This document presents a comparative study of brain tumor segmentation and detection techniques in MRI images. It discusses several techniques used for brain tumor segmentation including fuzzy transform, morphological operations, thresholding, edge-based and region-based methods. The document also reviews various literature on brain tumor detection algorithms using techniques such as fuzzy clustering, neural networks, watershed segmentation, and support vector machines. Accuracy levels of different algorithms are presented ranging from dice coefficients above 0.90 to overall accuracy of 96.51%.
HIV AND INFULENZA VIRUS PPT HIV PPT INFULENZA VIRUS PPT
285-IJAEMA-FEBRUARY-3487.pdf
1. A Comparative Study of Brain Tumor
Segmentation and Detection
R.Vinayagamoorthy1
, Dr.R.Balasubramanian2
,
Manonmaniam Sundaranar University Constituent Model College, Manonmaniam Sundaranar University
vinayagamoothy3@gmail.com
rbalus662002@yahoo.com
Abstract
In the recent years, brain tumor detection and segmentation by the MRI Images in the
medical field has been very useful in recent years. Due to accuracy of the MRI Images the
investigation can detect brain tumor in an exact manner. The MRI images help in the detection
of unusual growth of tissues and blocks of blood in nervous system. The initial step in the
detection of brain tumor comprises of the checking of symmetric and asymmetric shape of brain
to define the abnormal development in the brain. After these steps, the next step is to segment the
images. It is based on two technical processes. One is the F-Transform (Fuzzy Transform)
technique and the other is Morphological operation technique. Basically, these two techniques
are used to design the image in Magnetic Resonance Imaging. The design of the image can help
to detect the precincts of brain, brain tumor and calculate the concrete area of the tumor in the
brain. The f-transform method is used to give certain information like the rebuilt of missing
edges and extracting the silent edges in the images. The accuracy and clarity of the MRI Images
depends on each other. Hence, the study pursues a comparative study of the development of
Brain Tumor segmentation and detection in MRI.
Keywords: MRI Images, Brain Tumor, Fuzzy Transform, Morphological operation.
I. INTRODUCTION
In the modern medical field, MRI Images are very useful in Medical image processing.
The brain tumor is an unusual growth of tissues and uncontrolled cells and its rise. Due to this
state, the natural pattern of cell growth and death is failed and the tumor develops contaminating
the neighboring cells. The brain tumor consists of Primary stage and Secondary stage. When
Brain tumor spreads in any part of the brain, it is known as brain tumor. Brain tumor can be
The International journal of analytical and experimental modal analysis
Volume XII, Issue II, February/2020
ISSN NO: 0886-9367
Page No:2447
2. identified through a number of symptoms like seizures, mood swing, difficulty in limb
movements, and muscular movement etc. Medically, brain tumor is classified as Gliomas,
Medulloblastoma, Epeldymomas, CNS lymphoma and Oligodendrogloma. At the primary stage,
there is possibility of the tumor to be removed but it goes on to the secondary stage without
identification at early stages, the tumor disease could spread even after the removal of tumor it
remains and surfs back again and infects the area too vigorously. This is the biggest problem in
the development of the second stage tumor. Why this problem of second stage tumor occurs in
the modern medical times? It occurs due to the inaccurate location finding of the tumor area. The
next step is the problem in the facility of detection techniques. Segmentation and detection are
tumor detection techniques of the imaging of brain tumor. It can be done by MRI scanning i.e.
Magnetic Resonance Imaging, CT scanning i.e. computer tomography and Ultra sound etc. There
are several methods to detect brain tumor and by the tumor methods one can detect and diagnose
them more easily. Some edges are of nuclear network algorithm, watershed and edge detection,
fuzzy c mean algorithm and the asymmetry of brain has been used to detect the abnormality.
Edge detection is the one of the most crucial problem for image processing due to various
application techniques. Candy-edge detection method is one of the most constructive features in
image segmentation. This detection has been used for the extraction of edges. F-transform
method is an intelligent mode to handle uncertain information in image segmentation process.
