International Journal of Computer Information Systems, Vol. 3, No.

3, 2011

Brain Tumour Detection Using Colour-Based K-Means Clustering Segmentation

P.Daniel Ratna Raju G.Neelima Dr.K.Prasada Rao
Principal, Chandrakala P.G College
Tenali, India prasudh@yahoo.com

HOD CSE Department CSE Dept Priyadarshini Institute of Technology & Science Acharya Nagarjuna University, Tenali, India Guntur, INDIA ratnaraju.daniel@gmail.com neelima_raju2003@yahoo.co.in

AbstractIn this paper we proposed a colour based segmentation method that uses the K-means clustering technique to track tumour objects in magnetic resonance (MR) brain images. The key concept in this colour-based segmentation algorithm with K-means is to convert a given gray-level MR image into a colour space image and then separate the position of tumour objects from other items of an MR image by using Kmeans clustering and histogram-clustering. Experiments demonstrate that the method can successfully achieve segmentation for MR brain images to help pathologists distinguish exactly lesion size and region. Keywords cluster, magnetic resonance, thresholding..

I. INTRODUCTION Imaging is an essential aspect of medical science to visualize the anatomical structures of the human body, several new complex medical imaging modalities, such as X-ray, magnetic resonance imaging (MRI), and ultrasound, strongly depend on computer technology to generate or display digital images. With computer techniques, multidimensional digital images of physiological structures can be processed and manipulated to help visualize hidden diagnostic features that are otherwise difficult or impossible to identify using planar imaging methods. Image segmentation is the partition of an image into a set of non-overlapping regions whose union is the entire image. In the simplest case, one would only haven object region and a background region. A region cannot be declared a segment unless it is completely surrounded by edge pixels. It is not an easy task to make it known to a computer what characteristics constitutes a meaningful segmentation. For this reason, a set of rules in general segmentation procedures is required: : Regions of an image segmentation should be uniform and homogeneous with respect to some characteristic (e.g. grey level or texture) . Region interiors should be simple and without many holes. Adjacent regions of a segmentation should have significantly varying values with respect to the characteristic on which they are uniform . Boundaries of each segment should be simple, not ragged, and must be spatially accurate.

Image processing is the process of identifying features in images and marking them as different from one another. These features can be things like gray and white matter in an MRI of the brain or individual organs in an MR or CT image. Magnetic resonance imaging(MRI) provides detailed images of living tissues, and is used for both brain and body human studies. Data obtained from MR image is used for detecting tissue deformities such as cancers and injuries: MR is also used extensively in studies of brain pathology, where regions of interest (ROIs) are often examined in detail, for example in multiple sclerosis (MS) studies. In order to perform good quantitative studies, ROIs within the brain must be well defined. In traditional methods, a skilled operator manually outlines the ROIs using a mouse cursor. More recently, computer-assisted methods have been used for specific tasks such as extraction of MS lesions from MRI brain scans, or extraction of the cerebral ventricles in schizophrenia studies. Many of these computers- assisted tasks require segmentation of the whole brain from the head. Classically, image segmentation is defined as the partitioning of an image into non-overlapping, constituent regions that are homogeneous with respect to some characteristic such as intensity or texture. Importance of Image Segmentation: Fully automatic brain tissue classification from magnetic resonance images (MRI) is of great importance for research and clinical studies of the normal and diseased human brain. Segmentation of medical imagery is a challenging problem due to the complexity of the images. Accurate and robust tissue classification is the basis for many applications such as the quantitative analysis of tissue volume in healthy and diseased populations. II.SEGMENTATION Segmentation methods can be classified in the following way: 1. Pixel intensity based methods. The intensity values of the pixel are used to segment the image. The space continuity is not frequently considered in this type of methods. Within this group, they stand out the method of classification of pixels, which uses a sort of statistical algorithms to assign a label pixel of the image. 2. Region based methods. The image segmentation of is based on the similarity of the adjacent intensities of pixels at image

