Development of Computational Tool for Lung Cancer Prediction Using Data Mining

International Journal of Computer Applications Technology and Research
Volume 5– Issue 7, 417 - 421, 2016, ISSN:- 2319–8656
www.ijcat.com 417
Development of Computational Tool for Lung Cancer
Prediction Using Data Mining
Divya Chauhan
Shoolini University
Solan, Himachal Pradesh
India
Varun Jaiswal
Shoolini University
Solan, Himachal Pradesh
India
Abstract: The requirement for computerization of detection of lung cancer disease arises ever since recent-techniques which involve
manual-examination of the blood smear as the first step toward diagnosis. This is quite time-consuming, and their accurateness depends
upon the ability of operator's. So, prevention of lung cancer is very essential. This paper has surveyed various techniques used by previous
authors like ANN (Artificial Neural Network), image processing, LDA (Linear Dependent Analysis), SOM (Self Organizing Map) etc.
Keywords: Lung Cancer, Classification, Neural Network, SOM, LDA, PCA, Chi-Square, Feature Extraction.
1. INTRODUCTION
1.1 Background
Lung cancer research is one of the most concerning area of
interest in medical field. The early diagnose of the cancer can
help in increasing the mortality rate of humans [1]. Lung cancer
is customarily a contagion which takes place because of the
element linked with unimpeded cell or conveniently progress
in zones present in lung area. According to American Cancer
Society it is approximated that 48,610 persons (27,880 men and
20,730 women) will be detected with and 23,720 men and
women will have high percentage of lung cancer in 2013 only
[2]. In turn, it is part of the even broader set of diseases
disturbing the tuberculosis, Silicosis and Interstitial Lung
Disease (ILD), which are all known as diffuse parenchymal
lung disease (DPLD) [3].
1.2 Data Mining in Medical Field
Data mining is the process in which valuable information is
extracted from the large dataset. It has reached the high growth
over past few years. Due to the usefulness of data mining
approaches in health world, it has become the good technology
in healthcare domain [4]. This realization leads to explosion of
data mining approaches [5]. Medical data mining can exploit
the hidden patterns present in voluminous medical data which
otherwise is left undiscovered. Data mining techniques which
are applied to medical data include association rule mining for
finding frequent patterns, prediction, classification and
clustering. Traditionally data mining techniques were used in
various domains. However, it is introduced relatively late into
the Healthcare domain [6]. Nevertheless, as on today lot of
research is found in the literature. This has led to the
development of intelligent systems and decision support
systems in Healthcare domain for accurate diagnosis of
diseases, predicting the severity of various diseases, and remote
health monitoring. Especially the data mining techniques are
more useful in predicting heart diseases, lung cancer, and breast
cancer and so on.
1.3 CET Images and its Importance in
Medical Field
A CET scanner uses the digital processing to get 3-D image of
an object [7]. A CET scanner emits the radiation from a device
then scans the whole body to get 3-D image [8]. CET scan is
very important as CET scans are a valuable diagnostic tool.
They are able to detect some conditions that conventional XY-
rays cannot, since CET scans can show a "3-D" view of the
section of the body being studied. CET scans are also useful
for monitoring a patient's progress during or after treatment [9].
In this paper, various techniques to detect lung cancer will be
presented along with brief outline of lung cancer detection.
2. RELATED WORK
Hossein GhayoumiZadeh, et al. [10], 2013 represented an
image analysis approach for automated detection,
preprocessing-smoothing, enhancement, segmentation, feature
extraction-morphological and calorimetric and then detection
and categorization of particular cells, particularly the cancer
cells from usual cells is complete.
Lim Huey Nee, et al. [11], 2012 presented the incline scale,
thresholding, morphological operation and division change to
perform cell segmentation. In this paper 50 imageries were
used to test the planned method and the effect showed that the
process hasmanaged to obtain qualitatively good segmentation
consequences.
FauziahKasmin [12], 2012 presented the recognition of blood
disorder is through visual inspection of tiny images of blood
cells. From the recognition of blood disorders, it can lead to
classification of certain diseases related to blood. This
document describes a first round study of developing a
detection of leukemia types using microscopic blood sample
imagery. Here, analyzing through images is very significant as
from images; disease can be detected and diagnosed at earlier
stage. From there, further actions like scheming, monitoring
and prevention of diseases can be done. Imagery is used as they
are despicable and do not need expensive testing and lab
equipment’s. The system will focus on white blood cells
disease, leukemia. The system will use features in microscopic
images and look at changes on texture, geometry, color and
statistical analysis. Changes in these features will be used as a
classifier input. A text appraisal has been done and
Reinforcement Learning is proposed to classify types of
leukemia. A small conversation about issues concerned by
researchers also has been ready.
WaidahIsmail [13], 2011 presented a method for the detection
and classification of blast cells in M3 with others sub types
using computer generated annealing and neural networks. In
this paper, we greater than before our test result from 10 images
to 20 images. We perform Hill Climbing; Simulated Annealing
and Genetic Algorithms for detect the blast cells. As a result,

