IRJET- An Effective Brain Tumor Segmentation using K-means Clustering

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 03 | Mar-2018 www.irjet.net p-ISSN: 2395-0072
© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 1015
AN EFFECTIVE BRAIN Tumor SEGMENTATION USING K-means
clustering
1K.Pravallika, 1G.Veena Lokeswari, 1B.Buelah, 1G.Venkata Sowmya, 2V.Purna Chandra Reddy.
1B.Tech Student, Dept of ECE, VVIT, Nambur, Guntur District, A.P., India.
2Assistant Professor, Dept of ECE, VVIT, Nambur, Guntur District, A.P., India.
----------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Image segmentation technology has always
been one of key technologies in image processing, in recent
years, many algorithms are applied in the field of image
segmentation. Image segmentation technology divides the
image into several areas, and needed information is
extracted. General segmentation technology are based on
threshold segmentation method, based on region
segmentation method, the segmentation method based on
edge, and the segmentation method based on the specific
theory. In this paper study the different clustering methods
in image segmentation. In view of the traditional clustering
image segmentation algorithm for image segmentation
accuracy is low problem, put forward a kind of fuzzy control
based on C-means clustering image segmentation method.
Methods firstly in clustering image segmentation algorithm
based on fast, using fuzzy C-means clustering algorithm for
image segmentation. The experimental results show that the
algorithm in clustering, to optimize the performance of the
same premise, image segmentation edge clear,
segmentation better than traditional clustering algorithm
for image segmentation..
Key Words: Image segmentation, fuzzy c-means
clustering algorithm and optimization
1.INTRODUCTION
Stroke or cerebrovascular accident is a disease which
affects the vessels that supply blood to the brain. The
blockage of the blood vessel and the bursts of the blood
vessels causes the brain stroke [1-2]. There are three main
kinds of stroke: Ischemic strokes (severely reduced blood
flow), Hemorrhagic strokes (leakage of blood) and
Transient ischemic attacks (TIAs-also referred to as mini-
strokes) [3, 4]. Ischemic brain stroke is one of the leading
causes of death and disability in major industrialized
countries [5]. To detect the Ischemic brain cells, the
computed tomography (CT) and Magnetic Resonance
Image (MRI) is the two important screening tests [6, 7].
The existing methods used the conventional threshold
techniques, but this method only has the limited accuracy
and repeatability [8, 9]. The clear image of the affected
region of the brain image is not provided by the above
techniques, but, for the early identification of the brain
infection, the MRI is more efficient than the CT [10].
The brain stroke detection techniques consolidate many
methods such as pre-processing, segmentation, feature
extraction and classification [11-13]. The researchers
proposed many techniques for the segmentation and
classification of images. A fuzzy clustering approach [14]
to the segmentation followed by 3D connected
components to build the tumor shape, Atlas-based medical
image segmentation techniques [15] which convert the
segmentation of an MR image into a non rigid registration
problem between the MR image of the patient and the MR
image used to create the brain atlas. Unfortunately, this
requires using a large seed that mask atlas structures,
potentially leading to erroneous results. The drawback of
FCM (Fuzzy C-Mean) [16] is that the spatial neighborhood
term is computed at each iteration step, which is very
time-consuming. Fast generalized FCM (FGFCM) algorithm
[17] which introduces a local similarity measure, but the
computational cost is very high in this method.
After extracting the features of the segmented
image, the researchers introduce many types of
classification techniques to classify the stroke and non-
stroke regions. The Artificial Neural Networks (ANN) [18]
which is used to anticipate the fate of ischemic tissues on
three different stroke groups. But it may suffer from an
overfitting problem. The Random Decision Forests [19] is
a popular classifier, but it can contain errors, noise and
image artifacts, which may lead to uncertainties.
1.1 Fuzzy C-means (FCM)
FCM clustering is an unsupervised method for the data
analysis. This algorithm assigns membership to each data
point corresponding to each cluster centre on the basis of
distance between the cluster centre and the data point.
The data point near to the cluster centre has more
membership towards the particular centre. Generally, the
summation of membership of each data point should be
equal to one. After each iteration, the membership and
cluster centres are updated according to the formula.

