brain_4

Performance Analysis of Unsupervised Clustering Methods for Brain Tumor

Segmentation

Tushar H Jaware and Dr. K B Khanchandani
Department of E&TC, Shri Sant Gajanan Maharaj College of Engineering,

Shegaon, MS, India

Abstract:
Medical image processing is the most challenging and emerging field of neuroscience. The
ultimate goal of medical image analysis in brain MRI is to extract important clinical features that
would improve methods of diagnosis & treatment of disease. This paper focuses on methods to
detect & extract brain tumour from brain MR images. MATLAB is used to design, software tool for
locating brain tumor, based on unsupervised clustering methods. K-Means clustering algorithm is
implemented & tested on data base of 30 images. Performance evolution of unsupervised clustering
methods is presented.

Keywords: MRI, clustering, tumor

1. Introduction
The influence and impact of digital images on modern society is tremendous, and image

processing is now a critical component in science and technology. The rapid progress in
computerized medical image reconstruction, and the associated developments in analysis methods
and computer-aided diagnosis, has propelled medical imaging into one of the most important sub-
fields in scientific imaging.

In past few years, MRI has drawn considerable attention for its possible role in tissue
characterization. Brain tumor is an abnormal mass of tissue in which cells grow and multiply
uncontrollably, seemingly unchecked by the mechanisms that control normal cells. Image
segmentation is a process to identify regions of interest from digital images. In brain tumor studies,
accurate and reproducible segmentation and characterization of abnormalities are not
straightforward. For instance, a major problem in tumor treatment planning and evaluation is
determination of the tumor extent.

Figure 1. Diagnosis Rate in different Countrie

In MRI images, the amount of data is too much for manual segmentation. The procedure is

tedious, time, labor consuming, subjective and requires expertise. This gave way to methods that are
computer-aided with user interaction at varying levels. These methods are automatic and objective
and the results are highly reproducible. We designed software tool for locating brain tumor, based
on unsupervised clustering methods and analyzed its performance.

BRAIN. Broad Research in Artificial Intelligence and Neuroscience

Volume 4, Issues 1-4, October 2013, ISSN 2067-3957 (online), ISSN 2068 - 0473 (print)

2. Material and methods
In this paper we use luminosity-based segmentation method. This paper analyses k mean

clustering techniques to locate tumors in brain MRI. The input is axial view of the human brain.
The contrast of the gray level MR image is adjusted and then clustering algorithm is applied. The
position of tumor objects is separated from other items of an MR image by using clustering
algorithms and histogram-clustering.

After the clustering process, the cluster containing the tumor is selected as the primary
segment. Histogram clustering is applied to eliminate the pixels which are not related to the tumor
pixels. The performance of the clustering algorithms is found based on volumetric analysis of tumor
& computational speed by applying to a database of 30 images.

3. Algorithm
The algorithm has two stages, first is pre-processing of given MRI image and after that

segmentation and then perform morphological operations.

Figure 2. Stages of software implementation

4. Contrast Enhancement
Due to the less contrast of MR images, first we have to increase the contrast of an image.

The three techniques can be used for this:
� Imadjust increases the contrast of the image by mapping the values of the input intensity image to new

values such that, by default, 1% of the data is saturated at low and high intensities of the input data.
� Histeq performs histogram equalization. It enhances the contrast of the images by transforming the

values in an intensity image so that the histogram of the output image approximately matches a
specified histogram(uniform distribution by default)

� Adapthisteq performs contrast-limited adaptive histogram equalization. Unlike histeq, it operates on
small data regions (tiles) rather than the entire image. Each tile’s contrast is enhanced so that the
histogram of each output region approximately matches the specified histogram (uniform distribution by
default). The contrast enhancement can be limited in order to avoid amplifying the noise which might be
present in the image.

5. Clustering Algorithms
This system is implemented using morphological operations and the algorithms i.e. K-

Means algorithm

5.1. K-Means clustering:
K-means is one of the simplest unsupervised learning algorithms that solve the well known

clustering problems. The procedure follows a simple and easy way to classify a given data set
through a given no. of clusters (assume k clusters) fixed a priority. The main idea is to define k
centroids, one for each clusters. These centroids should be placed in a cunning way because of
different location causes different results. So the better choice is to place them as much as possible
far away from each other. The next step is to take each point belonging to given data set and
associate it to the nearest centroid. When no point is pending, the first step is completed and an
early group age is done. At this point, we need to re-calculate k new centroid as barycentres of the
cluster resulting from the previous step. After we have these k new centroids, a new binding has to
be done between the same data set points and the nearest new centroid. A loop has been generated.
As a result of this loop, we may notice that the k centroid change their location step by step until no

T. H Jaware & K B Khanchandani - Performance Analysis of Unsupervised Clustering Methods for Brain Tumor

Segmentation

more changes are done. In other word, centroids do not move any more. Finally, this algorithm aims
at minimizing an objective function, in this case a squared error function. The objective function

Where is chosen distance measure between a data point and the cluster centre is
an indicator of the distance of the n data points from their respective cluster centre.

