Lung Cancer Image- Feature Extraction and Classification using GLCM and SVM Classifier

Abstract

Lung cancer is the second most causing cancer when compared to all the other cancers. According to WHO (World Health Organization) lung cancer contributes about 14 per cent among all the cancers. Therefore, early detection and treatment is very much required. Now-a-days, image processing techniques are playing a major role in early detection of disease which is very helpful in further treatment stages. These techniques help in detecting the abnormality of the tissues-tumor in target cancer images. In this research, the proposed methodology is majorly carried out in five phases. In phase one lung cancer and non-lung cancer, images are collected from the lung cancer database. In phase two preprocessing is done by using the Median filter. Median filter is chosen as it preserves the edges i.e, sharp features are preserved. In Phase three, segmentation of the target image is done using Fuzzy C Means. Fuzzy C Means Clustering is chosen as it gives better performance than K-means Clustering. In phase four, the features are extracted using GLCM (Gray Level Co-occurrence Matrix). GLCM have high discrimination accuracy and less computational speed. In phase five, these extracted features are given to SVM classifier for classification of lung cancer from normal lung. The SVM classier achieved accuracy of 96.7% for detecting and classification of lung cancer.

Full text

International Journal of Innovative Technology and Exploring Engineering (IJITEE)
ISSN: 2278-3075, Volume-8, Issue-11, September 2019
2211
Published By:
Blue Eyes Intelligence Engineering
& Sciences Publication
Retrieval Number K20440981119/2019©BEIESP
DOI: 10.35940/ijitee.K2044.0981119

Abstract: Lung cancer is the second most causing cancer when
compared to all the other cancers. According to WHO (World
Health Organization) lung cancer contributes about 14 per cent
among all the cancers. Therefore, early detection and treatment is
very much required. Now-a- days, image processing techniques
are playing a major role in early detection of disease which is very
helpful in further treatment stages. These techniques help in
detecting the abnormality of the tissues-tumor in target cancer
images. In this research, the proposed methodology is majorly
carried out in five phases. In phase one lung cancer and non-lung
cancer, images are collected from the lung cancer database. In
phase two preprocessing is done by using the Median filter.
Median filter is chosen as it preserves the edges i.e, sharp features
are preserved. In Phase three, segmentation of the target image is
done using Fuzzy C Means. Fuzzy C Means Clustering is chosen
as it gives better performance than K-means Clustering. In phase
four, the features are extracted using GLCM (Gray Level
Co-occurrence Matrix). GLCM have high discrimination
accuracy and less computational speed. In phase five, these
extracted features are given to SVM classifier for classification of
lung cancer from normal lung. The SVM classier achieved
accuracy of 96.7% for detecting and classification of lung cancer.
Keywords: Median filter, FCM (Fuzzy C Means clustering),
GLCM (Gray Level Co-occurrence Matrix), , SVM (Support
Vector Machine).
I. INTRODUCTION
Now a days cancer is becoming the most common reason for
human deaths, in which lung cancer occupies the second
position. These cancer cells are produced due to the abnormal
growth of cells in the human body. These cells can originate in
any part of the body thereby leading to the cause of cancer [1].
Lung cancer can be caused in two ways. The first way is
spreading of cancer cell in the lungs itself which is known as
primary lung cancer. The second way is the spreading of
cancer cells from other parts of the body to the lungs which is
known as secondary lung cancer. According to WHO, 85 per
cent of the lung cancer caused is of type NSCLC (Non-small
cell lung cancer) and the remaining 15 per cent caused is due
to SCLC (Small cell lung cancer) [2]. Detection of cancer at
Revised Manuscript Received on September 03, 2019
* Correspondence Author
R.Ankita , M.Tech Scholar, Embedded Systems, Department of ECE,
GRIET, Hyderabad. Email:ankitha16995@gmail.com
Ch. Usha Kumari*, Professor, Department of ECE, GRIET, Hyderabad,
Telangana, India. Email:ushakumari.c@gmail.com
Mr. Mohd. Javeed Mehdi, Assistant Professor. Department of ECE,
GRIET, Hyderabad, Telangana, India. Email:mmjaveedk@gmail.com
N.Tejashwini is M.Tech Scholar , Embedded Systems, Department of
ECE, GRIET, Hyderabad
Dr T. Pavani, Professor, Department of ECE, GRIET, Hyderabad,
Telangana, India
late stages is the main reason for increasing cancer rates
leading to a greater number of death counts.
These death counts can be decreased by detecting the cancer
cells at early stages itself. Image processing techniques can be
used in early detection of this cancer cells and thus can help in
treatment at initial stages.
The proposed methodology has five phases for the
classification of lung cancer. In phase one, the required data is
collected from the database https://www.kaggle.com/datasets.
In phase two removal of noise is done by using the Median
filter. In phase three, the taken input images are segmented
using FCM (Fuzzy C Means) clustering. After segmentation,
in phase four, features are extracted using GLCM (Gray Level
Co-occurrence Matrix). These extracted features are used in
phase five for classification purpose which is carried by SVM
(Support Vector Machine).
Figure: Proposed Methodology for Cancer Detection
II. RELATED WORKS
In this paper [3], detection of cancer was carried out with
the help of ANN (Artificial Neural Network back-
propagation). From the lung cancer database, a total of 50
images were considered and these images were divided into
two groups- cancer and non-cancer images. The features were
extracted with the help of GLCM (Gray Level Co-occurrence
Matrix) and these were given to ANN. The cancer was
detected and 80% of accuracy was obtained.
In this paper [4][20], two segmentation methods were used
for early detection of lung cancer. The first method was
Hopfield Neural Network (HNN) and the second method was
Fuzzy C-Mean (FCM) algorithm. With the help of HNN,
nuclei and cytoplasm regions were extracted successfully.
HNN was preferred as FCM failed in extracting these features
with accuracy.
This paper [5][19], proposed an automatic detection of
lung cancer by suing CT scan
images. While diagnosing lung
cancer, lung nodule detection
Lung Cancer Image- Feature Extraction and
Classification using GLCM and SVM Classifier
R. Ankita, Ch.Usha Kumari, Mohd Javeed Mehdi, N. Tejashwini, T. Pavani