Where,
‘n’ is the number of data points
‘Vj’ represents the jth cluster centre
‘m’ is the fuzziness index m €[1,∞]
‘c’ represents the number of cluster centre
‘μij’ represents the membership of ith data to jth cluster
centre.
‘dij’represents the Euclidean distance between ith data
and jth cluster centre.
‘xi’ is the ith of d-dimensional measured data
‘cj’ is the d-dimension centre of the cluster
is any norm expressing the similarity between any
measured data and the centre.
The main objective of fuzzy c-means algorithm is to
minimize
Where, is the Euclidean distance between ith data
and jth cluster centre
1) Algorithmic steps for fuzzy C-means clustering:
Let X = {x1, x2, x3, ....xn}be the set of data points and V = {
v1, v2, v3, ....vc }be the set of cluster centres.
Step1: Randomly select ‘c’ cluster centres
Step2: Calculate the fuzzy membership ‘μij’using the
equation
Step3: Compute the fuzzy centres ‘vj’using
Step4: Repeat step2 and step3 until the minimum ‘J’ value
is achieved or ║U(k+1)-U(k)║< β
Where, ‘k’ is the iteration step
‘β’ is the termination criterion between [0,1]
is the fuzzy membership matrix
‘J’ is the objective function
The first loop of the algorithm calculates membership
values for the data points in clusters and the second loop
recalculates the cluster centres using these membership
values. When the cluster centre stabilizes the algorithm
ends.
The FCM algorithm gives best result for overlapped data
set and also gives better result than k-means algorithm.
Here, the data point can belong to more than one cluster
centre. The main drawback of FCM is 1) the sum of
embership value of a data point xi in all the clusters must
be one but the outlier points has more membership value.
So, the algorithm has difficulty in handling outlier points.
2) Due to the influence of all the data members, the cluster
centres tend to move towards the centre of all the data
points [Cox (2005)].
It only considers image intensity thereby producing
unsatisfactory results in noisy images [Hall et al. (1992)].
A bunch of algorithms are proposed to make FCM robust
against noise and in homogeneity but it’s still not perfect
[Hall et al. (1992), Lions et al. (1992), Acton et al. (2000),
Tolias and Panas (2008), Dave (1991), Zhang and Chen
(2004)].
2.Proposed k-means clustering
We have proposed segmentation of the brain MRI images
for detection of tumors using K-Means clustering
technique. A cluster can be defined as a group of pixels
where all the pixels in certain group defined by similar
relationship. Clustering is also unsupervised classification
because the algorithm automatically classifies objects
based on user given criteria. Here K-Means clustering
algorithm for segmentation of the image is used for tumor
detection from the brain MRI images. The proposed block
diagram is as shown.
Fig 1. Proposed Methodology
MRI scans of the human brain forms the input images for
our system where the grayscale MRI input images are
given as the input. The preprocessing stage will convert
the RGB input image to grayscale. Noise present if any, will
be removed using a median filter. The image is sharpened
using Gaussian filtering mask

The pre-processed image is given for image segmentation
using K-Means clustering algorithm.
Fig.2. Proposed K-means Clustering Algorithm
2.1.Experiment and Analysis
In order to test the performance of the algorithm, this
paper adopts brain MRI images of the Montreal
neurological institute for simulation. Traditional clustering
image segmentation result is shown in Fig.1, the noise
elimination of image using the median filter is shown in fig
2. The tradition approach for the detection of tumor is
implemented using K means Clustering is shown in fig 3.
Final tumor detection using K-means Clustering is shown
in fig 4. Traditional FCM image segmentation result is
shown in Fig.5, and image segmentation result of
proposed algorithm of section 2. The proposed fast image
segmentation is superior to the traditional algorithm, the
edge details is clear, this algorithm can ensure clustering
optimization performance unchanged, reduce the cost of
operation, and obviously improves the segmentation
efficiency.
Fig 3: input image and histogram for K-means Clustering
method
Fig4 noise elimination using Filtering Technique
Fig5: K-means Clustering processing for tumor detection
Fig6 Tumor Detection using K-means Clustering
3. CONCLUSIONS
Many image segmentation methods have been developed
in the past several decades for segmenting MRI brain
images, but still it remains a challenging task. A
segmentation method may perform well for one MRI brain
image but not for the other images of same type. Thus it is
very hard to achieve a generic segmentation method that
can be commonly used for all MRI brain images. In this
work, the merits and demerits of various automated
techniques for brain tumour identification is analyzed in
detail. Finally simulation is carried out for the two
clustering techniques i.e K-means and Fuzzy C means. The
Fuzzy techniquesgives the good performance for detection
of tumor in MRI images.