5.2. K-Means clustering Algorithm:
No. of cluster= k

1. Pick k cluster centre, either randomly
2. Assign each pixel in the image to the cluster that minimizes the variance between the pixel and
the cluster centre

3. Re-compute the cluster centre by averaging all of the pixels in the cluster.
4. Repeat steps 2 &3 until convergence is attained (e.g.no pixels change clusters)

6. Cluster Selection
After the clustering process, the cluster containing an area of interest (tumor) is selected as

the primary segment.

7. Histogram Based Clustering
To eliminate the pixels which are not related to the interest in the selected cluster, histogram

clustering is applied.

8. Region Elimination
The output of histogram clustering consists of tumor region as well as the other regions

which has the same luminance and colour values as the tumor. The regions which are smaller than
the tumor are eliminated.

9. Discussion
In the proposed paper, we locate the brain tumor in MR image using morphological

operations and the clustering methods. We designed the graphical user interface (GUI) for user
friendly environment. The results obtained for this clustering method are given below:

10. Results for K-Mean clustering:

Figure 3 shows three different original brain MR images, contrast enhancement of the
images, segmented images using K-means algorithm and finally located tumor. Fig 1.4 shows the
performance of the unsupervised clustering methods with the no. of tumor pixels and execution
time to locate the brain tumor.

BRAIN. Broad Research in Artificial Intelligence and Neuroscience

Volume 4, Issues 1-4, October 2013, ISSN 2067-3957 (online), ISSN 2068 - 0473 (print)

(a) (b) (c) (d)

Figure 3:(a) Input MR Image (b) Enhanced Image (c) Segmented Tumor (d) Located brain tumor

Figure 4.Performance based on no. of tumor pixel & execution time

In this paper we segmented the brain tumors in axial view of MR images with the help of

unsupervised clustering method i.e. K-means clustering. The unsupervised clustering methods gave
the better results than traditional method.

The performance analysis and comparison is done f on the basis of no. of tumor pixels in
segmented brain tumor and the execution time for the same. Regarding the no. of tumor pixels, K-
means clustering gave a better result than the other methods. The clustering algorithms were tested
with a data base of 20 MRI brain images. K-means clustering achieved almost 90%result.

References
[1]M. Mancas, B. Gosselin, B. Macq, 2005, "Segmentation Using a Region GrowingThresholding",
Proc. of the Electronic Imaging Conference of the International Society for Optical Imaging
(SPIE/EI 2005), San Jose (California, USA).
[2] Dong-yong Dai; Condon, B.; Hadley, D.; Rampling, R.; Teasdale, G.; "Intracranial deformation
caused by brain tumors: assessment of 3-D surface by magnetic resonance imaging"IEEE
Transactions on Medical Imaging Volume 12, Issue 4, Dec. 1993 Page(s):693 – 702
[3] Matthew C. Clark “Segmenting MRI Volumes of the Brain With Knowledge- Based Clustering”
MS Thesis, Department of Computer Science and Engineering, University of South Florida, 1994
[5] http://noodle.med.yale.edu
[6] http://documents.wolfram.com/

T. H Jaware & K B Khanchandani - Performance Analysis of Unsupervised Clustering Methods for Brain Tumor

Segmentation

[7] Dzung L. Pham, Chenyang Xu, Jerry L. Prince;"A Survey of Current Methods in Medical
Medical Image Segmentation" Technical Report JHU / ECE 99-01, Department of Electrical and
Computer Engineering. The Johns Hopkins University, Baltimore MD 21218, 1998.
[8] M. Sezgin, B. Sankur " Survey over image thresholding techniques and quantitative
performance evaluation" J. Electron. Imaging 13 (1) (2004) 146-165.
[9] Chowdhury, M.H.; Little, W.D,;"Image thresholding techniques" IEEE Pacific Rim Conference
on Communications, Computers, and Signal Processing, 1995. Proceedings. 17-19 May 1995
Page(s):585 – 589
[10] Zhou, J.; Chan, K.L.; Chong, V.F.H.; Krishnan, S.M “Extraction of Brain Tumor from MR
Images Using One-Class Support Vector Machine” 27th Annual International Conference of the
Engineering in Medicine and Biology Society, 2005. IEEE-EMBS 2005, Page(s):6411 – 6414
[11]T. Logeswari, M. Karnan, ―An improved implementation of brain tumor detection using
segmentation based on soft computingǁ, Page(s): 006-014, Journal of Cancer Research and
Experimental Oncology Vol. 2(1), March 2010.