Lung Cancer Image- Feature Extraction and Classification using GLCM and SVM Classifier
2212
Retrieval Number K20440981119/2019©BEIESP
DOI: 10.35940/ijitee.K2044.0981119
Published By:
Blue Eyes Intelligence Engineering
& Sciences Publication
plays an important role. Therefore, FCN (Fully Convolutional
Network) was selected for gaining more accuracy. First, FCN
was used for segmentation and then lung nodules were
detected. The proposed method was able to detect the lung
nodules with 100 percent accuracy.
In this paper [6][18], CAD (Computer Aided Diagnosis)
was preferred for detection of lung nodule. Here the disease
can be detected after extracting the lung nodule region from
the taken input CT scan images. Firstly, threshold value is set
and then optimal thresholding is performed. After selection
and application of this threshold value, region growing with
connectivity analysis is applied to extract the cavity region.
In this paper [7], a novel technique was proposed for easing
the segmentation technique. During analysis of lung cancer
many images are considered and thus segmentation process
becomes difficult for these huge number of data. Therefore,
BNMB (Bayesian Network Model) was selected which is
graphical model based on probability. This graphical
approach helped the images resulting from
over-segmentation.
In this paper [8], input color images were first converted
into grey scale images as processing of grey scale image is
easier than that of the color images. histogram equalization
was then applied to the images for obtaining the sharp borders
which can help in further analysis as it highlights the borders
thereby increasing the contrast of the images. thresholding
was applied and features were extracted based on the pixel
value. By applying Baack Propagation Network, an overall
efficiency of 78 percent was achieved.
In this paper [9], proposed methodology used CT
(Computerized Tomography) scan images. The methodology
has several steps such as pre-processing where the noise of the
images is removed. The second step is image segmentation
where over segmentation is removed using Marker Controlled
Watershed Segmentation. The segmented image was then
applied to Binarization and masking technique. Watershed
achieved more accuracy when compared with Threshold
which is 85.27 percent.
In this paper [10][17], lung cancer was detected using
X-ray images. Median filter was used for removing noise. In
segmentation stage, K-mean clustering and fuzzy C-mean is
used for feature extraction. These extracted features were
given to SVM classifier for classification. MATLAB tool
with GUI where normality and abnormality of lungs can be
shown easily.
In this paper [11], detection of lung cancer was done using
CT images. The proposed system uses median filter and the
wiener filter for preprocessing stage to remove the noise in the
image. Features were extracted using GLCM technique and
these extracted features were given to classifier. Here two
classifiers were used- SVM and BPNN. The accuracy
obtained by SVM classifier is 96.32 percent which is more
when compared with that of BPNN which is 83.07 percent
respectively.
III. METHODOLOGY
The process of classifying lung cancer is shown in Figure 1.
The entire process of lung cancer detection is divided into five
phases: image collection, image Preprocessing, image
segmentation, feature extraction and classification.
A. Image Collection
The cancer effected and non-cancerous images are
collected from https://www.kaggle.com/datasets. Al the CT
(Computed Topography) scan images are in Dicom format.
Here CT scan images are selected because CT scan images
are an advancement to X-ray images. CT scan images are a
combination of X-ray images observed from different angles.
CT scan images also have an advantage of high resolution,
less noise. In this research, a total of 100 images are collected
from the internet of which 70 are cancerous effected images
and 30 are non-cancerous images.
B. Image Preprocessing
All the images collected are grayscale images. Grayscale
images are easy to process when compared to that of RGB
images. These collected images may contain noise, which can
be of any type. So, filters are applied for removal of this noise.
In this research, Median filter which is a non-linear digital
filter is utilized for removing of noise. The advantage of
median filter is, during removal of noise from the image the
edges are preserved without any disturbance but under certain
conditions only.
C. Image Segmentation
Segmentation plays an important role in classification of
diseases. In this process the images are portioned into
multiple segments. These segments combinedly cover the
entire region or image. In this research, Marker-controlled
watershed segmentation is utilized for segmentation purpose
[12]. The main advantage of this technique is it provides fast
segmentation results than the other techniques and moreover
it helps in eliminating over-segmentation formed in watershed
transformation technique [13].
D. Fuzzy C-means Clustering
Fuzzy C-means Clustering algorithm can be used for both
classification and for clustering algorithm. It is used to find
out clusters present in the provided data but this method is not
similar to K-means clustering. It groups the similar data in one
group and different types of data in different group. One of the
advantages of FCM method is it gives accurate result when
compared with that of K-means clustering [14].
E. Feature Extraction
There are many techniques available for extracting features
from the images. in this research GLCM (Gray level
co-occurrence matrix) is used for extracting texture features
from the target image. This technique is preferred because it
deals with the pixel information about position of identical
gray level values. A total of 13 different features are
extracted-Autocorrelation, Contrast, Correlation, Correlation,
Cluster Prominence, Cluster Shade, Dissimilarity, Energy,
Entropy, Homogeneity, Homogeneity, Maximum probability,
Sum of squares, and Variance
F. Classification
SVM (Support Vector Machine) [15] is a machine learning
algorithm used for
classification. It is used for
analyzing data and recognize