This way of method is useful for the detection of tumor boundaries in the brain. Hence, this is a
very easy method for detection and a promising method for future edge extraction progress.
II. BASIC METHODOLOGY
The figure shows the basic block diagram of brain tumor detection and segmentation
procedure in MRI. The MRI images of brain are taken for processing. In the process image
acquisition is very prominent in segmentation.
The first step is in the method is Image Acquisition and the first considered MRI scan
images are of the given patient are either color, Gray-scale or intensity images displayed with a
default size of 220×220. If the color image needs such image acquisition, a Gray-scale converted
image is distinct by using a large matrix whose entries are of the numerical values between 0 and
255. In them, the value 0 corresponds to black and 255 to white part of the image. The brain
tumor detection of a patient consists of two key stages. They are image segmentation and edge
detection.
The International journal of analytical and experimental modal analysis
Volume XII, Issue II, February/2020
ISSN NO: 0886-9367
Page No:2448
3. The next step is the Pre-processing Stage. It consists of Noise removal methods. This
method can be done by using various spatial filters or linear or nonlinear filters. Other artifacts
like text are removed by some sort of morphological operations.
Fig. 1 Basic Block Diagram of Brain Tumor Detection and Segmentation Process
In the pre-processing stages, RGB to grey conversion and reshaping also takes place. It
includes of median filter for noise removal methods. In general, the possibilities of noise
interaction in modern MRI scan are very low. It is because of the thermal Effect.
Image Smoothing is the method of simplifying an image while at the time of preserving
important information in the image. The chief aim is to reduce noise or useless details without
introducing high level of distortion and to simplify subsequent analysis in the procedure.
Image Registration is the next process of bringing two or more images into proper
alignment. In medical imaging, image registration allows for the concurrent use of images taken
with different modalities such as MRI and CT scanning methods at dissimilar times or with
dissimilar patient positions. For example, images are acquired preoperative, as well as during
intra-operative surgery. Due to time constraints, the real-time intra-operative images have lower
resolution than the pre-operative images. Moreover, deformations that occur at the time of
surgery makes difficult to compare with the high resolution pre-operative image to the low
resolution intra-operative surgery of the patient. Hence, Image registration attempts to help the
surgeon relate the two sets of images [8].
The International journal of analytical and experimental modal analysis
Volume XII, Issue II, February/2020
ISSN NO: 0886-9367
Page No:2449
4. Image Segmentation is the key stage to identify the image properly because it affects the
accuracy of the subsequent diagnostic steps. On the other hand, proper segmentation is intricate
because of the different verities in lesion shapes, sizes, and colors along with different types of
skin and textures. Moreover, some lesions have asymmetrical boundaries and some have smooth
transition between the lesion and the skin. Several algorithms have been proposed to solve the
problem. They are classified as thresholding, edge-based or region-based methods and of
supervised and unsupervised techniques. The processes are termed as Threshold segmentation,
Water shed segmentation, Gradient Vector Flow, K-mean Clustering and Fuzzy C-means
Clustering.
After proper segmentation, morphological processing has been applied to remove the
unwanted parts. This process consists of image opening, image closing, dilation, and erosion
operations.
At the end of these segmentation and detection process, decision has been taken weather
that MRI image consists of any tumor or not and the normal or the abnormal state of weather has
been checked.
III. REVIEW OF LITERATURE
The 2016 World Health Organization says on the classification of tumor of central
nervous system is a conceptual one as well as pertain overview of predecessor. WHO classifies
CNS tumor by molecular parameters for its diagnosis structure. Further than 2016 CNS WHO
presence the new diffuse glomas and other tumor and defines the new feature like both histology
as well as molecule [1].
There are different types of brain tumors. They are Glioma, Papillary, Glioneuronal
tumor etc. The histological variants are capable of different edge distribution, location,
symptoms and the behaviours or clinical [2].
Fuzzy clustering is method which has been widely used in biomedical field to detect the
image. Effective fuzzy clustering algorithm is used in abnormal MRI brain image segmentation.
