BRAIN CANCER CLASSIFICATION USING BACK PROPAGATION NEURAL NETWORK AND PRINCIPLE COMPONENT ANALYSIS

International Journal of Technical Research and Applications e-ISSN:2320-8163,
www.ijtra.com Volume 2, Issue 4 (July-Aug 2014), PP. 26-31
26 | P a g e
BRAIN CANCER CLASSIFICATION USING
BACK PROPAGATION NEURAL NETWORK
AND PRINCIPLE COMPONENT ANALYSIS
Ganesh Ram Nayak1, Mr. Toran Verma2
1
M.Tech. (CSE), 2
Sr. Associate Professor,
Rungta College of Engg. & Technology,
Bhilai, Chhattisgarh, India – 490021.
Abstract— Classification of Brain Cancer is implemented
by using Back Propagation Neural network and Principle
Component Analysis, Magnetic Resonance Imaging of brain
cancer affected patients are taken for classification of brain
cancer. Image processing techniques are used for processing
the MRI images which are image preprocessing, image
segmentation and feature extraction is used. We extract the
Texture feature of segmented image by using Gray Level Co-
occurrence Matrix (GLCM). Steps involve for brain cancer
classification are taking the MRI images, remove the noise by
using image pre-processing, applying the segmentation
method which isolate the tumor region from rest part of the
MRI image by setting the pixel value 1 to tumor region and 0
to rest of the region, after this feature extraction technique
has been applied for extracting texture feature and feature
are stored in knowledge based, this features are used for
classification of new MRI images taken for testing by
comparing the feature of new images with stored features. We
implemented three classifiers to classify the brain cancer, first
classifier is back propagation neural network which perform
classification in two phase which are training phase and
testing phase, second classifier is the combination of PCA and
BPNN means by using PCA to reduce the dimensionality of
feature matrix and by using BPNN to classify the brain
cancer, third classifier is Principle Component Analysis which
reduce the dimensionality of dataset and perform
classification. And finally compare the performance of that
classifiers.
Key words— Brain Cancer; MRI; Segmentation; Gray
Level Co-occurrence Matrix; Principle Component Analysis;
Back Propagation Neural Network.
I.INTRODUCTION
The term Brain tumor is any mass that results from
abnormal growths of cells in the brain. It may affect any
person at almost any age. Brain tumor effects may not be
the same for each person, and they may even change from
one treatment session to the next. Brain tumors can have a
variety of shapes and sizes; it can appear at any location and
in different image intensities. Brain tumors can be benign or
malignant [4].
Magnetic Resonance Imaging (MRI) has become a
widely used method of high quality Medical imaging,
especially in brain imaging where MRI’s soft tissue contrast
and noninvasiveness is a clear advantage. MRI provides an
unparalleled view inside the human body. The level of
detail we can see is extraordinary compared with any other
imaging modality. Reliable and fast detection and
classification of brain cancer is of major technical and
economic importance for the doctors. Common practices
based on specialized technicians are slow, have low
responsibility and possess a degree of subjectivity which is
hard to quantify [1] [7]. The advantage of magnetic
resonance imaging (MRI) over other diagnostic imaging
modalities is its high spatial resolution and excellent
discrimination of soft tissues. MRI provides rich
information about anatomical structure, enabling
quantitative pathological or clinical studies [5].MRI is also
a safe and valuable adjunct to the clinical examination of
the knee and an aid to efficient preoperative planning. It is
the most commonly used imaging modality in the
evaluation of the knee joint [10].
A lot of research efforts have been directed towards the
field of medical image analysis with the aim to assist in
diagnosis and clinical studies. The medical images are
obtained from different imaging systems such as MRI scan,
CT scan and Ultra sound B scan. The computerized
tomography has been found to be the most reliable method
for early detection of tumors because this modality is the
mostly used in radio therapy planning for two main reasons.
The first reason is that scanner images contain anatomical
information which offers the possibility to plan the direction
and the entry points of radio therapy rays which have to
target only the tumor region and to avoid other organs. The
second reason is that CT scan images are obtained using
rays, which is same principle as radio therapy. This is very
important because the intensity of radio therapy rays have
been computed from the scanned image [3]. The Medical
Image for Brain Cancer Classification can be Magnetic
Resonance Imaging (MRI) [1], or it can be Computed
Tomography (CT) Scan [3].
The system uses many image processing techniques and
classifiers which are explain in next section, and the
classification result of brain cancer is shown in result
section.
II.METHOD
The system is built by using many image processing
techniques and classifiers, MRI images of brain cancer
affected patients are taken as input and system give the class
of that input MRI images. We organized a MRI images into
five different classes in which some images from each
classes are used for training the network and remaining
images are for testing. To do so we used an image
processing techniques which are image preprocessing,
image segmentation and feature extraction. Finally
classifiers are implemented for classifying the brain cancer.