International Journal of Innovative Technology and Exploring Engineering (IJITEE)
ISSN: 2278-3075, Volume-8, Issue-11, September 2019
2213
Published By:
Blue Eyes Intelligence Engineering
& Sciences Publication
Retrieval Number K20440981119/2019©BEIESP
DOI: 10.35940/ijitee.K2044.0981119
the patterns and classifies by identifying different patterns
from the given data. First a set of input data is considered,
from these two classes are categorized. These classes are
given to classifier for detection[16].
In this research SVM classifies the abnormal images with
that of the normal image and the result is obtained. The result
provides the accuracy for the taken classification.
IV. RESULTS AND DISCUSSIONS
Phase 1: Here figure 2a and 2b represents the normal and
abnormal images taken from the dataset for classification
purpose. All the collected cancerous and non-cancerous
images are in Dicom format.
Figure 2a: Normal Lung iamge
Figure 2b: Abnormal Lung image
Phase 2: Noise is removed successfully with the help of
median filter while preserving all the edges carefully for
classification purpose. The filtered image is as shown in
Figure 3.
Phase 3: After filtering of image, segmentation of the image
was done using Fuzzy C-Means clustering. FCM was
successfully applied and better results were shown as in
Figure 4.
Phase 4: Table 1 represents the extracted features of the
taken input image by using GLCM. These are given to the
classifier for the further classification.
Figure 3: Filtered image
Figure 4: FCM Segmented image
S.no
Features
Extracted value for
taken input image
1
Autocorrelation
4.9129
2
Contrast
2.8974
3
Correlation
9.0993
4
Cluster Prominence
1.0292
5
Cluster Shade
1.5734
6
Dissimilarity
2.2932
7
Energy
4.2073
8
Entropy
1.5343
9
Homogeneity
8.9457
Table 1: Feature values extracted for taken input image.
Phase 5: The dataset used for this research consists of total 60
CT lung images. Among these 30 are cancerous images and
30 are non-cancerous images.
Total 9 features were selected for classification purpose
which gave the below scatter plot as shown in Figure 5. This
plot was obtained by plotting the obtained autocorrelation
values against contrast values. In this plot the blue color
represents the cancerous images, red represents the
non-cancerous images and cross marks represents the
misclassified images.
Here Figure 6 represents the parallel co-ordinate plot for
the taken 9 features. In this plot the blue color curve
represents the cancerous images
curve, red represents the
non-cancerous images curve