By using clustering in brain tumor segmentation we can diagnose accurately the region of cancer
[3].
Now-a-days, brain tumor is one of the major hazardous diseases. Its detection should be
fast and accurate and can be detected by automated tumor detection techniques. One of the
automated tumor detection techniques is the use of MRI images. It defines the tumor growth
The International journal of analytical and experimental modal analysis
Volume XII, Issue II, February/2020
ISSN NO: 0886-9367
Page No:2450
5. region and the edges detection. As compare to other techniques with this it gives more accurate
as well as clear and advantages of automated tumor detection techniques is used for removal of
tumor if needed [4].
The neural networks are a new technology has been discovered. The neural networks are
an “HOT” research area, like a cardiology, radiology, oncology etc. To solve highly complex
problem three is combination of neurons into layers permits for artificial neural network. In an
medical applications the neural network are like ANNs etc. and the medical application the
neural network are used to map an input into a desired output [5].
It is a new technique of detection of brain tumor and for very good result and accuracy.
The watershed method is combined with edge detection operation. The color brain MRI images
can be obtained by this algorithm. In this the RGB image is converts into on HSV color image so
that the image is separated in 3 regions which are known as hue, saturation and intensity. The
canny edge detector is applied is applied to an output image for rebuilt process of edge occurs in
this .at last combining the three images and the final resultant brain tumor segmented image is
obtained. This algorithm is applied on 20 brain MRI images for excellent result [6].
In an MRI image the highly irregular boundaries of tumor tissues is seen. For a
segmentation of medical image, the deformable modes and region base methods are used. The
main problems are there in MRI images like undefined location of tumor are unseen boundaries
or data loss at boundaries and a silent edge not extended. By using this algorithm the silent edge
is extended and found boundary of tumor location or area and once the boundary or location of
tumor is seen clearly. Then removal of tumor can be take place [7].
Mariam Saii, Zaid Kraitem (2017) in their research “Automatic Brain Tumor Detection in
MRI Using Image Processing Techniques” offers a fully automatic method for tumor
segmentation on Magnetic Resonance Images MRI. In this method, at first in the preprocessing
level, anisotropic diffusion filter is applied to the image by 8-connected neighborhood for
removing noise from it. [9] In the second step, Support Vector Machine SVM Classifier had
been used for tumor detection accurately. After creating appropriate mask image, based on its
symmetry in axial and coronary MRI, the tumor had been detected and segmented (Dice
coefficient > 0.90) in a few seconds. In short, the method applied on several MRI images with
different types had been regardless of the degree of complexity.
The International journal of analytical and experimental modal analysis
Volume XII, Issue II, February/2020
ISSN NO: 0886-9367
Page No:2451
6. Bahadure et.al (2017) in their article “Image Analysis for MRI Based Brain Tumor
Detection and Feature Extraction Using Biologically Inspired BWT and SVM” had studied on
the segmentation, detection, and extraction of infected tumor area from MRI images. It is the
primary concern and a tedious task performed by radiologists and other clinical experts. The
accuracy depends on the experience of the experts. So in the modern times, the use of computer
aided technology in medical field becomes very necessary to overcome such limitations. The
researchers had studied to improve the performance and condense the complexity involves in the
MRI image segmentation process. They have investigated Berkeley wavelet transformation
(BWT) based brain tumor segmentation in the study. In addition, to improve the accuracy and
quality rate of the support vector machine (SVM) based classifier; relevant features are extracted
from each segmented tissue. [10] The results of the proposed technique have been evaluated and
validated for performance and quality analysis on MR brain images, based on accuracy,
sensitivity, specificity, and dice similarity index coefficient. The experimental results achieved
96.51% accuracy, 94.2% specificity, and 97.72% sensitivity, demonstrating the effectiveness of
the proposed technique for identifying normal and abnormal tissues from brain MR images. [10]
Parasuraman Kumar, B. Vijay Kumar (2019) in their article “Brain Tumor MRI
Segmentation and Classification Using Ensemble Classifier” had discussed on Brain tumor and
abnormal formation within the brain. They stressed it is becoming a major cause for death in
many cases. The detection of these cells has been a difficult problem and MRI gives the solution
to it. It is very essential to compare brain tumor from the MRI treatment. It is very difficult to
have vision about the abnormal structures of human brain using simple imaging techniques.