International Journal of Technical Research and Applications e-ISSN: 2320-8163,
27 | P a g e
Figure 1.1 – Block diagram of the system
The block diagram of developed system is shown in the
figure 1.1, which works in two phases, first phase is
Training phase and second phase is Testing phase for two
classifiers- BPNN and PCA with BPNN, in first phase it
takes the MRI images of brain cancer affected patients and
perform the image processing techniques shown in the
figure 1.1 and feature are stored in the knowledge base and
it is used for training the neural network, and in second
phase testing MRI images are taken as a input and perform
the same operation we performed before on training images
and features of testing images are compared with stored
image in knowledge base. The third classifier is Principle
Component Analysis which takes features of both training
and testing images to classify the brain cancer.
III.STAGES IN CLASSIFICATION
The stages of classification are shown in the figure 1.1,
and detail of each stage is given below –
A. MRI Image
We used the MRI images of brain cancer affected
patients, we takes the MRI images of five different diseases
of brain cancer. The five types of MRI image are –
Astrocytoma, Glioma, Meningioma, Metastasis
bronchogenic carcinoma and Sarcoma. Each of these
disease are organize to class means class I to class V.
B. Image Pre-processing
Brain images are noisy, inconsistent and incomplete,
thus preprocessing phase is needed to improve the image
quality and make the segmentation results more accurate
[3].
For image preprocessing we used a Median filter,
Median filtering is similar to using an averaging filter, in
that each output pixel is set to an average of the pixel values
in the neighborhood of the corresponding input pixel.
However, with median filtering, the value of an output pixel
is determined by the median of the neighborhood pixels,
rather than the mean. The median is much less sensitive
than the mean to extreme values (called outliers). Median
filtering is therefore better able to remove these outliers
without reducing the sharpness of the image. The medfilt2
function implements median filtering.
C. Image Segmentation
Image segmentation is the process of partitioning a
digital image into multiple segments (sets of pixels, also
known as super pixels). The goal of segmentation is to
simplify and/or change the representation of an image into
something that is more meaningful and easier to analyze.
The result of image segmentation is a set of segments that
collectively cover the entire image, or a set of contours
extracted from the image (see edge detection). Each of the
pixels in a region are similar with respect to some
characteristic or computed property, such as color, intensity,
or texture. Adjacent regions are significantly different with
respect to the same characteristic(s).
By using image segmentation we isolate the tumor
region from rest of the image. We applied the Optimum
global Thresholding using Otsu method for image
segmentation, this method computed the threshold value,
each pixel’s intensity of the image is compare with the
threshold value, if the pixel’s intensity is greater than
threshold than pixel value is set to 1 otherwise set 0 and
finally we get a segmented image.
D. Texture Feature Extraction
The work involves extraction of the important features
for image recognition. The features extracted give the
property of the texture, and are stored in knowledge base.
[1]. the extracted features are compare with the unknown
sample means the testing image for classification.
We used a Gray Level Co-occurrence matrix for texture
feature extraction. Gray level co-occurrence matrix
(GLCM) was firstly introduced by Haralick. A gray-level
co-occurrence matrix (GLCM) is essentially a two-
dimensional histogram. The GLCM method considers the
spatial relationship between pixels of different gray levels.
The method calculates a GLCM by calculating how often a
pixel with a certain intensity i occurs in relation with
another pixel j at a certain distance d and orientation Ɵ. A
co-occurrence matrix is specified by the relative frequencies
P (i, j, d, Ɵ). A co-occurrence matrix is therefore a function
of distance d, angle Ɵ and gray scales i and j. [6] [8] [9].
Figure 3.4.1 - Direction for generation of GLCM
We extracted the following texture feature which are
given below –
1) Contrast:-
Contrast returns a measure of the intensity contrast
between a pixel and its neighbor over the entire image
[6][9].

