Lung Cancer Detection using Machine Learning

International Journal of Trend in Scientific Research and Development (IJTSRD)
Volume 4 Issue 6, September-October 2020 Available Online: www.ijtsrd.com e-ISSN: 2456 – 6470
@ IJTSRD | Unique Paper ID – IJTSRD33659 | Volume – 4 | Issue – 6 | September-October 2020 Page 1399
Lung Cancer Detection using Machine Learning
Harpreet Singh1, Er. Ravneet Kaur2
1Research Scholar, 2Assistant Professor,
1,2Baba Banda Singh Bahadur Collage, Fatehgarh Sahib, Punjab, India
ABSTRACT
Modern three-dimensional (3-D) medical imaging offers the potential and
promise for major advances in science and medicine as higher fidelity images
are produced. Due to advances in computer aided diagnosis and continuous
progress in the field of computerized medical image visualization, there is
need to develop one of the most important fields within scientific imaging.
From the early basis report on cancer patients it has been seen that a greater
number of people die of lung cancer than from other cancers such as colon,
breast and prostate cancers combined. Lung cancer arerelatedtosmoking(or
secondhand smoke), or lessoftentoexposureto radonor other environmental
factors that’s why this can be prevented. But still it is not yet clear if these
cancers can be prevented or not. In this research work, approach of
segmentation, feature extraction and Convolution Neural Network (CNN)will
be applied for locating, characterizing cancer portion.
KEYWORDS: Lung Cancer, Image Processing, Machine Learning, K-means, Gray-
Level Co-Occurrence Matrix (GLCM)
How to cite this paper: Harpreet Singh |
Er. Ravneet Kaur | "LungCancerDetection
using Machine Learning" Published in
International Journal
of Trend in Scientific
Research and
Development(ijtsrd),
ISSN: 2456-6470,
Volume-4 | Issue-6,
October 2020,
pp.1399-1402, URL:
www.ijtsrd.com/papers/ijtsrd33659.pdf
Copyright © 2020 by author(s) and
International Journal ofTrendinScientific
Research and Development Journal. This
is an Open Access article distributed
under the terms of
the Creative
CommonsAttribution
License (CC BY 4.0)
(http://creativecommons.org/licenses/by/4.0)
INTRODUCTION
The image processing is a technique which is used for the
enhancement of unprocessed pictures or images captured
from different cameras from different origins. With the help
of image processing, the significant data can be retrieved
efficiently. In the past decades, various methods have been
evolved in image processing techniques for the extraction of
complicated information in an effective manner. Image
processing approach is widely utilized in army, clinical and
investigational areas [1]. Some associations also use image
processing approach for simplifying the manual workload
and execution of positive actions. The image processing is
applied inside numerous applications inclusively in order to
improve the optical description of pictures. For the
preparation of pictures, different calculations are
implemented as well.Image processingalsoknownasDigital
Image Processing (DIP) comprises both visual and analog
image processing which involves different methods. Image
acquisition is also termed as imaging [2]. The visual and
digital image processing can be performed with the help of
imaging. This technique utilizes several domains like
computer graphics for the generation of pictures [3]. This
technique also provides assistance in the manipulation and
modification of pictures. The picture or image is analyzed
with the help of processor hallucination or computer vision.
In lung cancer, anomalous cells multiply and grow in the
form of a tumor. The lymph fluid which environs lung tissue
carries the cancerous cells from lungs to blood. The lymph
streams via lymphatic vessels. These lymph fluid drainsinto
lymph nodules deployed in the lungs and in the middle
region of chest area. The growth of lung tumor always
carried out towards the middle area of chest due to the
regular flow of lymph fluid towards the chest center.Whena
cancer cell leaves its origin area, metastasis happens [4].
This cancerous cell now goes towards a lymph nodule or to
different body part with the help of blood flow. The prime
lung tumor is a kind of cancer which originates from the
lung. The compilation of lung pictures for the creation of
data sample in the initial step. In Image Enhancement,image
is processed and smoothened. This process enhances the
picture quality and also eliminates noise from the picture.
Thus, this process offers superior key for the digital image
processing [5]. Image enhancement is an important pillar of
image pre-processing.Image Segmentationallocatesa digital
picture into different sections like sets of pixels also
recognized as super-pixels. The key objective of this process
is the alteration of a picture demonstration in an easier
investigative manner. Picture sectioning is utilized for
identifying the location of objects, limits and borders in
pictures. In this process, a label is assigned to each pixel in a
picture and thus the pixels with the identical label share
definite features [6]. In Feature Extraction feature plays an
extremely significant character. Different image
preprocessing approaches such as binarization,
thresholding, normalization, masking approach etc. are
implemented on the sampled picture before the attainment
of features. Various classifiers are used for performing the
classification on the basis of retrieved characteristics. SVM
(Support Vector Machine) is a classificationalgorithmthatis
based on optimization theory. As it maximizes the margin it
is also known as a binary classifier. All the data points of an
individual class are separated by the best hyperplane, this
can be identified through the classification provided by
IJTSRD33659