www.ijcat.com 418
simulated annealing is the “best” heuristic search for detecting
the leukemia cells. From the detection, we perform features
extraction on the blast cells and we classify based on M3 and
other sub-types using neural networks. We received persuasive
result which has targeting around 97% in classify of M3 with
other sub-types. Our consequences are based on real world
image data from a Hematology Department
3. VARIOUS TECHNIQUES FOR
CANCER DETECTION AND
PREVENTION
3.1 ANN (Artificial Neural Network)
An artificial neural network does not shot to be like the thought
process and if/ then sense of the people brain as completed by
an expert system. It mimics exact aspects of the in turn
dispensation and objective sympathetic of the brain by means
of a network of neural link [14]. As a result, a number of writers
record it as a “microscopic”, “white box” structure and a
professional system as a “macroscopic”, “black box” system.
An Artificial Neural Network consists of a huge amount of
simple dispensation elements that are dependable and covered
[15].
Figure 1: Basic Diagram of A.N.N
Inputs: x1, x2, x3, x4……………………….xn
Weights: w1j, w2j, w3j, w4j………………..wnj
TransferFunction: 
Activation Function: α
Output: x1w1j, x2w2jj…………….xnwnj
3.2 LDA (Linear Dependent Analysis)
Linear Discriminant Analysis is utmost commonly utilized as
dimensionality lessening method in the pre-processing stage
for machine learning applications in addition to design-
classification. The main objective is to project a specific dataset
on top of a lower-dimensional space using virtuous class
reparability so as to decrease computational prices as well as
also evade overfitting [16]. The novel linear discriminant was
first designated for a two-class issue, in addition it was then
afterwards widespread as "Multiple Discriminant Analysis" or
"multi-class LDA" through C. R. Rao in the year of 1948.
Linear Discriminant Analysis is "controlled" as well as
calculates the guidelines ("linear discriminants") which would
probably signify the axes that are applied to make the most of
the separation amongst multiple type of classes. Below are the
five basic steps utilized for implementing a LDA technique
[17].
A necessary and sufficient condition for the set of functions:
f1(x), f2(x)...fn(x) to be linearly independent is that
c1 f1(x) + c2 f2(x) + ... + cn fn(x) = 0
only when all the scalars ci are zero.
Figure 2: Basic Diagram of L.D.A
3.3 Self-Organizing Map (SOM)
The Self-Organizing Map is one of the commonly used network
model. It belongs to the learning networks. The Self-
Organizing Map is un-supervised learning method. If Self-
Organizing Map is used for feature extraction then it is called
Self-Organizing Feature Map [18].
Below figure shows that there are 5 cluster units, Yi and 7 input
units, Xi. Clusters are arranged in linear array [19].
Calculate
dimensional mean
Calculate
disseminate metrics
Find eigen vectors
Sort eigen vectors
Utilize eigen matrix
Convert into new
space
Utilizing N spaces End LDA