28 | P a g e
Contrast =
Range = [0 (size(GLCM,1)-1)^2]
Contrast is 0 for constant image.
2) Correlation:-
Returns a measure of how correlated a pixel is to its
neighbor over the whole image. [6][9].
Range = [-1 1]
Correlation is 1 or -1 for a perfectly positively or
negatively correlated image. Correlation is NaN for a
constant image.
Correlation =
Where mr and mc are mean computed along rows and
column respectively, are in form of standard
deviations computed along rows and column respectively.
3) Angular Second Moment (Uniformity or Energy):-
Angular Second Moment is also known as Uniformity
or Energy. It is the sum of squares of entries in the GLCM
Angular Second Moment measures the image homogeneity.
Angular Second Moment is high when image has very good
homogeneity or when pixels are very similar [6][9].
Energy =
Rang= [0,1]
Energy is 1 for constant image.
4) Inverse Difference Moment (Homogeneity):-
Inverse Difference Moment (IDM) is the local
homogeneity. It is high when local gray level is uniform and
inverse GLCM is high. It return a value that measure the
closeness of the distribution of element in the G to the
diagonal of G.[6][9]
Rang = [0,1]
Homogeneity is 1 for diagonal G.
Homogeneity =
5) Entropy:-
Entropy is a statistical measure of randomness that can
be used to characterize the texture of the input image.
Entropy is defined as
Entropy = -
Where k is the rows (or column) dimension of square
matrix G probability is the ij-th element of G/n, where n is
equal to sum of the element of G, and G is referred simply
as co=occurrence matrix. [6][9].
E. Classifier Implementation
Classifiers are used to classify the brain cancer. We
implemented three classifiers two classify the brain cancer
which are given below –
1) Back Propagation Neural Network
Back propagation is a supervised learning method. In
supervised learning, each input vector needs a
corresponding target vector. Input vector and target vector
are presented in training of the network. The output vector
(i.e. actual output) which is result of the network is
compared with the target output vector then an error signal
is generated by the network. This error signal is used for
adjustment of weights until the actual output matches the
target output. Algorithm stages for BPN are initialization of
weights, feed forward, back propagation of Error and
updating of weights and biases [6] [2].
After the weights are adjusted on the training set,
their value is fixed and the ANN's are used to classify
unknown input images. The generalized delta rule involves
minimizing an error term defined as [1]
In this equation, the index p corresponds to one input
vector, and the vectors tp and op are the target and observed
output vectors corresponding to the input vector p,
respectively [1].
Figure 3.5.1 – Performance graph of BPNN
2) PCA with Back Propagation Neural Network
In this classifier we reduce the no of features by using
Principle Component Analysis. We takes a 15 samples of
MRI images for training the neural network, feature is
extracted using gray level co-occurrence matrix and we got
a 113 features value of each images and features of all
images are reduce by using Principle Component Analysis
and we got a reduce data set. Set procedure is applied for
testing MRI images and classification is done by using Back
Propagation Neural network explained in previous section.
Figure 3.5.2 – Performance graph of PCA with BPNN
3) Principle Component Analysis
Here Principle Component Analysis is used as a
classifier, in which it derives the new feature set from large
extracted feature using GLCM. After deriving the new
feature it applies a Euclidian distance for brain cancer
classification.

