IRJET- Image Segmentation using Classification of Radial Basis Function of Neural Network in Brain Tumor Detection

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 06 Issue: 05 | May 2019 www.irjet.net p-ISSN: 2395-0072
© 2019, IRJET | Impact Factor value: 7.211 | ISO 9001:2008 Certified Journal | Page 959
IMAGE SEGMENTATION USING CLASSIFICATION OF RADIAL BASIS
FUNCTION OF NEURAL NETWORK IN BRAIN TUMOR DETECTION
P. PRIYADHARSHINI1, R. THILLAIKKARASI2, S. SARAVANAN3
1PG scolar, ME., (Applied Electronics) Department of ECE, Salem College of Engineering And Technology, Salem.
2 Assistant Professor, Department of ECE, Salem College of Engineering and Technology, Salem.
3 Professor /Head, Department Of EEE, Muthayammal Engineering College, Namakkal, Tamilnadu, India
---------------------------------------------------------------------***----------------------------------------------------------------------
ABSTRACT - The location of with tumors in the brain is one
of the factors that determine how a brain tumor affects an
individual's functioning and whatsymptomsthe tumorcauses.
Along with the Spinal cord, the tumor forms the Central
Nervous System (CNS). The brain tumor can occur at any
stage. Image pre-processing techniques are used to improve
the quality to an image before processing and sending into an
application. The image processing techniques use a small
neighborhood of a pixel in an input image to get a new
brightness value in the output image. These pre-processing
techniques are also called as separation and discerning
enhancement. The algorithm incorporates steps for
pre‐processing, image segmentation, andfeatureextractionof
the GLCM. The RBFN is a 3-layer network where the input
vector is the first layer, the second "hidden" layer is the RBF
neurons, and the third layer is the output layer containing
linear combination neurons. Finally, using the segmentation
technique and morphological operations tumorous region is
isolated from an abnormal input image.
INTRODUCTION
The size can vary extensively. Real-time diagnosis
of tumors by using more reliable algorithms has been the
main focus of the latest developments in medical imaging
and identification taking part in brain tumor using MRI
images. The segmentation of the cells and their nuclei from
the rest of the image content is one of the main problems
faced by most of the medical imagery diagnosissystems.The
process of determining in most powerful and diagnosis
segmentation. Image Segmentation is performed on the
input images.
A. Operations and Types of Tumor:
3D appearance segmentation aids in the automated
diagnosis of brain diseases and helps in qualitative and
quantitative analysis of images such as measuring accurate
size and volume of the detected portion.
B. Tumor:
An enlargement part of the body, generally without
inflammation, caused by an unusual development of fleshy
tissue, whether nonthreatening or malignant.
C. Types of Tumor:
There are three types of tumor:
1) Benign; 2) Pre- Malignant; 3) Malignant.
1) Benign Tumor:
A benign tumor is a tumor is the one that does not
develop in an unexpected way; it doesn't affect its
neighboring healthy tissues and also does not enlarge to
non-neighboring tissues. The secret agent is the common
example of benign tumors.
2) Pre-Malignant Tumor:
It is also a precancerous stage, imitatedasa disease,
if not precisely treated it may lead to the tumor.
3) Malignant Tumor:
Malignant is fundamentally a medical word that
defines a severe progressing disease.Themalignanttumoris
a term which is normally used for the description of cancer.
The Magnetic Resonance Imaging (MRI) is to view the
internal structures of the body in element particularly for
imaging soft tissues and it does not use any particle
emission. The major problem in image segmentation is the
inaccurate diagnosis of the tumor regionwhichgetsreduced
mainly due to the contrast, blur, noise, artifacts, and
distortion. Even a small amount of noise can change the
classification. (1)
Fig.1 input MRI image

