result
in optimizing lesion selection for biopsy and pathology
review.

.So, the aim of this scheme to improve two schemes in
order to develop an interfac
e that can assist diagnosis
dermatologists in the diagnostic phase. In this paper, a
preprocessing technique is performed to separate noise and
unnecessary structure from the image. Therefore, automatic
segmentation is introduced to locate the skin lesion.

Then,
feature extraction followed by ABCD rule to make the
diagnosis through the calculation of the TDV score. The best
results were obtained by using 50 images which contains
apprehensive melanoma skin cancer. The best fully automated
method is TDV. By e
valuating results we can reason out that
the accuracy of the system is better than previous work.

© 2017 WJTER All rights reserved.

301

I.

INTRODUCTION

Cancer arises through genetic alterations in cells that are subject to further selection.
During this guesstionary process, control over genomic integrity becomes indecorously
compromised, resulting in loss of tumor
–
suppressor genes and activation of
ontogenesis.
As a result of these changes, cancer cells differ substantially from their regular
counterparts. The incidence of the melanoma has shown histrionic increases over the past
thirty five years. Melanoma represents a small subset, it is the most e
xtreme cutaneous
neoplasm. In

the United

States,

in

2014

there

w
ere

an

estimated

76,100

new cases of
skin

cancer

diagn
o
sed

and

9,710

deaths

from

this disease[1]. The incidence of
melanoma increases by 4.1 percent per year, faster than any other
malignancy a skin
cancer will develop in one in five people during the lifetime. Skin cancer is divided into
two major part which are melanoma and non melanoma skin cancers. Early detection of
this cancer can help its cure ability. Melanoma arises from the

cancerous growth in the
pigmented spots. Dermatologists can diagnose melanoma in about 80% of the cases
according to BCD process.So; the digital dermatoscopy could give dermatologist a closer
look at apprehensive lesions in an early step. Based on images
obtained by digital
dermoscopy, our conclusive aim is to develop an aided
–
diagnostic system for the
identification of early stage melanomas. This would enable supervised classification of
melanocytic lesions. The melanoma detection process is composed of f
ive steps that are
the preprocessing, the segmentation, the post
–
processing, the feature extraction by the
ABCD rule and finally the classification based on the Total Dermatoscopic Value
calculation (TDV). The ABCD rule presents the following features whic
h are the
Asymmetry, the Border Irregularity, the Color variation and the Diameter. This latter must
be greater than 6mm in the case of melanoma.

Earlier detection and therapy also leads to
decreased cost of therapy. An estimated 90% of costs spent on mela
noma therapy in the
United States are related to those with advanced disease.7 Therefore, a significant savings
in health care cost can be realized if melanoma incidence could be lowered or if it could be
detected in an earlier, more easily treated phase.
However, skin cancer is one of the few
cancers in which the cause of most cases is known
—
excessive sun exposure,8 and it
would be expected that lowering an individual’s ultraviolet (UV) exposure should
similarly improve later risk of developing skin cancer
. However, other factors influence
the risk of developing and dying from this cancer.

This paper is organized as follows. In Section 2 we present a brief overview of previous
related work. In Section 3, we illustrate our approach to melanoma skin cancer de
tection.
Experimental results and analysis are presented and discussed in Section 4. Finally, we
present a brief conclusion.

1.

Melanoma Skin Cancer Detection

The digital dermatoscopic method is developed for image analysis. Among these
methods, the colo
r enhancement which is based on the color correction with HSV model.
In this approach, linear regression models are constructed for each channel, which allows
automatic adjustment of the hue and saturation [2]. A second approach is based on a
scheme of sta
ndardization based on two steps; removing the color variations and
strengthening contrast images [3]. Although this approach has in some cases good results,
it remains limited since lesions classification depends on other parameters like asymmetry
and bord
er irregularity.

Other approaches developed in the literatures are based on feature
extraction. The first describes a new method based on graphs which will enable networks
to extract pigments from images of dermoscopy [4].The second implements the

302

classification of skin lesions based on the characteristics of ”granularity” [5]. The third
proposes a classification with white areas on the images [6]. However ,multi
–
scale
growing [7], fuzzy c means based on anisotropic mean shift[14],

Based on the expe
riment
using 30 samples images, the accuracy of the system proposed by Fatichah et al. [8] is
85% but the accuracy using fuzzy region growing proposed by Fatichah et al. [9] is 86.6%.
Despite region
–
based approaches have difficulties, especially, when the
lesion or the skin
region is textured, this category is widely used by researchers. For that, in this paper we try
to improve region growing algorithm for the detection of lesion.

II.