According to them, Ensemble methods are the most influential development in Data Mining and
Machine Learning in early decades. They combine the procedure of neural network, extreme
learning machine (ELM) and support vector machine classifiers to identify the tumours
accurately. This system consists of manifold phases such as Preprocessing, segmentation, feature
extraction, and classification. This phase clearly classifies brain images into tumor and non-
tumors areas using Feed Forwarded Artificial neural network based classifier. Hence, these
experiments have proved robust to initialization, faster and accurate finding of the tumors. [11]
IV COMPARISON AND GROWTH OF YESTER RESEARCHES
Table-1 Techniques and Results in Brain Tumor Segmentation and Detection
Author Year Paper Name Technique Result
The International journal of analytical and experimental modal analysis
Volume XII, Issue II, February/2020
ISSN NO: 0886-9367
Page No:2452
7. P. Kleihues 1993
The new WHO
classification
of brain tumors
Brain Pathology
It gives different edge
distribution, Syptoes.
D. N. Louis 2007
The 2007 WHO
classification of tumors
of the
central nervous system
Detection of
CNS(Central
Nervous
System)
The molecular parameter
is used for its diagnosis
structure.
D. J.
Hemanth 2009
"Effective Fuzzy
Clustering Algorithm
for
Abnormal MR Brain
Image Segmentation
Abnormal MR
Brain Image
Segmentation
It gives abnormal MR
brain image
segmentation accurate
region of cancer and
better identification of
branch i.e. stage of cancer.
A. A.
Abdullah
2012
Implementation of an
improved cellular
neural network
algorithm for brain
tumor detection
Neural network
It solves high complex
problem and it is used to
map an input into a
desired output.
Maiti
and M.
Chakraborty
2012
A new method for brain
tumor segmentation
based on watershed and
edge detection
algorithms in HSV
color model
Watershed and
edge detection
algorithms in
HSV
color model
It gives color brain
MRI image foe very
good accuracy result.
S.Charutha
and M. J.
Jayashree
2014
An efficient brain tumor
detection byintegrating
modified texture based
region
growing and cellular
automata edge detection
Automated and
efficient brain
tumor detection
The proposed method
efficient in treatment
of brain tumor and
also in removal of
tumor.
R.Preetha
and G. R.
Suresh
2014
Performance Analysis of
Fuzzy C Means
Algorithm in Automated
Detection of Brain Tumor
Fuzzy C Means
Algorithm in
Automated
Detection of
Brain Tumor
The boundary of tissues
can be seen clearly.
Mariam
Saii, Zaid
Kraitem
2017
Automatic Brain Tumor
Detection in MRI Using
Image Processing
Techniques
Support Vector
Machine SVM
Classifier had
been used for
tumor detection
Using appropriate mask
image, based on its
symmetry in axial and
coronary MRI, the tumor
had been detected and
The International journal of analytical and experimental modal analysis
Volume XII, Issue II, February/2020
ISSN NO: 0886-9367
Page No:2453
8. segmented (Dice
coefficient > 0.90) in a few
seconds regardless of the
degree of complexity.
Bahadure
et.al
2017
Image Analysis for MRI
Based Brain Tumor
Detection and Feature
Extraction Using
Biologically Inspired
BWT and SVM
Berkeley
wavelet
transformation
(BWT) based
brain tumor
segmentation
and Support
Vector Machine
(SVM) based
classifier
The results of the proposed
technique have been
evaluated and validated for
performance and quality
analysis on MR brain
images, based on accuracy,
sensitivity, specificity, and
dice similarity index
coefficient.
