An artificial neural network approach for detecting skin cancer

TELKOMNIKA, Vol.17, No.2, April 2019, pp.788~793
ISSN: 1693-6930, accredited First Grade by Kemenristekdikti, Decree No: 21/E/KPT/2018
DOI: 10.12928/TELKOMNIKA.v17i2.9547  788
Received April 17, 2018; Revised June 8, 2018; Accepted June 29, 2018
An artificial neural network approach for detecting
skin cancer
Sugiarti
1
, Yuhandri
2
, Jufriadif Na’am*
3
, Dolly Indra
4
, Julius Santony
5
1,4
Faculty of Computer Science, Universitas Muslim Indonesia, South Sulawesi, Indonesia
2,3,5
Computer Science Faculty, Universitas Putra Indonesia YPTK, Padang, 25221, Indonesia
*Corresponding author, e-mail: sugiarti@umi.ac.id
1
, yuyu@upiyptk.ac.id
2
,
jufriadifnaam@gmail.com
3
, dolly.indra@umi.ac.id
4
, juliussantony@yahoo.co.id
5
Abstract
This study aims to present diagnose of melanoma skin cancer at an early stage. It applies feature
extraction method of the first order for feature extraction based on texture in order to get high degree of
accuracy with method of classification using artificial neural network (ANN). The method used is training
and testing phases with classification of Multilayer Perceptron (MLP) neural network. The results showed
that the accuracy of test image with 4 sets of training for image not suspected of melanoma and melanoma
with the lowest accuracy of 80% and the highest accuracy of 88.88%, respectively. The 4 sets of training
used consisted of 23 images. Of the 23 images used as a training consisted of 6 as not suspected of
melanoma images and 17 as suspected melanoma images.
Keywords: artificial neural network, first-order feature extraction, melanoma, multilayer perceptron
Copyright © 2019 Universitas Ahmad Dahlan. All rights reserved.
1. Introduction
Melanoma is one type of skin cancer which serious enough and could not be handled
perfectly, so it can be lead to death, disability and requires large medical expenses [1]. There
are many method used in detecting Melanoma and one of the successful one is based on image
processing, where it can be used to help the process of diagnosing skin cancer [2-5]. Now is
done dermatologist to diagnose wounds skin cancer using two imaging techniques that are
macroscopic with accuracy of 65-80% and dermatocopic with accuracy of 75-97% [6]. Boundary
detection for skin cancer image is early stage and very important for diagnose support system
for skin diseases based on computer [7]. Accuracy of diagnosis is determined by accuracy of
boundary detection of the skin cancer lesions. So that the image that experience impaired
become easier at the interpretation, then the image need to be done image manipulation
process known as image processing [8-12]. Early detection of skin cancer will be easier in
handling and treatment based on types of the cancer suffered. Experience for dermatological
expert showed the difficult to distinguish melanoma from other pigment wound on the skin, such
as typical wound and not typical (harmless) [10]. One of the solutions in detecting skin cancer is
based on computational intelligence techniques.
Numerous computational intelligence (CI) techniques have emerged motivated for
solving many real world problems by real biological systems, namely, artificial neural networks
(NNs) [13-22] , evolutional computation, simulated annealing and swarm intelligence, which
were enthused by biological nervous systems, natural selection, the principle of
thermodynamics and insect behavior, respectively. Despite the limitations associated with each
of these mentioned techniques, they are robust and have been applied in solving real life
problems in the areas of science, technology, business and commerce including the
works of [23-25].
Based on the above explanation, this study aims to present diagnose of melanoma skin
cancer at an early stage. It applies feature extraction method of the first order for feature
extraction based on texture in order to get high degree of accuracy with method of classification
using artificial neural network (ANN). The method used is training and testing phases with
classification of Multilayer Perceptron (MLP) neural network.