ARCHITECTURE SYSTEM OF MELANOMA SKIN CANCER
DETECTION

Dermatoscopic
images used in this work are collected with the Department of
Dermatology, University Hospital Hedi Chaker Sfax. They are coded on 24 bits storing the
three color components r (red), g (green) and b (blue). These rgb images are 640×480
pixels in size. Unfo
rtunately, the image acquisition leads to the permanent presence of
illumination. The existence of shadows, reflections on wet tissue or fatty report more
difficulties.

2.1

Preprocessing

The image preprocessing is an major step in image diagnosis. It is used to correct
defects illumination, eliminating noise and small spots and enhance. it starts by using a
median filter aiming at cleaning the image by eliminating certain defects. Th
is spatial filter
is based on calculating for each pixel the median of the gray levels of the neighbors pixels.
Since each component of the image can be viewed as a grayscale image, we can apply this
filter on each component separately. Then, we performed
a morphological closing aiming
at eliminating all artifacts such as hair. According to the images provided, it was found
empirically that setting a 5×5 median filter and a disk with a diameter of 3 gives the best
results. Fig.1 shows the influence of the m
edian filter and the morphological closing on the
image. Finally, we must improve the color between the lesion and healthy skin. Since our
goal is to extract the lesion apart, we improve the contrast of the image by adjusting
intensity values after convert
ing the image to grayscale. The influence of the Histogram
adjustment on the grayscale image is shown in Fig. 2.

2.2

Segmentation

To do a classification of skin lesions, and particularly to distinguish melanoma from
benign lesions, we must begin by isol
ating the lesion from healthy skin that surrounds each
color image using a segmentation methodology. The detection of this skin lesion is a
critical problem in dermatoscopic images because the transition between the lesion and the
surrounding skin is diffi
cult to detect accurately. For this, segmentation method chosen
must be precise. In this paper, the segmentation process used is region growing. This
segmentation technique begins with a pixel as the seed point. Then it aggregates the pixels
according to t
wo criteria: homogeneity and adjacency [10]. These two manual methods,
based on a random selection of the threshold and the germ, are not recommended. In fact,
it will have a bad Influence on the segmentation result. In this work, we tried to adapt a
simpl
e method of region growing for our application to detect skin lesions. To optimize
the result of the segmentation, the proposed method is based on an automatic choice of the
threshold and the seed point from which and at each level of growth we add the

303 nei
ghboring pixels having similar properties. Since only one region must be segmented
and the lesion is always darker than the normal skin, the seed pixel will be automatically
selected as the pixel with the gray level which is darker and more similar to its
neighbors.
This characteristic is achieved by calculating the difference between the pixel P and the
average of a 5×5 window of its neighbors. This method is illustrated in the diagram
presented in Fig. 3.

(a)

(b)

Fig. 1
: Preprocessing: (a) Original image (b) image given after applying the median filter and the morphological closing

(a)

(b)

Fig.
2
: Influence of the histogram adjustment on grayscale image:(a) Filtered image converted to grayscale (b) Influence of the

histogram

adjust

Fig.
3
: Seed point selection method
.

304

Fig. 4
: Triangle method

Fig. 5
: Choice of the triangle method: a) An image presenting a

Lesion b) Corresponding histogram.

The threshold is calculated using the following steps: Finding the optimal threshold T
of the image automatically us
ing the triangle method which allows constructing a line
between the histogram peak and the farthest end. The threshold is the maximum distance
between the line and the histogram as shown in Fig. 4. The choice of this method is based
on the fact that all t
he images showing a lesion are still presented as a dark spot
surrounded by a lighter area (skin) as shown in Fig. 5(a). The histogram of such an image
is always given as presented in Fig. 5(b).Calculating the average of the region where

is
the Number of

pixels belonging to the region:

=
∑
values

of

the

pixels

belonging

to

the

regions

The threshold is calculated by measuring the difference between the average and the
optimal threshold:

ℎ

ℎ

=
|

−

|

(
1
)

Then, the growth process used the following steps:

(1)Adding the seed point neighboring pixels belonging to a window of 3×3 pixels to a list
o
f neighbors.

(2) Calculating the difference between the gray
–
level of each neighboring pixel and the
region average.

(3) Adding the neighboring pixel of the seed point, with the smallest difference, in the
region and mark it as a seed pixel.

(4) Updating t
he average Avr, the threshold and the gray level difference.

(5) Deleting the pixel from the list of neighbors

(6) Returning to step 1 each time the difference found in (4) is less than the threshold.

Hence, the approach is to grow the region around the st
arting pixel. This aggregation does
not stop until the

305 difference exceeds a certain threshold:

|
P
(
i
,
j
)
−

Avr
|
>
=

threshold

(
2
)

2.3

Post
Processing:

To remove isolated pixels within the region of the lesion by the segmentation, a further
step of morphological closing is

produced. In our work, the disc diameter of this morphological closing is set to 5 as it
gives the best result.
Finally, the resulting image is a binary mask used to separate the
lesion from the skin. It is then superimposed with the original image. Fig. 6 illustrates this
step.