Volume: 06 Issue: 03 | Mar 2019 www.irjet.net p-ISSN: 2395-0072
MATERIALS AND METHODS
In this phase, an image is improved in the way that
finer details are enhanced and noise is uninvolved from the
image. Most commonly used enhancement and noise
reduction techniques are applied that can give the best
possible results. Development of result in more prominent
edges and a perfected image is developed, a noise will be
concentrated on the distorting effect of the image.
The Magnetic Resonance Imaging (MRI) is to view
the internal structures of the body in detail exclusively for
imaging soft tissue and it does not use any radioactivity. The
brain tumor is an abnormal growth of tissues in the brain
and is mainly caused by radiation to the head, genetic risk,
HIV infection, cigarette smoking and also due to
environmental toxins. The major problem in image
segmentation is the inaccurate diagnosisofthetumorregion
which gets reduced mainly due to the contrast. (6)
Fig 2 Flow chart of Proposed Methodology
A. Image Acquisition
Images are obtained using MRI scan & displayed in
2D having pixels as its elements. MRI scan was stored in a
database of images in JPEG image formats. These images are
displayed as grayscale images. The entries of grayscale
images are ranging from 0 to 255, where o point to total
black color and 255 signifies the whole white color.
B. Pre-Processing Stage
The sum of contrast augmentation for some
greatness is directly proportional to the slope of the
Cumulative Distribution Function (CDF).
Fig 3 Filtered Images
Fig 4 Adaptive Histogram Equalization
C. Extraction of Texture Feature
Gray-level co-occurrence matrix (GLCM) is the
Statistical method of investigating the texturesthatconsider
the spatial relationship of the pixels. The GLCM functions
characterize the texture of an image by calculating how
frequently pairs of a pixel with specific values and in a
specified spatial relationship that presentinanimage,forms
GLCM. It is the most broadly used and more generally
applied method because of its high accuracy and less
computation time. Let Nh be the total number of pairs, then
Cij = cij/Nh is the elements of normalized GLCM.

Texture feature extraction by using GLCM
It even provides a contrast between malignant and
normal tissue, which may be below the threshold of human
perception. The statistical features of MR images are
obtained using Gray Level Co-occurrence Matrix (GLCM),
which is also known as Gray Level Spatial Dependence
Matrix (GLSDM)? GLCM, introduced by Heraldic is a
statistical approach that can well describe the longitudinal
connection between pixels of dissimilar gray intensities.
GLCM is a two-dimensional histogram in which (i, j)the
element is the frequency of event I that occurs with j.
D) Classification using RBF
The next step in the proposed system is to classify
and train the extracted signal. We use the RBF network as
described in the below Figure 3 to train and test the signals
of MRI. Learning is of 2 stages: using two methods such as
unsupervised methods, supervised methods14. The basic
RBF is of the three-layer network, as shown in Figure 3.
The main features of RBF are as follows and shown
in the above Figure 3. They are 2 layer feed forward net-
work. RBF consists of a set of the hidden layer. MLP is used
to implement the output nodes. It covers even smaller
regions also. It is much faster than BPN since it has 2 stages
of learning. The different learning algorithms are:
The architecture of the Radial Basis Function
Network
RBF has wide-ranging researchimportancebecause
they are world-wide approaches, fast learning speed due to
locally tuned neurons and they have compressed topology
than other neural networks. Radial basisfunctionnetwork is
used for a wide range of applications primarily because it
can approximate any regular function and its training speed
is faster than multi-layerperceptron(MLP).Thearchitecture
of the RBF network is given below.
x1
x2
x3
input lay er
(fan-out)
hidden lay er
(weights correspond to cluster centre,
output function usually Gaussian)
output layer
(linear weighted sum)
y 1
y 2
Fig 4 Architecture of RBF
However, if pattern classification is required,
then a hard-limiter or sigmoid perform can be placed on
the output neurons to convey0/1 output values and also
the cluster. The single feature of the RBF network is the
process performed in the out of sight layer. Furthermore,
this distance measure is made non-linear, so that if a design
an area that is close by to a cluster interior it gives a value
close to 1.Then the Gaussian operate most typically used
radial-basis operate may be a Gaussian operate.
In associate degree RBF network, r is the distance from
the cluster center. The equation represents a Gaussian bell-
shaped curve.
The distance measured from the cluster center is
usually the Euclidean distance. For each neuron in the
hidden layer, the weights represent the coordinates of the
center of the cluster. The root-mean-square distance
between the current cluster center and its P nearest
neighbors is calculated, and this is the valuechosenfor. So,if
the current cluster center is cj, the value consider.
Advantages/Disadvantages
 RBF trains faster than an MLP
.
 It is slower than an MLP, so wherever speediness is
a factor an MLP may be more applicable.
Fig 5 The Given Image is Abnormal
E) Image Segmentation
Image segmentation can partition the
brain imaging scan image into multiple segments (sets of
pixels, conjointly called super pixels). One ofthesimplest
methods is that of histogramming and Thresholding. If we
plot the Number of pixels which have a speciﬁc grey value
versus that value, we create the histogram of the image.
Properly normalized, the histogram is essentially the
probability density function of the grey values of the image
Assume that we have an image consisting of a bright object
on a dark background and Assume that we want to extract
the object. For such an image, the histogram will have two
Peaks and a valley between them. We can choose as the
threshold then the grey value which corresponds to the
valley of the histogram, indicated by t in ﬁgure 6.2a, and
label all pixels with grey values greater than as object pixels
and all pixels with grey values smaller than as background
pixels.
Image segmentation is typically used to locate
objects and boundaries (lines, curviest) in images is that