REFERENCES
[1] Y. Wang, A. Katsaggelos, X. Wang, and T. Parrish, "A
deep symmetry convnet for stroke lesion segmentation",
In: Proc. of the IEEE International Conference on Image
Processing (ICIP), Phoenix, AZ, USA, pp.111-115, 2016.
[2] S. Bauer, P. Gratz, J. Gralla, M. Reyes, and R. Wiest,
"Towards automatic MRI volumetry for treatment
selection in acute ischemic stroke patients", In: Proc. of the
36th Annual International Conference of the IEEE
Engineering in Medicine and Biology Society, Chicago, IL,
USA, pp. 1521-1524, 2014.
[3] N.H. Rajini and R. Bhavani, "Computer aided detection
of ischemic stroke using segmentation and texture
features", Measurement, Vol. 46, No. 6, pp. 1865-1874,
2013.
[4] N. Rost, S. Sadaghiani, A. Biffi, K. Fitzpatrick, L. Cloonan,
J. Rosand, D. Shibata, and T. Mosley, "Setting a gold
standard for quantification of leukoaraiosis burden in
patients with ischemic stroke: The Atherosclerosis Risk in
Communities Study", Journal of Neuroscience Methods,
Vol. 221, No. 1, pp. 196-201, 2014.
[5] N. Nabizadeh, N. John, and C. Wright, "Histogram-based
gravitational optimization algorithm on single MR
modality for automatic brain lesion detection and
segmentation", Expert Systems with Applications, Vol. 41,
No. 17, pp. 7820-7836, 2014.
[6] O. Maier, M. Wilms, J. von der Gablentz, U. Krämer, T.
Münte, and H. Handels, "Extra Tree forests for sub-acute
ischemic stroke lesion segmentation in MR sequences",
Journal of Neuroscience Methods, Vol. 240, No. 1, pp. 89-
100, 2015.
[7] M. Artzi, O. Aizenstein, T. Jonas-Kimchi, V. Myers, H.
Hallevi, and D. Ben Bashat, "FLAIR lesion segmentation:
Application in patients with brain tumors and acute
ischemic stroke", European Journal of Radiology, Vol. 82,
No. 9, pp. 1512-1518, 2013.
[8] M. Benders, N. van der Aa, M. Roks, H. van Straaten, I.
Isgum, M. Viergever, F. Groenendaal, L. de Vries, and F. van
Bel, "Feasibility and Safety of Erythropoietin for
Neuroprotection after Perinatal Arterial Ischemic Stroke",
The Journal of Pediatrics, Vol. 164, No. 3, pp. 481-486.e2,
2014.
[9] N. Ghosh, Y. Sun, B. Bhanu, S. Ashwal, and A. Obenaus,
"Automated detection of brain abnormalities in neonatal
hypoxia ischemic injury from MR images", Medical Image
Analysis, Vol. 18, No. 7, pp. 1059-1069, 2014.
[10] A. Patel, B. van Ginneken, F. Meijer, E. van Dijk, M.
Prokop, and R. Manniesing, "Robust cranial cavity
segmentation in CT and CT perfusion images of trauma
and suspected stroke patients", Medical Image Analysis,
Vol. 36, No. 2, pp. 216-228, 2017.
[11] J. Griffis, J. Allendorfer, and J. Szaflarski, "Voxel-based
Gaussian naïve Bayes classification of ischemic stroke
lesions in individual T1-weighted MRI scans", Journal of
Neuroscience Methods, Vol. 257, No. 1, pp. 97-108, 2016.
[12] N. Yassi, B. Campbell, B. Moffat, C. Steward, L.
Churilov, M. Parsons, P. Desmond, S. Davis, and A. Bivard,
"Know your tools—concordance of different methods for
measuring brain volume change after ischemic stroke",
Neuroradiology, Vol. 57, No. 7, pp. 685-695, 2015.
[13] B. Menze, K. Van Leemput, D. Lashkari, T. Riklin-
Raviv, E. Geremia, E. Alberts, P. Gruber, S. Wegener, M.
Weber, G. Szekely, N. Ayache, and P. Golland, "A Generative
Probabilistic Model and Discriminative Extensions for
Brain Lesion Segmentation— With Application to Tumor
and Stroke", IEEE Transactions on Medical Imaging, Vol.
35, No. 4, pp. 933-946, 2016.
[14] D. Chyzhyk, R. Dacosta-Aguayo, M. Mataró, and M.
Graña, "An active learning approach for stroke lesion
segmentation on multimodal MRI data", Neurocomputing,
Vol. 150, No. 6, pp. 26-36, 2015.
[15] L. Chen, P. Bentley, and D. Rueckert, "Fully automatic
acute ischemic lesion segmentation in DWI using
convolutional neural networks", NeuroImage: Clinical, Vol.
15, No. 6, pp. 633-643, 2017.
[16] S.A. Kumar and B. Harish, "A Modified Intuitionistic
Fuzzy Clustering Algorithm for Medical Image
Segmentation", Journal of Intelligent Systems, Vol. 0, No. 0,
pp. 1-15, 2017.
[17] K. Murphy, N. van der Aa, S. Negro, F. Groenendaal, L.
de Vries, M. Viergever, G. Boylan, M. Benders, and I. Išgum,
"Automatic quantification of ischemic injury on diffusion-
weighted MRI of neonatal hypoxic ischemic
encephalopathy", NeuroImage: Clinical, Vol. 14, No. 1, pp.
222-232, 2017.
[18] L. Mirtskhulava, J. Wong, S. Al-Majeed, and G. Pearce,
"Artificial Neural Network Model in Stroke Diagnosis", In:
Proc. of the 17th UKSim-AMSS International Conference
on Modelling and Simulation (UKSim), Cambridge, UK, pp.
50-53, 2015.
[19] I. Ibrahim and T. Khatib, "A novel hybrid model for
hourly global solar radiation prediction using random
forests technique and firefly algorithm", Energy
Conversion and Management, Vol. 138, No. 8, pp. 413-425,
2017.
[20] A. Subudhi, S. Jena and S. Sabut, "Delineation of the
ischemic stroke lesion based on watershed and relative
fuzzy connectedness in brain MRI", Medical & Biological
Engineering & Computing, Vol. 55, No. 1, pp. 1-13, 2017.