2.4

ABCD Feature Extraction:

The ABCD rule used by dermatologists in recognition p
rocess of skin lesions to
assess the risk of malignity of a pigmented lesion. This method is able to provide a more
objective and reproducible diagnostic of skin cancers in addition to its speed of
calculation. It is based on four parameters: (Asymmetry)
concerns the result of evaluation
of lesions asymmetry, (Border) estimates the character of lesions border, (Color) identifies
the number of colors present in the investigated lesion, and (Diameter).

2.4.1
Asymmetry:

The asymmetry of the lesion is
the most important indicator characterizing the
malignancy. This characteristic is considered in terms of form and color. To facilitate the
task of calculating the score of asymmetry, we translated the lesion to the center of the
image. Now it is easy to d
ivide the image according to the major axis and the minor axis of
the lesion .To measure the asymmetry in terms of form, we are basing our calculation on
the asymmetry index presented as follows:

(a)

(b) (c) (d)

Figure
6
: The result of the segmentation: (a) original image (b)lesion detected by region growing (c)
lesion detected after post
–
processing (d) segmented image
.

=

∆

(
3
)

S is the

major axis and the minor axis is the area of non
–
overlapping zone and NM is the
total area of the lesion. To measure the asymmetry in terms of color, our calculations are
based on the histograms of the three RGB components of each part of the lesion and
the
chi
–
square distance.

,

=

(
e
1
(
x
)

−

e
2
(
x
)
)
2
e
1
(
x
)

+

e
2
(
x
)

(
4
)

where

1

and

2

are the two histograms of size

.

Finally, the asymmetry score N is calculated as the average between the asymmetry
score in terms of form and that in terms of color.

306

2.4.2.
Border Irregularity
:

Generally, the borders of the benign lesions are clearly defined. The irregularity

of the
border can usually report a cancer during growth and propagation. For an evaluation, the
lesions are divided into eight sectors as presented in Fig. 7. Within each sector,a strong
and sharp cut pat
–

tern at the periphery receives a score of 1. In c
ontrast, a gradual
approach, indistinct within each segment has a score of 0. Thus, the maximum irregular
border score is 8, and the minimum one is 0. This calculation is based on the Euclidean
distance and the standard derivation in each sector.

=

(

2
−

1
)

+
(

2
−

1
)

(
5
)

where j2 and j2 are the coordinates of the center of the lesion. i1 and j1 are the coordinates
of

pixel x.

=

(
6
)

With k is the number of pixels in the edge belongin
g to the considered area.

is the
Euclidean distance between the center of the lesion and the pixel x. Then the standard
deviation is calculated for each sector with the following equation:

=
(
1


(

−

)
2
)

(
7
)

Where n is the number of elements in the sector.

Fig.
7
: Calculating the border score: B=3

Fi
g. 8
: Color score calculation for a lesion:D=2(light brown and dark brown)

Fig. 9
: The scales used to measure the actual diameter

2.4.3. Colors:

Melanomas are characterized by the presence of six different colors, namely, white,
red, light brown,

dark brown, blue
–
gray and black. For each present color, we add 1 to the
score. When the six colors are present, the maximum score is 6. The minimum score is
1.To verify the existence of each color in the lesion, the image was converted from the

307 RGB color

space to the CIE Labcolor space because the distance between two colors in the
RGB color space does not reflect the ”difference” perceived by the naked eye contrary to
the CIE Labcolor space. This has shown in Fig. 8. This difference can be calculated usi
ng
the Euclidean distance delta B
:

∆

=

(

1
−

2
)

+
(

1
−

2
)

+
(

1
−

2
)

(
8
)

where C1, m1 and n1 are the components of the CIE Lab colo
rimetric space of the desired
color and C2, m2 and n2 those of each pixel of the image.

2.4.4.
Diameter:

Since the dermatoscope gives the possibility of tracing the marks on the image during
acquisition shown in Fig. 9, we are based on these scales to
calculate the actual diameter
of the lesion.

2.4.5. Classification:

After obtaining the values of the four characteristics, the Total Dermatoscopic value
(TDV) is calculated based on the following formula where each one of the presented
characteristics i
s multiplied by a weighting factor:

=
1
.
3
∗

+
0
.
1
∗

+
0
.
5
∗

+

.
5
∗

(
9
)

III.

EXPERIMENTAL RESULTS

The dermatoscopic image suspected as melanoma is used as
input data. To evaluate the
performance of our developed system, this research use dataset of 40 dermatocospic
images. The measures used for the evaluation are accuracy, sensibility and specificity
measures presented as follows:

=
(

)
(

)

(10)

To evaluate the performance of experiment we ha
ve used melanoma and Non
–
Melanoma
dermatocospic images, shown in Fig. 10. Based on the experimental results, the small
value of color variation causes, in some cases false diagnosis. To improve this accuracy,
we must tuning the calculation of this paramete
r. We can use for example a color variation
model for dermoscopic images.