29 | P a g e
The mathematical background required for PCA is
given below-
a) Mean
b) Standard Deviation
c) Variance
d) Covariance
e) Covariance Matrix
f) Eigen Vector and Eigen value
a) Mean
Notice the symbol X̅ (said “X bar”) to indicate the mean
of the set. All this formula says is “Add up all the numbers
and then divide by how many there are” [11].
b) Standard Deviation
The Standard Deviation (SD) of a data set is a measure
of how spread out the data is. How do we calculate it? The
English definition of the SD is: “The average distance from
the mean of the data set to a point”. The way to calculate it
is to compute the squares of the distance from each data
point to the mean of the set, add them all up, divide by, (n-
1) and take the positive square root. As a formula [11]
c) Variance
Variance is another measure of the spread of data in a
data set. In fact it is almost identical to the standard
deviation. The formula is this:[11]
d) Co – variance
Standard deviation and variance only operate on 1
dimension, so that you could only calculate the standard
deviation for each dimension of the data set independently
of the other dimensions. However, it is useful to have a
similar measure to find out how much the dimensions vary
from the mean with respect to each other. Covariance is
such a measure. Covariance is always measured between 2
dimensions. If you calculate the covariance between one
dimension and itself, you get the variance. So, if you had a
3-dimensional data set (x, y, z ), then you could measure the
covariance between the x and y dimensions, the x and z
dimensions, and the y and z dimensions. Measuring the
covariance between x and x, or y and y , or z and z would
give you the variance of the < , = and > dimensions
respectively. The formula for covariance is very similar to
the formula for variance. The formula for variance could
also be written like this:[11]
e) Covariance Matrix
Covariance is always measured between 2 dimensions.
If we have a data set with more than 2 dimensions, there is
more than one covariance measurement that can be
calculated. For three dimensional (x, y, z) data set calculate
cov(x,y), cov(x,z) and cov(y,z) In fact for n dimensional
data set, we can calculate total covariance values.[11]
f) Eigenvector and eigenvalue
As you know, you can multiply two matrices together,
provided they are compatible sizes. Eigenvectors are a
special case of this. Consider the two multiplications
between a matrix and a vector in In the first example, the
resulting vector is not an integer multiple of the original
vector, whereas in the second example, the example is
exactly 4 times the vector we began with. Why is this?
Well, the vector is a vector in 2 dimensional space The
other matrix, the square one, can be thought of as a
transformation matrix. If you multiply this matrix on the left
of a vector, the answer is another vector that is transformed
from its original position. What properties do these
eigenvectors have? You should first know that eigenvectors
can only be found for square matrices. And, not every
square matrix has eigenvectors.[11]
STEPS TO BE FOLLOWED IN PCA [11]
1. Get some data
2. Subtract the mean
3. Calculate the covariance matrix
4. Calculate the eigenvectors and eigenvalues of the
covariance matrix
5. Choosing components and forming a feature vector
6. Deriving the new data set
7. Calculate the Euclidian distance between feature of
unknown MRI image and All stored Feature, and find out
minimum distance and according to these distance unknown
samples are classified.
IV.RESULTS
The developed system classifies the brain cancer of MRI
images of brain cancer affected patients. We implemented
three classifiers, each classifiers have their own efficiency
to classify the brain cancer. We takes the five types of brain
cancer which are organize into five different classes from
class I to class V. We isolate the tumor region from rest of
the MRI image by using segmentation, which shows the
tumor region of MRI images which is shown in the figure
below –