[21] O. Maier, B. Menze, J. von der Gablentz, L. Häni, M.
Heinrich, M. Liebrand, S. Winzeck, A. Basit, P. Bentley, L.
Chen, D. Christiaens, and F. Dutil, "ISLES 2015 - A public
evaluation benchmark for ischemic stroke lesion
segmentation from multispectral MRI", Medical Image
Analysis, Vol. 35, No. 1, pp. 250-269, 2017.
[22] N.H. Rajini and R. Bhavani, "Computer aided detection
of ischemic stroke using segmentation and texture
features", Measurement, Vol. 46, No. 6, pp. 1865-1874,
2013.
[23] M. Gong, Y. Liang, J. Shi, W. Ma, and J. Ma, "Fuzzy C-
Means Clustering With Local Information and Kernel
Metric for Image Segmentation", IEEE Transactions on
Image Processing, Vol. 22, No. 2, pp. 573-584, 2013.
[24] P.R. Filho, R. Sarmento, G. Holanda, and D. de Alencar
Lima, "New approach to detect and classify stroke in skull
CT images via analysis of brain tissue densities", Computer
Methods and Programs in Biomedicine, Vol. 148, No. 11,
pp. 27-43, 2017.
[25] P.R. Filho, E. Rebouças, L. Marinho, R. Sarmento, J.
Tavares, and V. de Albuquerque, "Analysis of human tissue
densities: A new approach to extract features from
medical images", Pattern Recognition Letters, Vol. 94, No.
10, pp. 211-218, 2017.

IRJET- An Effective Brain Tumor Segmentation using K-means Clustering

More Related Content

What's hot (19)

Similar to IRJET- An Effective Brain Tumor Segmentation using K-means Clustering (20)

More from IRJET Journal (20)

Recently uploaded (20)

IRJET- An Effective Brain Tumor Segmentation using K-means Clustering