(a)

(b)

Fig. 10
: Skin Cancers

; (a) Non Melanoma; (b) Melanoma

0
0.2
0.4
0.6
A7_15B
A7_35B
N7_18B
N6_43B
Melanoma and Non
Melanoma
param A
param B
param C
param D
0
0.2
0.4
0.6
1
2
3
4
ABCDRULE
ABCDRULE

308

(a)

(
b)

Fig. 13: (a)
Melanoma and non
–
Melanoma graph Based on the parameter. (b) Graph based on ABCD rule

IV.

CONCLUSION:

The objective of the proposed method has involved the detection of melanoma based on region growing
segmentation and the ParamA ParamB
ParamC ParamD rule used for the detection of malignancy of
pigmented skin lesion, this as shown in Fig. 13
(a)
. The use of region growing in our method based on
automatic selection of the seed pixel and the threshold ensures the best results and avoids over
lap between
the lesion and healthy skin. The result of the segmentation was then used in the next step by the ABCD rule
in order to classify pigmented skin lesion as benign or malignant, this as shown in Fig. 14
(b)
. There are three
diagnosis that are used
in this research i.e. melanoma, suspicious, and benign skin lesion. The experiment
uses samples of dermatoscopic images containing a lesion that is suspicious melanoma skin cancer
.

REFERENCES

[1] S. Stechschulte, C. Ricotti, C. J. Cockerell, (2008)
Advances in diagnostic testing for
skin cancer, TOUCH BRIEFINGS pp.73
–
76.

[2] R.Lucas, T.McMichael, W.Smith, B.Armstrong, (2006) Solar ultravio
–

let
radiation.global burden of disease from solar ultraviolet ra
–

diation, Environmental Burden
of Disease. Se
ries 13 117.

[3] D.Rigel, J.Russak, R.Friedman, (2010) The evolution of melanoma di
–
agnosis, A
Cancer Journal for Clinicians 60 301316.

[4] F. Xie, A. Bovik, Automatic segmentation of dermoscopy im
–

ages using self
–
generating neural networks seeded by gene
tic al
–

gorithm, Pattern Recognition,
ELSEVIER (2013) 1012
–
1013.

[5] H. Iyatomi, M. E. Celebi, G. Schaefer, M. Tanaka, (2011) Automated color calibration
method for dermoscopy images, Computerized

Medical Imaging and Graphics, pp. 89
–
98.

[6] G. Schaefer, M
. I. Rajab, M. E. Celebi, H. shi Iyatomi, (2011) Colour and contrast
enhancement for improved skin lesion segmenta
–

tion, Computerized Medical Imaging and
Graphics, pp.99
–
104.

[7] M. Sadeghi, M. Razmara, T. K. Lee, M. Atkins, (2011) A novel method for dete
ction
of pigment network in dermoscopic images using graphs, Computerized Medical Imaging
and Graphics, pp.137
–
143.

[8] W. V. Stoecker, M. Wronkiewiecz, R. Chowdhury, R. J. Stanley,J. Xu, A. Bangert, B.
Shrestha, D. A. Calcara, H. S. Rabinovitz,M. Olivier
o, F. Ahmed, L. A. Perry, R. Drugge.,
(2011), Detection ofgranularity in dermoscopy images of malignant melanoma using color
and texture features, Computerized Medical Imaging and Graphics, pp.144
–
147.

[9] A. Dalal, R. H. Moss, R. J. Stanley, W. V. Stoeck
er, K. Gupta, D. A. Calcara, J. Xu, B.
Shrestha, R. Drugge, J. M. Mal
–

ters, L. A. Perry, (2011), Concentric decile segmentation
of white and hypopig
–

mented areas in dermoscopy images of skin lesions allows
discrimination of malignant melanoma, Computeriz
ed Medical Imaging and Graphics,
pp.148
–
154.

[10] C. Grana, G. Pellacani, R. Cucchiara, S. Seidenari, (2003), A new al
–

gorithm for
border description of polarized light surface micro
–

scopic images of pigmented skin
lesions, IEEE Transactions on Medical
Imaging 22 , 959964.

[11] P. Rubegni, A. Ferrari, G. Cevenini, D. Piccolo, M. Burron, (2001), Differentiation
between pigmented spitz naevus and melanoma by digital dermoscopy and stepwise
logistic discriminant analysis, Melanoma Research 11, 3744.

[12] H
. Zhou, G. Schaefer, M. C. nad F. Lin, T. Liu, (2011), Gradient vector with mean
shift for skin lesion segmentation, Computerized Medical Imaging and Graphics 35 ,
121127.

View publication statsView publication stats