TELKOMNIKA ISSN: 1693-6930 
An artificial neural network approach for detecting skin cancer (Sugiarti)
789
2. Research Method
In this section, the system design for skin cancer detection program using artificial
neural network (ANN) with input data in the form of color image dermatoscopic is presented.
The diagram for whole the process mentioned above can be illustrated in Figure 1 as follows:
Figure 1. The diagram for skin cancer detection
with this program expected to help in the early detection of melanoma. System design stages
are described as follows:
1) Select image Input
2) Preprocessing by changing RGB original image into gray image.
3) Possibility distribution image quality improvement with distribution possibility algorithm.
4) Segmentation with threshold method to separate object with the background.
5) Feature extraction based on texture by using first-order characteristic extraction method.
6) Image classification with Artificial Neural Networks.
7) The final result of diagnose type of melanoma skin cancer and non-melanoma.
3. Results and Discussion
Gray scaling is a technique used to change color image i.e. RGB into grey level (from
black to white) with this modification then matrix to compose the image that previous 3 matrix
will change to just 1 matrix [10]. Preprocessing to improve the quality of the image using
Distribution Possibility Algorithm (DPA) aims to improve image quality using fuzzy logic
approach using 5 parameters i.e. α, β1, γ, β2 and max. Distribution Possibility Algorithm
described as follows :
1) Step-1 : Initialization Parameter
Set β1 = ( min + mean )/2
Set β2 = ( max + mean )/2
2) Step-2 : Fuzzification
For all pixels (i,j) in the image
a) if (data(i,j)>= min) && (data(i,j) <β1)
new-grey(i,j) = 2*(((data(i,j)-min)/(means-min))^2)
b) if (data(i,j)>= b1) && (data(i,j) < means)
new-grey(i,j) = 1-(2*(((data(i,j)-means)/(means-min))^2))
c) if (data(i,j)>= means) && (data(i,j) < b2)
new-grey(i,j)=1-(2*(((data(i,j)-means)/(maks-means))^2))
d) if (data(i,j)>= b2) && (data(i,j) < maks)
new-grey(i,j)=2*(((data(i,j)-means)/(maks-means))^2)
3) Step-3 : Modification
Fuzzy-data2(i,j) = new-grey(i,j)^2
4) Step-4: Defuzzification
For all pixels (i, j) in the quality image (i,j) = fuzzydata2(i,j)*data(i,j)

 ISSN: 1693-6930
TELKOMNIKA Vol. 17, No. 2, April 2019: 788-793
790
Next, thresholding process to grayscale image that aim to produce binary image that
mathematically can be written as follows:
𝑔(𝑥,𝑦) = {
0, (x, y) ≥ 𝑇
1, (𝑥, 𝑦) < 𝑇
(1)
The purpose of Otsu Thresholding algorithm using to perform image segmentation with
a differentiated way into 2 classes, that is background (the value is set by 0) and the object (the
value is set by 1) use certain level as a boundary. The steps to perform thresholding Otsu
is [13]:
1) From the image that have increased its image, taken number of pixels at i level, which is
represented by n total of the number of pixels defined by N = n1 + n2 + .... + nn. Then
obtained a p-value of the number of pixels divided by the total of the number of pixels.
𝑝𝑖 =
𝑛 𝑖
𝑁
, 𝑝𝑖 ≥ 0 , ∑ 𝑝𝑖 = 1𝐿
𝑖=1 (2)
2) Afterwards performed division pixels into 2 classes, that is C0 as background and C1 as
object.
ω0 = Pr(C0) = ∑ 𝑘
𝑖=1 Pi=ω(k) (3)
ω1 = Pr(C1) = ∑ 𝑘
𝑖=𝑘+1 Pi=1 − ω(k) (4)
3) Then to get mean value is represented by formula as follows:
µ (k) = ∑ 𝑖𝑝𝑖
𝐿
𝑖=1 (5)
µ 𝑇 = µ(𝐿) ∑ 𝑖𝑝𝑖
𝐿
𝑖=1 (6)
4) To get class variance is defined by formula as follows:
𝜎 =𝑊
2
𝜔0 𝜎 +0
2
𝜔0 𝜎1
2
(7)
𝜎 =𝐵
2
𝜔0(µ0
− µ 𝑇
)2
+ 𝜔1(µ1
− µ 𝑇
)2
=𝜔0 𝜔1(µ1
− µ0
)2
(8)
𝜎 = ∑ (i − µ 𝑇
)2
𝐿
𝑖=1
𝑇
2
𝑃𝑖 (9)
5) Afterwards the search is done for maximum value from all values visible, and threshold value
is determined using the average of the discovery of variance values. If not found then
threshold value is equal to zero (0.0).
𝜎 (𝑘∗) =𝐵
2
max 𝜎 𝐵
2
(𝑘), for 1 ≤ 𝑘< L (10)
𝑆∗
={𝑘; 𝜔0 𝜔1 = 𝜔(𝑘)[1 − 𝜔(𝑘)] > 0, 𝑜𝑟 0 < 𝜔(𝑘) < 1} (11)
The first-order feature extraction is the characteristic collection methods which are
based on characteristic of image histogram. The histogram is shows probability the emergence
of value from the degree of gray pixels in an image. For the values in the histogram is generated
then it can be counted several parameters on the first-order characteristic, among other things:
mean, skewness, variance, kurtosis, and entropy [14].
1) Mean (µ) shows the dispersion size of an image.
µ = ∑ fn p(fn)𝑛 (12)