www.ijcat.com 419
Figure 3: SOM Example
Self-Organizing Map was designed by Kohonen. The SOM has
been useful in many applications. It maps the high dimensional
space to map units for preserve mapping. Neuron units
commonly made lattice onto a plane. Preserving property
means reserving the distance between points. In addition to that
Self-Organizing Map has the capability of generalizing. It
means recognizing the patterns that never met before. The Self-
Organizing Map I 2-D can be represented as following:
𝑌 = { 𝑥…...𝑥 𝑎𝑐𝑤 } (1)
The neurons are connected to adjacent neurons by a relation.
Commonly, the neurons are connected to each other via
rectangular or hexagonal topology. Topologies of neurons are
represented above.
Randomly choose a vector
Determine output node wi.
wi x >= wk
Weight update is given as below:
w(new) = w(old) + υ
3.4 Support Vector Machines (SVM)
Support Vector Machine (SVM) is first and foremost a classier
technique which executes classification tasks through building
hyperplanes in a multi-dimensional space, which divides cases
of different and dissimilar class labels. We can define the
matrix
(H)ij = yiyj(xi ∙ xj), (2)
And introduce more compact notation [20]:
Minimize:
W (a) = -aT
1 + ½ aT
Ha
Subject to:
aT
y = 0
0 ≤ a ≤ C1
Support Vector machines are also called kernel machines and
they have two phases of training:
 Transform input data to high dimensional data.
 Solve quadratic problem [21].
Figure 4: SVM Planar Division
3.5 Genetic Algorithm (GA)
Genetic algorithm is the type of algorithm that is used to solve
both constrained and non-constraint problems based on
selection criteria. Genetic algorithm modifies the new
population and generate new solutions until best solution has
not been reached. From large set of population, genetic
algorithm uses the random chromosomes to make it parent then
make it to produce children [22].
Choose initial population
From left population, select individual chromosomes.
Choose best selected chromosomes
Do crossover
Do repetition
End
Figure 5: Genetic Algorithm Process
3.6 Principal Component Analysis (PCA)
Principal components analysis (PCA) is basically useful for
reducing the number of variables that consists a data set while
retaining the inconsistency in the data and to identify unknown
patterns in the data and to classify them according to how much
of the information, stored in the data, they report for [23].