Parasuram
an Kumar,
B. Vijay
Kumar
2019
Brain Tumor MRI
Segmentation and
Classification Using
Ensemble Classifier
Ensemble
methods for
Data Mining and
Machine
Learning. They
combine the
procedure of
neural network,
extreme learning
machine (ELM)
and support
vector machine
classifiers to
identify the tumor
It clearly classifies brain
images into tumor and non-
tumors areas using Feed
Forwarded Artificial neural
network based classifier
and faster, accurate in
finding the tumors.
V CONCLUSION
In this research article, different techniques to detect and segment Brain tumor from MRI
images are discussed and comparatively related to find the growth of segmentation and detection
techniques. To extract and segment the tumor, different techniques such as SOM Clustering, k-
mean clustering, Fuzzy C-mean technique, curvelet transform, Support Vector Machine SVM
Classifier, Berkeley wavelet transformation and Ensemble methods are used to get accurate
results. Hence, it is evident that detection of Brain tumor from MRI images can be done by
various methods, and in future different automatic methods like multi SVM techniques can be
The International journal of analytical and experimental modal analysis
Volume XII, Issue II, February/2020
ISSN NO: 0886-9367
Page No:2454
9. adopted to achieve more accuracy and more efficient results; and these developments can avoid
fatal causalities by accurate identification and target based proper diagnosis.
REFERENCES
[1] D.N.Louis, et al, “The 2007 WHO classification of tumor of central nervous system,” Act d
neuropathological, Vol 114, pp 97-109, 2007.
[2] P.Kleihues, et al. “The new WHO classification of brain tumor, brain pathology”, Vol 3, pp.
255-268, 1993.
[3] D.J.Hemanth, et al, “Effective fuzzy clustering algorithm for abnormal MR brain image
segmentation,” Advance Computing Conference 2009, pp-609-614.
[4] S.Chrutha and M.J.Jayashree, “An efficient brain tumor detection by integrating modified
texture based region growing and cellular automata edge detection,” Control
Instrumentation, Communication and Computational Technology (ICCICCT), 2014,
pp.1193-1199.
[5] A. Abdullah, et al., “Implementation of an improved cellular neural network algorithm for
brain tumor detection," Biomedical Engineering (Isobel), 2012, pp. 611- 615.
[6] I. Maiti and M. Chakra borty, “A new method for brain tumor segmentation based on
watershed and edge detection algorithms in HSV color model,” Computing and
Communication Systems (NCCCS), 2012, pp. 1-5.
[7] R. Preetha and G. R. Suresh, "Performance Analysis of Fuzzy C Means Algorithm in
Automated Detection of Brain Tumor," Computing and Communication Technologies
(WCCCT), 2014, pp. 30-33.
[8] Bandana Sharma et al. “Review Paper on Brain Tumor Detection Using Pattern Recognition
Techniques” International Journal of Recent Research Aspects, ISSN: 2349-7688, Vol. 3,
Issue 2, June 2016, pp. 151-156
[9] Mariam Saii, Zaid Kraitem. “Automatic Brain Tumor Detection in MRI Using Image
Processing Techniques.” Biomedical Statistics and Informatics. Vol. 2, No. 2, 2017, pp. 73-
76. doi: 10.11648/j.bsi.20170202.16
[10] Nilesh Bhaskarrao Bahadure et.al. “Image Analysis for MRI Based Brain Tumor Detection
and Feature Extraction Using Biologically Inspired BWT and SVM”. International Journal
of Biomedical Imaging, Volume 2017, Article ID 9749108,
The International journal of analytical and experimental modal analysis
Volume XII, Issue II, February/2020
ISSN NO: 0886-9367
Page No:2455
10. https://doi.org/10.1155/2017/9749108.
[11] Parasuraman Kumar, B. VijayKumar “Brain Tumor MRI Segmentation and Classification
Using Ensemble Classifier” International Journal of Recent Technology and Engineering
(IJRTE), ISSN: 2277-3878, Volume-8, Issue-14, June 2019.
The International journal of analytical and experimental modal analysis
Volume XII, Issue II, February/2020
ISSN NO: 0886-9367
Page No:2456