791
2) Variance ( σ2) shows the elements variance in the histogram in an image.
σ
2=
(fn-µ)
2
p (fn) (13)
3) Skewness (σ
3
) shows the level relative skewness on histogram curve in an image.
𝑎3 =
1
𝑎3
∑ (f 𝑛 − µ)3
p(f 𝑛)𝑛
(14)
4) Kurtosis (σ
4
) shows the level of skewness relative on histogram curve in an image.
𝑎4 =
1
𝑎4
∑ (f 𝑛 − µ)4
p(f 𝑛)𝑛
− 3 (15)
5) Contrast shows the size of the deployment (moment of inertia) elements of the image
matrix.
   



1
0,
2
,
G
ii
jipjiContrast (16)
6) Standard Deviation is used to measure the average contrast of image intensity [26].
𝜎 = ∑ 𝑃𝑖,𝑗(i − µ𝑖
)2
𝑁−1
𝑖,𝑗=0
𝑖
2
(17)
7) Smoothness is used to measure the relative smoothness of the intensity of the image.
R Value = 0 for image with a constant intensity, while R Value that approaching 1 for image
with the scattered intensity.
R= 1-1/(1+ σ
2
) (18)
8) Entropy (H) shows the size of irregularity shape in an image.
H = − ∑ p(f 𝑛)4
log p(f 𝑛)𝑛
(19)
Artificial neural network is an information processing paradigm that is inspired by
biological neural cell system, just like the brain that processes information. In this study, we are
using a classification method Artificial Neural Networks (ANN) Multilayer Perceptron (MLP). The
ANN-PB training algorithm that formulated by Rumelhart in 1986, briefly as follows:
1) Initialization of weights, which can be done at random.
2) The calculation of activation value, each neural calculates activation value from input that it
receives. Activation value input layer is identity function. In the hidden layer and the output of
value activation is calculated by activation function.
3) Adjustment of weights, where in adjustment of weights is affected by magnitude of error
value output targets and the output value of the current network
4) Iterations will continue to be done until certain minimum error criteria are fulfilled.
In feature extraction stage, image based on texture with first-order characteristic
extraction method using 6 parameters i.e. contrast, variance, standard deviation, kurtosis, mean
and smoothness obtained varying parameter values. Next process is stages of identification
that it performed with training, testing and system performance. Images used 23 images, that
they were ordinary wounds image are not cancer and melanoma image.
The training process with Artificial Neural Network (ANN) with the image of the train is a
combination of the image is non melanoma and melanoma that generates parameter values.
Testing on test image with 4 sets of testing with combination training image and test image
as shown in Tables 1-4. From Table 1, the combination of set I (50%) equals to 50%; from
Table 2 the combination of set II (60%) equals to 40%; from Table 3 the combination of set III
(70%) equals to 30%; and from Table 4 the combination of set IV (80%) equals to 20%.