www.ijcat.com 420
PCA allows calculating a linear alteration that maps
information as of a high dimensional space to a lower
dimensional space [24].
B1 = t11 a11 +………t1n an
B2 = t21 a1 +………..t2an
Linear transformation implied by PCA
The linear transformation RN
-> RK
3 that performs the
dimensionally reduction
B1 Ut
1
B2 Ut
2 (x- |x| ) = Ut
(x- |x| )
3.7 Discreet Wavelet Transform (DWT)
With the enlargement in utilization of internet, communication
of data has turn out to be quiet easy. In contrast with the data
communication in analog form, digital communication offers
us several aids for instance enhanced/superior quality, high
speed, compression of data etc [25]. However, image
acquisition has some shortcomings also, such as the noise
present during transmission. The recognition of the specific
data is one of the significant necessities in the arena of
information transmission, whether it is the transmission of
information/data in military-applications or transmission of
pictures on internet that desires to be safer than before [26].
The wavelet transform has grown pervasively approval in
denoising of image as well as signal processing. It is the
breaking down a specific signal into scaled along with shifted
versions of the unique wavelet. A wavelet is a type of
waveform of efficiently restricted duration which has average
value of zero. And for signals; the identity of the specific signal
is specified through the component of low-frequency.
We can approximate a discrete signal in k2
(X) 1
by
𝑓[𝑏] =
1
√𝑁
∑ 𝑄 𝜙[ℎ0, 𝑗]𝑗 𝜙ℎ0,𝑗[𝑏] +
1
√𝑁
∑ ∑ 𝑄 𝜓[ℎ, 𝑗]𝜓ℎ,𝑗[𝑏]𝑗
∞
ℎ=ℎ0
(3)
Here, 𝑓[𝑏], 𝜙ℎ0,𝑗[𝑏] 𝑎𝑛𝑑 𝜓ℎ,𝑘[𝑏] are discrete functions which
are defined in [0, N-1], to-tally N points. For the reason that the
sets {𝜙ℎ0,𝑗[𝑏]}
𝑗∈𝑋
and {𝜓ℎ,𝑗[𝑏]}
(ℎ,𝑗)∈𝑋2,ℎ ≥ 𝑗0
are orthogonal
to each other. We can simply take the inner product to obtain
the wavelet coefficients:
𝑄 𝜙[ℎ0, 𝑗] =
1
√𝑁
∑ 𝑓[𝑏]𝑏 𝜙ℎ0,𝑗[𝑏] (4)
𝑄 𝜓[ℎ0, 𝑗] =
1
√𝑁
∑ 𝑓[𝑏]𝑏 𝜓 ℎ,𝑗[𝑏] ℎ ≥ ℎ0 (5)
(4) are called approximation coefficients while (5) are called
detailed coefficients.
3.8 Chi Square Test Analysis
The chi-squared one-variable test serve a principle comparable
to the binomial test, excluding that it can be used when there
are more than two categories to the variable. Thus, if you want
to resolve if the numbers of people in each of several categories
vary from some predict values, the chi-squared one-variable
test is proper. The chi-square goodness-of-fit test is a single-
sample non-parametric test, also referred to as the one-sample
goodness-of-fit test [27].
4. CONCLUSION AND FUTURE SCOPE
Lung cancer is one of the major health problems in all over
world. Cancer constitutes 10.3% of medically certified deaths,
which is the most leading cause of death after disease of the
circulatory system, accidents and disease of the respiratory
system. There are over 100 different types of cancer and one of
them is lung cancer. In lung cancer treatment delay results in
high mortality rate. So, this paper has reviewed cancer cell
detection using various methods.
Use of support vector machines will be considered in the future
work as a classification tool. Support Vector Machine (SVM)
is also called Support Vector Networks are supervised learning
models that analyze data and recognize patterns.
5. REFERENCES
[1] Raje, C.; Rangole, J., "Detection of Leukemia in
microscopic images using image processing," in
Communications and Signal Processing (ICCSP), 2014
International Conference on , vol., no., pp.255-259, 3-5
April 2014.
[2] Kalyanmoy Deb, A. Raji Reddy, “Reliable classification
of two-class cancer datausing evolutionary algorithms”,
Elsevier, BioSystems , Vol.72, pp.111–129, 2003.
[3] SubrajeetMohapatra, SushantaShekharSamanta,
DiptiPatra and SanghamitraSatpathi, “Fuzzy based Blood
Image Segmentation for Automated Leukemia
Detection”, IEEE, 2012.
[4] Nimesh Patel, AshotoshMehra, “Automated Detection of
Leukimiausing microscopic images”, Elsevier, Vo. 58,
2015.
[5] Jafar, I., Hao Ying , Shields ,A.F., Muzik , O.
‘Computerized Detection of Lung Tumors in PET/CT
Images’, EMBS 2006, 28th Annual International
Conference of the IEEE Engineering in Medicine and
Biology Society, 2006.
Population Members
Generation=0
Evaluate fitness
Selection
Crossover
Mutation
Stop
Is last generation?