the method of allocating a label to each element in a
picture such pixels with the similar label share assured
visual physical characteristics. The result of image splitting
up is a set of segments that mutually cover the entire image
or a set of contours extracted from the image. The main
difficulties in the edge recognition process are that the
malignant cells near the spacious of the MRI are very fat,
thus look like very dark on the MRI, which is very mystifying
in the edge recognition process. To overcome the difficult,
two steps were completed.
a. Histogram Equalization
Histogram equalization is used to enhancecontrast.
In some case were histogram equalizationcanverypoorand
then decreasing the contrast.
Fig 6: a) The original MRI b) Histogram
Equalized MRI
b. Thresholding
Thresholding is the simplest method of image
segmentation. From a grayscale image, Thresholding can be
used to create binary images in many visionpresentations,it
is convenient to be ably isolated out the constituenciesof the
image.
Fig 7: Original MRI Image
Fig 8: Threshold Segmented Image
(White portion indicates tumor)
CONCLUSION
Finally, for the recommended system, the essential
effect Image has to be chosen for determining better
segmentation techniques. An effective database with high
recognition has to be implemented. Early detection of the
tumor will be useful to the patients for who are smaller
tumors that are class 1 and class 2 tumors which can be
cured easily.
REFERENCE
[1] Roopali R.Laddha, S.A. Ladhake ―A Review on Brain
Tumor Detection Using Segmentation And Threshold
Operations‖, International Journal of Computer Science
& Information Technologies‖, Vol.5, 2014.
[2] Ahmed Kharrat, Mohamed Ben Messaoud, Nacéra
Benamrane, Mohamed Abid (2009)―DetectionOfBrain
Tumor In Medical Images‖, IEEE International
Conference On Signals, Circuits And Systems
[3] S. Rajeshwari, T. Sree Sharmila ―Efficient Quality
Analysis Of MRI Image Using Preprocessing
Techniques‖, IEEE 2013
[4] N. Kwak, and C.H. Choi, "Input Feature Selection for
classification problem", IEEE Transactions on Neural
Networks, 13(1), 143- 159, 2002
[5] Prinyanka, Balwinder Singh ―AReviewOnBrainTumor
Detection Using Segmentation‖, International Journal of
Computer Science & Mobile Computibg , Vol 2,Issue.7,
2013.
[6] M.Karuna, Ankita Joshi ―Automatic detection and
Severity Analysis of Brain Tumours using GUI in Matlab,
International Journal of Research in Engineering and
Technology, Vol.2, Issue10, Oct 2013.

[7] P. Padmapriya*, K. Manikandan, K. Jeyanthi, V. Renuga
and J. Sivaraman ‘Detection and Classification of Brain
Tumor using Radial Basis Function’ Indian Journal of
Science and Technology, Vol 9(1, January 2016
[8] Wen-Bo Zhao1,2, De-Shuang Huang2, and Lin
Guo2!”Comparative Study between Radial Basis
ProbabilisticNeural Networks andRadial BasisFunction
Neural Networks1”Department of Automation,
University of Science and Technology ofChina Springer-
Verlag Berlin Heidelberg 2003
[9] Performance Comparison of Radial Basis Function
Networks and Probabilistic Neural NetworksforTelugu
Character RecognitionBy T. Sitamahalakshmi,
Dr.A.VinayBabu, M. Jagadeesh, Dr.K.V.V .ChandraMouli.

IRJET- Image Segmentation using Classification of Radial Basis Function of Neural Network in Brain Tumor Detection

More Related Content

What's hot (19)

Similar to IRJET- Image Segmentation using Classification of Radial Basis Function of Neural Network in Brain Tumor Detection (20)

More from IRJET Journal (20)

Recently uploaded (20)

IRJET- Image Segmentation using Classification of Radial Basis Function of Neural Network in Brain Tumor Detection