 ISSN: 1693-6930
TELKOMNIKA Vol. 17, No. 2, April 2019: 788-793
792
Table 5 presents the comparison of accuracy obtained from Tables 1-4. Table 5 based
on the testing have been done with 4 times set testing. It can be obtained that the accuracy of
the test image non melanoma and melanoma images. Results of detection had the low
accuracy of 80% and the highest accuracy of 88.88%.
Table 1. Testing Set I
Image Recognition
Result of
Testing
Non_03.JPG
Non-
Melanoma
TRUE
Non_04.JPG
Non-
Melanoma
TRUE
melanoma_01.JPEG Melanoma TRUE
melanoma_06.JPEG
Non-
Melanoma
FALSE
Table 2. Testing Set II
Image Recognition
Result of
Testing
Non_01.JPG
Non-
Melanoma
TRUE
Non_04.JPG Melanoma FALSE
Non_07.JPG
Non-
Melanoma
TRUE
Table 3. Testing Set III
Image Recognition
Result of
Testing
Non_07.JPG
Non-
Melanoma
TRUE
Non_08.JPG
Non-
Melanoma
TRUE
melanoma_14.JPEG
Non-
Melanoma
FALSE
Table 4. Testing Set IV
Image Recognition
Result of
Testing
Non_01.JPG
Non-
Melanoma
TRUE
Non_03.JPG
Non-
Melanoma
TRUE
Table 5. Comparison of the Accuracy
Set Comparison (%) Accuracy (%)
I 50 : 50 85,71
II 60 : 40 81.81
III 70 : 30 88.88
IV 80 : 20 80.00
4. Conclusion
In the system of skin cancer detection, performed feature extraction based on texture
with first-order feature extraction that produces varying parameters value so we need an
appropriate classification method. The development of future studies should use samples with
higher numbers so can be used as a comparison in order to produce a better accuracy.
References
[1] Morton DL, Wen DR, Wong JH, Economou JS, Cagle LA, Storm FK, Foshag LJ, Cochran AJ. 1992.
Technical details of intraoperative lymphatic mapping for early stage melanoma. Archives of surgery.
1992; 127(4): 392-399.
[2] Seidenari S, Pellacani G, Giannetti A. Digital videomicroscopy and image analysis with automatic
classification for detection of thin melanomas. Melanoma Research. 1999; 9(2):163-172.
[3] Barrett HH, Swindell W. 1996. Radiological imaging: the theory of image formation, detection, and
processing. Academic Press. 1996: 2.
[4] Ercal F, Chawla A, Stoecker WV, Lee HC, Moss RH. Neural network diagnosis of malignant
melanoma from color images. IEEE Transactions on biomedical engineering. 1994; 41(9): 837-845.