www.ijcat.com 421
[6] Nakao M, Kawashima A, Kokubo M, Minato K.
“Simulating Lung Tumor Motion for Dynamic Tumor-
Tracking Irradiation”. Nuclear Science Symposium
Conference Record, 2007. NSS 2007
[7] E, Donald, “Introduction to Data Mining for Medical
Informatics,” Clin Lab Med, pp. 9-35, 2008.
[8] R. Zhang, Y, Katta, “Medical Data Mining,” Data
Miningand Knowledge Discovery, pp. 305-308, 2002.
[9] Irene M. Mullins et al., “Data mining and clinical data
repositories: Insights from a667,000 patient data set,”
Computers in Biology and Medicine, vol. 36, pp. 1351-
1377, 2006.
[10] Zadeh, Hossein Ghayoumi, SiamakJanianpour, and
JavadHaddadnia, "Recognition and Classification of the
Cancer Cells by Using Image Processing and Lab VIEW,"
International Journal of Computer Theory and
Engineering (2013).
[11] L. H. Nee, M. Y. Mashor, R. Hassan,"White Blood Cell
Segmentation for Acute Leukemia Bone Marrow Images,"
International Conference on Biomedical Engineering
(ICoBE),Penang, Malaysia, 27-28 February 2012.
[12] Kasmin, Fauziah, Anton SatriaPrabuwono, and Azizi
Abdullah, "Detection ofLeukemia in Human Blood
Sample Based On Microscopic Images: A Study, " Journal
of Theoretical & Applied Information Technology46.2
(2012).
[13] Ismail, Waidah, et al. "The detection and classification of
blast cell in Leukaemia Acute PromyelocyticLeukaemia
(AML M3) blood using simulated annealing and
neuralnetworks." (2011).
[14] K.A.G. Udeshani, R.G.N. Meegama, T.G.I. Fernando,
“Statistical Feature-based Neural Network Approach for
the Detection of Lung Cancer in Chest X-Ray Images,”
International Journal of Image Processing (IJIP), Volume
(5), Issue (4) , 2011.
[15] Jinsa , “Lung cancer classification using neural networks
for CT images”, Computer Methods and Programs in
Biomedicine, Volume 113, Issue 1, January 2014, Pages
202-209
[16] J. Yang, D. Zhang, J.-Y. Yang and B. Niu , "Globally
max-imizing, locally minimizing: unsupervised
discriminant projection with applications to face and palm
biometrics" , IEEE transactions on pattern analysis and
machine intelligence , vol. 29 , no. 4 , pp.650 -664 , 2007
[17] Young Tae Lee; Yong Joon Shin; Cheong Hee Park,
"Extending Linear Discriminant Analysis by Using
Unlabeled Data," in Computer and Information
Technology (CIT), 2011 IEEE 11th International
Conference on , vol., no., pp.557-562, Aug. 31 2011-Sept.
2 2011.
[18] C. J. Lin, C.H. Chu, C.Y. Lee, Y.T. Huang, “2D/3D Face
Recognition Using Neural Networks Based on Hybrid
Taguchi Particle Swarm Optimization”, Eighth
International Conference on Intelligent Systems Design
and Application (ISDA), 307-312, DOI :
10.1109/ISDA.2008.286.
[19] Timothy Rumbell, Susan L. Denham, and Thomas
Wennekers, “A Spiking Self-Organizing Map Combining
STDP, Oscillations, and Continuous Learning”, IEEE
Transactions On Neural Networks And Learning Systems,
Vol. 25, No. 5, May 2014
[20] M. Hearst. Support vector machines. IEEE Transactions
on IntelligentSystems, 18 – 28, 1998.
[21] Detection of Lung Nodule Using Multiscale Wavelets and
Support Vector Machine. K.P.Aarthy, U.S.Ragupathy
[22] Man, K.F.; Tang, K.S.; Kwong, S., "Genetic algorithms:
concepts and applications [in engineering design]," in
Industrial Electronics, IEEE Transactions on , vol.43,
no.5, pp.519-534, Oct 1996, doi: 10.1109/41.538609.
[23] Taranpreet Singh Ruprah, “Face Recognition Based on
PCA Algorithm,” Special Issue of International Journal of
Computer Science & Informatics (IJCSI), 2231–5292,
Vol.- II, Issue-1, 2
[24] M. Turk and A. Pentland. Eigenfaces for face recognition,
"Cognitive Neuroscience Journal," vol. 3, no. 1, pp.71-86,
March 1991.
[25] Z. Tufekci and J. N. Gowdy, "Feature extraction using
discrete wavelet transform for speech recognition," IEEE
Inter.Conf. Southeastcon2000, pp. 116-123, April 2000.
[26] K. P. Soman and K. I. Ramchandran, Insight into
Wavelets from Theory to Practice, Printice-Hall of India,
2e, 2005.
[27] Yong Li, "Applications of Chi-Square Test and
Contingency Table Analysis in Customer Satisfaction and
Empirical Analyses," in Innovation Management, 2009.
ICIM '09. International Conference on , vol., no., pp.105-
107, 8-9 Dec. 2009

Development of Computational Tool for Lung Cancer Prediction Using Data Mining

More Related Content

What's hot (20)

Similar to Development of Computational Tool for Lung Cancer Prediction Using Data Mining (20)

More from Editor IJCATR (20)

Recently uploaded (20)

Development of Computational Tool for Lung Cancer Prediction Using Data Mining