International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
support vector machine[7]. The main aim of Naïve
Bayesclassifier is the implementation of a strategy where
future objects are assigned to a group in the presence of a
pattern of objects for every class. The applied variable
vectors are demonstrated with the help of future entities.
Decision Tree Classifier is considered as non-parametric
supervised learning techniques and used for categorization
and deterioration [8]. The main aim of this approach is the
development of a model for the accurate prediction of an
intended variable in accordance with several key variables.
K-Nearest neighbor classifier depends on the learning by
similarity. The ndimensional arithmeticqualitiesareutilized
for the description of training sets.
Literature Review
Amir R. et.al (2019) reviewed the development of inclusive
molecular description of tumor lump[9].Afundamental role
was played by the ailment biomarkersfortheearlydetection
and indulgent of tumor analysis. This work summarized the
speedy development of biosensor equipments for lung
tumor biomarkers discovery. More expansion in
nanobiotechniques in association with nanobiocomposite
and miniaturization approaches would considerably
improve existing biodiagnostic capability for sensing tumor
biomarkers in genuine organic models with sufficient
compassion, accuracy, sturdiness and price efficiency.
Guobin Z., et.al (2019) presented a serious evaluation of the
CADe scheme for automated lung cancer recognition with
the help of CT descriptions for summarizing the existing
developments [11].Thesemechanismsincludedinformation
attainment, preprocessing, lungimagesegmentation,nodule
recognition and false positive diminution. A brief summary
of superior nodule detection methods and classifiers was
also provided on the basis of understanding, false positive
value and other constrained data. After different studies it
was evaluated that Computer aided diagnosis(CAD)scheme
was essential for timely lung malignancy recognition.
JingS. et.al (2019) proposed a novel approachofmicroscopic
hyper spectral imaging for the identification of ALK affected
lung tumor [11]. In this approach, a household microscopic
hyper spectral imagingschemewasutilizedforcapturing the
pictures of five classes of lung tissues. In results, Group ALK
obtained more relative proportion of cytoplasm of 77.3%
than Group ALK-positive. The investigational outcomes
related to quantitative scrutiny and ethereal curves
demonstrated that the treatment of ALK affectedlungtumor
implemented with low concentrated medicines would be
developed towards the ALK non-affected lung tumor.
Moritz S., et.al (2018) estimated the usefulness of machine
learning for lung tumor recognition in FDG-PET imaging in
the scenario of ultralow amount PET scan [12]. In the
absence of pulmonary tumor, the recital of artificial neural
network on selective lung cancer patients was examined.
The sensitivity rate of 95.9% and 91.5% was attained by the
artificial neural system for lung cancer detection. The deep
learning approach for detecting lungcancerprovidedAUC of
.989 for standard dose images, 0.983 for reduced dose
images, and 0.970 for PET3.3% rebuilding. It was also
suggested that more advancements in this technique could
enhance the accurateness of lung tumor testing approaches.
Madhura J, et.al (2017) presented a review of noise
reduction approaches for lung cancer diagnosis [13]. It was
stated that lung cancer was a solemn ailment which caused
due to the abnormal growth of cells in the lung tissues.
Amongst all the other kinds of tumors, the lung tumor was
identified as the most incident cancer. Therefore,thiscancer
became the reason of several cancer patients’ deaths. This
review work also described the different kinds of noises
present in the pictures, techniques for the attainment of
apparent pictures and noise elimination methods. A brief
review on the existing noise elimination methods was also
provided in this paper.
Suren M., et.al (2017) stated that CT images could be used
for the lung tumor recognition. The major objective of this
study was the evaluation of different automated
technologies, investigation of existing finest method,
recognition of its restrictions and disadvantages and the
projection of a decisive system with several advancements
[14]. For this purpose, the lung tumor recognition
approaches were classified on the basis of their lung cancer
analyzing accurateness. In every stage, these lung cancer
recognition methods were examined and their restrictions
and disadvantages were considered. It was identified that
different lung cancer detection techniques showed different
precision. Some techniques showed least precision rate
while some techniques showed good precision rate for lung
cancer detection but no technique showed 100% precise
lung cancer detection.
Research Methodology
This research work is related to lung cancer detection from
the CT (Computed Tomography) scan image using image
processing techniques. The proposed methodology has the
four phases for the lung cancer localization and
characterization.
Figure 1: Proposed Flowchart
Input CT scan lung image dataset which collected
from different internet sources
Pre-process the input image in which images will be
denoised using filtering technique
Apply threshold-based segmentation technique called
outu’s segmentation to remove skull part from the
image
Apply GLCM to extract 13 textural features of input
MRI image
Train the model using CNN approach for the
categorization of tumor and non-tumor portion from
the image
Test the trained model for the tumor detection and
analyze performance in terms of accuracy, precision,
recall and execution time