30 | P a g e
Figure 4.1 – Original MRI image
Figure 4.2 – Preprocessed MRI image
Figure 4.3 – Segmented MRI image
Figure 4.3 shows the tumor region of MRI image which
image is used for feature extraction.
We developed Graphical User Interface for user friendly
operation on the system, the GUI contain all the component
of developed system means feature extraction of both
training images and testing images, dedicated button of all
three classifiers, training of neural network and training of
PCA with neural network. The performance of three
classifiers are shown in the table 4.1, which is given below
–
Number of Testing
No. of Testing images
successful classified by
sample classifiers
BPNN PCA with
BPNN
PCA
5 4 3 5
Table 4.1 – Performance of Classifiers
V.CONCLUSION & SCOPE OF FURTHER WORK
The brain cancer classification system has been
designed by using Back propagation neural network and
Principle Component Analysis which uses image processing
techniques like image preprocessing, image segmentation
and feature extraction using Gray Level Co-occurrence
matrix. The system uses three classifiers for classification of
brain cancer which are Back Propagation Neural Network,
PCA with BPNN and Principle Component Analysis. The
classification performance of three classifiers are shown in
the table 4.1, which shows that performance of Principle
Component Analysis is better as compare to two other
classifiers for classification of brain cancer.
This system classify a few type of brain cancer, the main
aim of this system is to compare the performance of
classifier which are used in this system. The system can be
implemented which classify all type of brain cancer by
using appropriate classifier for each type of cancer. The
scope of the system can further be improved by using other
types (e.g. PET, MRS, CTS) of Images.
REFERENCES
[1] Dipali M. Joshi, Dr. N. K. Rana and V. M. Misra. 2010.
“Classification Of Brain Cancer Using Artificial Neural
Network”, 2nd International Conference on Electronic Computer
Technology, pp. 112-116.
[2] Kailash D.Kharat, Pradyumna P.Kulkarni,
M.B.Nagori.2012. ”Brain Tumor Classification Using Neural
Network Based Method”, International Journal of Computer
Science and Informatics, vol. – 1, pp. – 85-90.
[3] A. Padma, R. Sukanesh. 2011. “Automatic Classification and
Segmentation of Brain Tumor in CT images Using Optimal
Dominant Gray Level Run Length Texture Feature”, International
Journal of Advanced Computer Science and Applications,Vol.
2, No. 10, pp. – 53-59.
[4] Ahmed Kharrat, Karim Gasmi, Mohamed BEN.
2010. “A Hybrid Approach for Automatic Classification of Brain
MRI Using Genetic Algorithm and Support Vector Machine”,
Leonardo Journal of Sciences, pp. 71-82.
[5] W. M. Wells, W. E. L. Grimson, R. Kikinis, F. A. Jolesz.
1996. “Adaptive Segmentation of MRI Data”, IEEE Transaction
on Medical Imaging, Vol. – 15, No. – 4, pp.429-442.
[6] Shweta Jain, Shubha Mishra. 2013. “ANN Approach Based
on Back Propagation Network and Probabilistic Neural Network
to Classify Brain Cancer”, International Journal of Innovative
Technology and Exploring Engineering, Vol. – 3, Issue – 3, pp. –
101-105.
[7] Matthew C. Clark, Lawrence O. Hall, Dmitry B. Goldgof,
Robert Velthuzen, F. Reed Murtagh, Martin S. Silbiger. 1998.
“Automatic Tumor Segmentation using Knowledge Based
Techniques”. IEEE Transaction on Medical Imaging, Vol. – 17,
No. – 2, pp. – 187-201.
[8] Minakshi Sharma, Dr. Sourabh Mukharjee. 2012. “Brain
Tumor Segmentation using hybrid Genetic Algorithm and
Artificial Neural Network Fuzzy Inference System”, International
Journal of Fuzzy Logic System, Vol. – 2, No. – 4, pp. – 31-42.
[9] P. Mohanaiah, P. Sathyanarayana, L. GuruKumar. 2013.
“Image Texture Feature Extraction Using GLCM Approach”.

31 | P a g e
International Journal of Scientific and Research Publications, Vol.
– 3, Issue – 5, pp. – 1-5.
[10] Mustafa Z. Mahmoud, Maram A. Fagiri, Atallh Fowzan
Al-Motrfi, Abdelmoneim Sulieman. 2013. “Magnetic Resonance
Imaging Findings in Knee Joint Pain at King Saud Medical City”.
International Journal of Science and Research, Vol. – 2, Issue – 6,
pp. – 4-7.
[11] Varada S.Kolge, Prof.K.V.Kulhalli. "PCA and PNN assisted
automated brain tumor classification", IOSR Journal of Electronic
and Communication Engineering, pp. - 19-23.
Author Profile
Ganesh Ram Nayak, completed B.E. Information
Technology from Shri Shankaracharya College of
Engineering and Technology, Bhilai (C.G) in 2012 and
Now he is pursuing M.Tech in Computer Science &
Engineering from Rungta College of Engineering and
Technology, Bhilai (C.G).
Mr. Toran Verma working as Associate Professor in
Department of Computer Science & Engineering at Rungta
College of Engineering and Technology, Bhilai (C.G).

BRAIN CANCER CLASSIFICATION USING BACK PROPAGATION NEURAL NETWORK AND PRINCIPLE COMPONENT ANALYSIS

Related slideshows

More Related Content

BRAIN CANCER CLASSIFICATION USING BACK PROPAGATION NEURAL NETWORK AND PRINCIPLE COMPONENT ANALYSIS