793
[5] Ganster H, Pinz P, Rohrer R, Wildling E, Binder M, Kittler H. 2001. Automated melanoma
recognition. IEEE transactions on medical imaging. 2001; 20(3): 233-239.
[6] H Sucianto. Analysis of image filtering using median. Gunadarma university library. 2007.
[7] Golston JE, Moss RH, Stoecker WV. Boundary detection in skin tumor images: an overall approach
and a radial search algorithm. Pattern Recognition. 1990; 23(11): 1235-1247.
[8] Erçal F, Moganti M, Stoecker WV, Moss RH. Detection of skin tumor boundaries in color images.
IEEE Transactions on Medical Imaging. 1993; 12(3): 624-626.
[9] Xu L, Jackowski M, Goshtasby A, Roseman D, Bines S, Yu C, Dhawan A, Huntley A. Segmentation
of skin cancer images. Image and Vision Computing. 1999; 17(1): 65-74.
[10] Sumijan, Madenda S, Harlan J, Wibowo EP., Hybrids Otsu method, Feature region and Mathematical
Morphology for Calculating Volume Hemorrhage Brain on CT-Scan Image and 3D Reconstruction.
TELKOMNIKA Telecommunication, Computing, Electronics and Control. 2017;
15(1): 283-291.
[11] Na`am J, Harlan J, Madenda S, Santony J, Suharianto C. Detection of proximal caries at the molar
teeth using edge enhancement algorithm. International Journal of Electrical and Computer
Engineering IJECE, 2018; 8(5): 3259-3266.
[12] Na`am J, Harlan J, Nurcahyo G W, Arlis S, Sahari, Mardison, Rani LN. Detection of Infiltrate on Infant
Chest X-Ray. TELKOMNIKA Telecommunication Computing Electronics and Control.
2017;17(4):1943-1951.
[13] Amaliah B, Arieshanti I, Fatichah C, Dewi SN, Widyanto MR. Klasifikasi Voted Perceptron Untuk
Identifikasi Melanoma. In Seminar Nasional Aplikasi Teknologi Informasi (SNATI). 2011.
[14] Chiroma H, Abdulkareem S, Herawan T. Evolutionary Neural Network model for West Texas
Intermediate crude oil price prediction. Applied Energy. 2015; 142: 266-273.
[15] Nawi N M, Khan A, Rehman MZ, Chiroma H, Herawan T. Weight optimization in recurrent neural
networks with hybrid metaheuristic Cuckoo search techniques for data classification. Mathematical
Problems in Engineering, 2015; 2015.
[16] Abubakar AI, Khan A, Nawi, NM, Rehman MZ, Wah TY, Chiroma H, Herawan T. Studying the Effect
of Training Levenberg Marquardt Neural Network by Using Hybrid Meta-Heuristic Algorithms. Journal
of Computational and Theoretical Nanoscience. 2016; 13(1): 450-460.
[17] Chiroma H, Abdul-kareem S, Ibrahim U, Ahmad I G, Garba A, Abubakar A, Hamza MF, Herawan T.
Malaria severity classification through Jordan-Elman neural network based on features extracted
from thick blood smear. Neural Network World. 2015; 25(5): 565.
[18] Husaini N A, Ghazali R, Nawi NM, Ismail LH, Deris MM, Herawan T. Pi-Sigma Neural Network For A
One-Step-Ahead Temperature Forecasting. International Journal of Computational Intelligence and
Applications. 2014;13(04): 1450023.
[19] Nawi NM, Rehman MZ, Aziz MA, Herawan T, Abawajy JH. Neural network training by hybrid
accelerated cuckoo particle swarm optimization algorithm. In International Conference on Neural
Information Processing. Springer International Publishing. 2014: 237-244
[20] Nawi NM, Rehman MZ, Aziz MA, Herawan T, Abawajy JH. An Accelerated Particle Swarm
Optimization Based Levenberg Marquardt Back Propagation Algorithm. In International Conference
on Neural Information Processing. Springer International Publishing. 2014: 245-253.
[21] Chiroma H, Abdul-Kareem S, Muaz SA, Khan A, Sari EN, Herawan T. Neural Network Intelligent
Learning Algorithm for Inter-related Energy Products Applications. In International Conference in
Swarm Intelligence. Springer International Publishing. 2014: 284-293.
[22] Ghazali R, Husaini NA, Ismail LH, Herawan T, Hassim YMM. The performance of a Recurrent HONN
for temperature time series prediction. In 2014 International Joint Conference on Neural Networks
(IJCNN). IEEE. 2014: 518-524.
[23] Husaini NA, Ghazali R, Ismail LH, Herawan T. A Jordan Pi-Sigma Neural Network for Temperature
Forecasting in Batu Pahat Region. In Recent Advances on Soft Computing and Data Mining.
Springer International Publishing. 2014: 11-24.
[24] Nawi NM, Khan A, Rehman MZ, Herawan T, Deris MM. CSLMEN: A New Cuckoo Search Levenberg
Marquardt Elman Network for Data Classification. In Recent Advances on Soft Computing and Data
Mining. Springer International Publishing. 2014: 173-182.
[25] Chiroma H, Abdulkareem S, Abubakar AI and Herawan T. Kernel functions for the support vector
machine: comparing performances on crude oil price data. In Recent Advances on Soft Computing
and Data Mining. Springer International Publishing. 2014: 273-281.
[26] Lasisi A, Ghazali R, Herawan T. Comparative performance analysis of negative selection algorithm
with immune and classification algorithms. In Recent Advances on Soft Computing and Data Mining.
Springer International Publishing. 2014: 441-452.

An artificial neural network approach for detecting skin cancer

More Related Content

What's hot (20)

Similar to An artificial neural network approach for detecting skin cancer (20)

More from TELKOMNIKA JOURNAL (20)

Recently uploaded (20)

An artificial neural network approach for detecting skin cancer