Following are the various phases of the lung cancer
detection: -
1. Pre-processing: -
The pre-processing is the first phase in which CT scan image
is taken as input. The technique of image de-noising will be
applied which will remove noise from the input image. The
output of this stage is an enhanced image. This is one of the
most crucial stages in lung cancer detection.
2. Segmentation:
In the second phase, the approach of region-based
segmentation will be applied which will segment the similar
and dissimilar regions from the CT scan image. The otsu’s
segmentation technique is applied forthesegmentation.The
sectioned picture attained from thresholding comprises
several benefits like lesser storage space, speedy
dispensation velocity and easiness in exploitation in
comparison with gray level picture that generally includes
256 steps. In the presented work, a gray scale picture is
utilized for thresholding process. In this process, rgbpicture
is converted into binary picture. The obtained picture is in
the form of black and white.
3. Feature Extraction: -
The feature extraction is the third phase, in which GLCM
algorithm will be applied for the feature extraction of the CT
scan image. In this step, the GLCM algorithm is applied for
the feature extraction. The GLCM algorithm will extract the
textural features of the input image. The GLCM algorithm
extracts 13 features of the image for the tumor detection
Energy =
Entropy=
Contrast=
4. Classification: -
In the last phase, the approach of CNN will be applied which
can categorize and localize the cancer part .All the data
points of an individual class are separated by the best hyper
plane, this can be identified through the classification
provided by CNN. In the CNN the largestthebesthyperplane
is described by the largest margin between the two classes.
There are no interior data points when there is maximum
width between the slabs parallel to the hyper plane which is
also known as margin. The maximum margin in hyper plane
is separated by the CNN algorithm.
Experimental Results
The proposed research is implemented in MATLAB and the
results are evaluated by comparing proposed and existing
techniques in terms of various performance parameters.
Figure 2: Accuracy Analysis
As shown in figure 2, the accuracy of the existing system
which is SVM approach is compared with the proposed
approach which is CNN approach. The system is tested on
different number of images and it is analyzed that CNN gives
better results as compared to SVM approach.
Figure 3: Sensitivity Analysis
As shown in figure 3, the sensitivity of the existing system
different number of images and it is analyzed that CNN give
best results as compared to SVM approach.
Fig 4: Specificity Analysis