Lung Cancer Image- Feature Extraction and Classification using GLCM and SVM Classifier
2214
Retrieval Number K20440981119/2019©BEIESP
DOI: 10.35940/ijitee.K2044.0981119
Published By:
Blue Eyes Intelligence Engineering
& Sciences Publication
and red dotted curve represents the misclassified images
curve.
Figure 5: Scatter plot for Autocorrelation versus
Contrast
Figure 6: Parallel Co-ordinate plot
The confusion matrix of taken 60 images is given as shown
in Figure. From the result it is clear that among these 60
images 2 of the images are misclassified.
ROC (Receiver Operating Characteristic) curve for the
taken input images is as shown in Figure. After a successful
100 iterations for the taken 60 images an accuracy of 97.6 %
was achieved. With the help of this images TP, TN, FN, FP
values are known which was used to calculate the accuracy.
Figure 7: Confusion matrix
Where,
A
TP=True Positive
TN=True Negative
FN=False Negative
FP=False Positive
Figure 8: ROC Curve
Therefore, from above plots we can say that performance of
the proposed research for classification of lung cancer had
obtained better results by achieving an accuracy of 96.7
percent.

International Journal of Innovative Technology and Exploring Engineering (IJITEE)
ISSN: 2278-3075, Volume-8, Issue-11, September 2019
2215
Published By:
Blue Eyes Intelligence Engineering
& Sciences Publication
Retrieval Number K20440981119/2019©BEIESP
DOI: 10.35940/ijitee.K2044.0981119
V. CONCLUSION
The proposed methodology was followed by five phases
image collection, image pre-processing, image segmentation,
features extraction and classification. Median filter was used
for image filtering in which the edges were preserved for
classification. After completion of filtering, Fuzzy C-Means
clustering is used for segmentation for fast and better results.
Features for the taken images were extracted using Gray
Level Co-occurrence Matrix. A total of 9 features were
considered for classification and these were given to the
classifier for classification purpose. Final result was obtained
by applying SVM classifier. Plots for the classifier were
successfully obtained and accuracy of 96.7 percent was
achieved with the help of derived values from the respective
plots.
REFERENCES
1. Al-Tarawneh, M.S. (2018). Lung Cancer Detection Using Image
Processing Techniques.
2. Dhaware, B. U., & Pise, A. C. (2016, September). Lung cancer
detection using bayasein classifier and FCM segmentation. In 2016
International Conference on Automatic Control and Dynamic
Optimization Techniques (ICACDOT) (pp. 170-174). IEEE.
3. Anifah, L..., Haryanto, Harimurti, R.P., Permatasari, Z., Rusimamto,
P.W., & Muhamad, A.R. (2017). Cancer lungs detection on CT scan
image using artificial neural network backpropagation based gray level
coocurrence matrices feature. 2017 International Conference on
Advanced Computer Science and Information Systems (ICACSIS),
327-332.
4. Taher, F., & Sammouda, R. (2011, February). Lung cancer detection
by using artificial neural network and fuzzy clustering methods. In
2011 IEEE GCC Conference and Exhibition (GCC) (pp. 295-298).
IEEE.
5. Chunran, Y., Yuanvuan, W., & Yi, G.X. (2018). Automatic Detection
and Segmentation of Lung Nodule on CT Images. 2018 11th
International Congress on Image and Signal Processing, BioMedical
Engineering and Informatics (CISP-BMEI), 1-6.
6. Memon, N. A., Mirza, A. M., & Gilani, S. A. M. (2006, August).
Segmentation of lungs from CT scan images for early diagnosis of lung
cancer. In Proceedings of world academy of science, engineering and
technology (Vol. 14).
7. Mohamed, A. S. E. D., Salem, M. A., Hegazy, D., & Shedeed, H. A.
(2015, December). Probablistic-based framework for medical CT
images segmentation. In 2015 IEEE Seventh International Conference
on Intelligent Computing and Information Systems (ICICIS) (pp.
149-155). IEEE.
8. Kalaivani, S., Chatterjee, P., Juyal, S., & Gupta, R. (2017, April). Lung
cancer detection using digital image processing and artificial neural
networks. In 2017 International conference of Electronics,
Communication and Aerospace Technology (ICECA) (Vol. 2, pp.
100-103). IEEE.
9. Patil, B. G., & Jain, S. N. (2014). Cancer cells detection using digital