As shown in figure 4, the specificity of the existing system
different number of images and it is analyzed that CNN give
better results as compared to SVM approach.
Conclusion
For lung cancer detection image processing is used. There
are three steps for the detection of cancer nodule. To detect
the presence of cancer nodule CT scan images are used.
Further the pre-processing composed of two processes.
Image enhancement and image segmentation are that two
processes. The imagesegmentationprocessaimstopartition
the image into meaningful format and identify the object or
relevant information from thedigital image.Theoutputfrom
the segmentation process is applied tothefeatureextraction
stage. Features such as area, perimeter and irregularity are
found out in feature extraction. On the basis of the extracted
features the abnormality in lung are found out by the cancer
cell identification module. The approach of GLCM and CNN
are implemented in this work for localizing and classifying
cancer part from the CT scan image. The proposed approach
is implemented in MATLAB and results are analyzed in
terms of accuracy. It is analyzed that the proposed approach
gives optimized results up to 8 percent.
References
[1] Anjali Kulkarni, Anagha Panditrao, “Classification of
Lung Cancer Stages on CT Scan Images Using Image
Processing”, 2014 IEEE International Conference on
Advanced Connnunication Control and Computing
Technologies (lCACCCT)
[2] Anam Tariq, M. Usman Akram and M. Younus Javed,
“Lung Nodule Detection in CT Images using Neuro
Fuzzy Classifier”, 2013 Fourth International
Workshop on Computational Intelligence in Medical
Imaging (CIMI), Pages: 49 – 53
[3] Christian Donner, Naoufel Werghi, Fatma Taher,
Hussain Al-Ahmad, “Cell Extraction from Sputum
Images for Early lung Cancer Detection”, 2012 16th
IEEE Mediterranean Electrotechnical Conference,
Pages: 485 – 488
[4] Fatma Taher, Naoufel Werghi and HussainAl-Ahmad,
“Comparison of Hopfield Neural Network and Mean
Shift algorithm in Segmenting Sputum Color Images
for Lung Cancer Diagnosis”, 2013 IEEE 20th
International Conference on Electronics,Circuits,and
Systems (ICECS), Pages: 649 – 652
[5] Janee Alam, Sabrina Alam, Alamgir Hossan, “Multi-
Stage Lung Cancer Detection and Prediction Using
Multi-class SVM Classifier”, 2018, International
Conference on Computer, Communication, Chemical,
Material and Electronic Engineering (IC4ME2)
[6] Moffy Vas, Amita Dessai, “Lung cancer detection
system using lung CT image processing”,2017,
International Conference on Computing,
Communication, Control and Automation (ICCUBEA)
[7] N. Werghi, C. Donner, F. Taher, H. Alahmad,
“Segmentation of Sputum Cell Image for Early Lung
Cancer Detection”, IET Conference on Image
Processing (IPR 2012), 2012, Pages: 1 – 6
[8] Shanhui Sun, Christian Bauer, and Reinhard Beichel,
“Automated 3-D Segmentation of Lungs with Lung
Cancer in CT Data Using a Novel Robust Active Shape
Model Approach”,IEEETRANSACTIONSON MEDICAL
IMAGING, VOL. 31, NO. 2, FEBRUARY 2012
[9] Amir Roointan, Tanveer Ahmad Mir, Shadil Ibrahim
Wani, Mati-ur-Rehman, Khalil Khadim Hussain, Bilal
Ahmed, Shugufta Abrahim, Amir Savardashtaki,
Ghazaal Gandomani, Molood Gandomani, Raja
Chinnappan, Mahmood H Akhtar, “Early detection of
lung cancer biomarkers through biosensor
technology: a review”,2019, PBA 12266
[10] Guobin Zhang, Shan Jiang, Zhiyong Yang, Li Gong,
Xiaodong Ma, Zeyang Zhou, Chao Bao, Qi Liu,
“Automatic nodule detection for lung cancer in CT
images: A review”, 2018, CBM 3128
[11] Jing Songa, Menghan Hua, Jiansheng Wanga, Mei
Zhoua,b, Li Suna, Song Qiua, Qingli Lia, Zhen Sun,
Yiting Wanga, “ALK positivelungcanceridentification
and targeted drugs evaluation using microscopic
hyperspectral imaging technique”,2019, Infrared
Physics & Technology
[12] Moritz Schwyzer, Daniela A. Ferraro, Urs J.
Muehlematter, Alessandra Curioni-Fontecedro,
Martin W. Huellner, Gustav K. von Schulthess, Philipp
A. Kaufmann, Irene A. Burger, Michael Messerli,
“Automated Detection of Lung Cancer at Ultralow
dose PET/CT by Deep Neural Networks - Initial
results”, 2018, LUNG 5827
[13] Madhura J, Dr. Ramesh Babu D R, “A Survey on Noise
Reduction Techniques for Lung Cancer Detection”,
2017, International Conference on Innovative
Mechanisms for Industry Applications
[14] Suren Makajua, P. W. C. Prasad,AbeerAlsadoona,A.K.
Singhb, A. Elchouemi, “Lung Cancer Detection using
CT Scan Images”, 2017, 6th International Conference
on Smart Computing and Communications, ICSCC

Lung Cancer Detection using Machine Learning

More Related Content

What's hot (20)

Similar to Lung Cancer Detection using Machine Learning (20)

More from ijtsrd (20)

Recently uploaded (20)

Lung Cancer Detection using Machine Learning