image processing methods. International Journal of Latest Trends in
Engineering and Technology, 3(4), 45-49.
10. Segmentation and Detection of Lung Cancer Using Image Processing
and Clustering Techniques Preeti Joon, Shalini Bhaskar Bajaj and
Aman Jatain
11. Li, L., Wu, Y., Yang, Y., Li, L., & Wu, B. (2018, June). A New
Strategy to Detect Lung Cancer on CT Images. In 2018 IEEE 3rd
International Conference on Image, Vision and Computing
(ICIVC) (pp. 716-722). IEEE.
12. Makaju, S., Prasad, P. W. C., Alsadoon, A., Singh, A. K., &
Elchouemi, A. (2018). Lung cancer detection using CT scan
images. Procedia Computer Science, 125, 107-114.
13. E. Ariascastro, and D. L. Donoho, “Does median filtering truly
preserve edges better than linear filtering,” Annals of Statistics, vol.
37,2009, pp.1172-1206, doi:10.1214/08-AOS604.
14. R. M. Haralick, “Statistical and structural approaches to texture,”
Proceedings of the IEEE, vol. 67, May. 1979, pp. 786-804,
doi:10.1109/PROC.1979.11328.
15. Joon, P., Bajaj, S. B., & Jatain, A. (2019). Segmentation and Detection
of Lung Cancer Using Image Processing and Clustering Techniques.
In Progress in Advanced Computing and Intelligent Engineering (pp.
13-23). Springer, Singapore.
16. Kumari, C.U., Rao, G.S.B. and Madhu, R., 2012. Erlang Capacity
Evaluation In Gsm And Cdma Cellular Systems. International Journal
Of Mobile Network Communications & Telematics (Ijmnct) Vol2, (5).
17. Kumari, C.U., Mounika, G. and Prasad, S.J., 2019, March. Identifying
Obstructive, Central and Mixed Apnea Syndrome Using Discrete
Wavelet Transform. In International Conference on E-Business and
Telecommunications (pp. 16-22). Springer, Cham.
18. Kumari, C.U. and Padma, T., 2019. Energy-Efficient Routing
Protocols for Wireless Sensor Networks. In Soft Computing and Signal
Processing (pp. 377-384). Springer, Singapore.
19. Kumari, C.U., 2018, April. Investigation: Life-Time and Stability
Period in Wireless Sensor Network. In 2018 3rd International
Conference for Convergence in Technology (I2CT) (pp. 1-5). IEEE.
20. Kumari, C.U. and Krishna, R., M.: High performance wireless
communication channel using LEACH protocols. Pak. J.
Biotechnol, 13, pp.52-56.
AUTHORS PROFILE
R.Ankita is currently pursuing M.Tech in the field of
Embedded Systems in Gokaraju Rangaraju Institute of
Engineering and Technology. She has pursued her
B.Tech in the field of Electronics and Communication
Engineering. Her research interests include Image
processing and Signal processing.
Dr Ch Usha Kumari is a professor in Department of
ECE, Gokaraju Rangaraju Institute of Engineering and
Technology (GRIET), Hyderabad. She completed her
Ph.D from Jawaharlal Nehru Technological University,
Hyderabad. She completed her M.Tech from Andhra
University Visakhapatnam. She had 14years of teaching
experience. She published many journals and research
papers in national and international conferences. She is
life associate member of IETE and fellow member of IEEE.
Mr. Mohd. Javeed Mehdi is an Assistant Professor
in the Dept. of E.C.E., Gokaraju Rangaraju Institute of
Engineering and Technology, Hyderabad, India. He
completed M.Tech.(Embedded Systems) from VIF
College of Engineering and Technology, Hyderabad .
He has 13 years of teaching experience, and published
and presented papers in national and international
conferences and journals. His areas of research interests are Embedded
Systems, IOT, Data Science, and Image Processing.
N.Tejashwini is currently pursuing M.Tech in the
field of Embedded Systems in Gokaraju Rangaraju
Institute of Engineering and Technology. She has
pursued her B.Tech in the field of Electronics and
Communication Engineering. Her research interests
include Image processing and Signal processing.
.
Dr T. Pavani received her AMIE with first class from
The Institution of Engineers (India), M.Tech and PhD
from Andhra University College of Engineering, Andhra
University. She is having 12 years of teaching and
research experience. Her areas of interest are Antennas,
Electromagnetics, EMI/EMC. and Applications of Soft
computing. She is a life member of the Institution of
Engineers and SEMCE. Presently working for a project sanctioned under
collaborative Research Scheme, with the Grant No:
JNTUH/TEQIP-III/CRS/2019/ECE/9