September Issue

Page 34 of 75

ISSN 2229 5208

pixels. Region growth methods, where the regions begin being small and soon after, by an iterative process of growth, regions that have similar characteristics are merged. 3. Model based methods. It begins with a model that grows updated accordingly to the image characteristics of the image. Region based segmentation The shape of an object can be described in terms of its boundary or the region it occupies. Image region belonging to an object generally have homogeneous characteristics,e.g., similar in intensity or texture. Region-based segmentation techniques attempt to segment an image by identifying the various homogeneous regions that corresponds to different objects in an image.Unlike clustering methods; region-based methods explicitly consider spatial interactions between neighbouring voxels. In its simplest form, region growing methods usually start by locating some seeds representing distinct regions in the image. The seeds are then grown until they eventually cover the entire image. The region growing process is therefore governed by a rule that describe the growth mechanism and a rule that check the homogeneity of the regions at each growth step. Region growing technique has been applied to MRI segmentation. A semi automatic, interactive MRI segmentation algorithm was developed that employ simple region growing technique for lesion segmentation. In an automatic statistical region growing algorithm based on a robust estimation of local region mean and variance for every voxel on the image was proposed for MRI segmentation. The best region growing parameters are automatically found via the minimization of a cost functional. Furthermore, relaxation labelling, region splitting and constrained region merging were used to improve the quality of the MRI segmentation.The determination of an appropriate region homogeneity criterion is an important factor in region growing segmentation methods. Edge detection segmentation MR image segmentation based on edge detection has been proposed, where a combination of Marr-Hildreth operator for edge detection and morphological operations for the refinement of the detected edges is used to segment 3D MR images. A boundary tracing method is proposed, where the operator clicks a pixel in a region to be outlined and the method then finds the boundary starting from that point. The method is however, restricted to segmentation of large, well defined structures but not to distinguish fine tissue types. Edge based segmentation methods usually suffers from over or under segmentation, induced by improper threshold selection. In addition, the edges found are usually not closed and complicated edge linking techniques are further required.

International Journal of Computer Information Systems, Vol. 3, No.3, 2011 corresponding to the different modes. A thresholding procedure attempts to determine an intensity value, called the threshold, which separates the desired classes. The segmentation is then achieved by grouping all pixels with intensities greater than the threshold into one class and all other pixels into another class. Determination of more than one threshold value is a process called multi thresholding. Thresholding is a simple yet often effective means for obtaining a segmentation of images in which different structures have contrasting intensities or other quantifiable features. The partition is only generated interactively, although automated methods do exist. Thresholding is often performed interactively based on the operators visual assessment of the resulting segmentation. Thresholding is often used as an initial step in a sequence of image-processing operations. It has been applied in digital mammography, in which two classes of tissue are typically present- healthy and tumours. Its main limitations are that, in its simplest form, only two classes are generated, and it cannot be applied to multi channel images. In addition, thresholding typically does not take into account the spatial characteristics of an image. This causes it to be sensitive to the noise and intensity in homogeneities, which can occur in MR images. Both of these artefacts essentially corrupt the histogram of the image, making separation more difficult. For these reasons, variations on classical thresholding have been proposed for medicalimage segmentation that incorporate information based on local intensities and connectivity.

IV. CLUSTERING Clustering algorithms essentially perform the same function as classifier methods without the use of training data. Thus, they are termed unsupervised methods. To compensate for the lack of training data, clustering methods iteratively alternate between segmenting the image and characterizing the properties of each class. In a sense, clustering methods train themselves, using the available data. Three commonly used clustering algorithms are the K-means or isodata algorithm, the fuzzy C-means algorithm and the expectation maximization (EM) algorithm. The K-means clustering algorithm clusters data by iteratively computing a mean intensity for each class and segmenting the image by classifying each pixel in the class with the closest mean. Figure 1(b) shows the result of applying the K means algorithm to a slice of an MR brain image in Figure 4(a). The number of classes was assumed to be three representing (from gray to white in figure 1) cerebrospinal fluid, grey matter, white matter. The fuzzy c means algorithm generalizes the K -means algorithm, allowing for soft segmentation based on fuzzy set theory. The EM applies the same clustering principles with underlying assumption that the data follow a Gaussian mixture model

III. THRESHOLDING Thresholding approaches segment scalar images by creating a binary partitioning of the image intensities. The histogram of a scalar image that possess different apparent classes,

September Issue

Page 35 of 75

ISSN 2229 5208

Fig.1 Segmentation of magnetic resonance brain image.(a) Original image (b) Segmentation using K-means algorithm (c) Segmentation using K-means algorithm with a Markov random field prior.

The posterior probabilities and computing maximum likelihood estimates of the means, covariances and mixing coefficients of the mixing model. Although clustering algorithms do not require training data, they do require an initial segmentation (or equivalently initial parameters) . The EM algorithm has demonstrated greater sensitivity to initialization than the K-means or fuzzy C-means algorithm. Like classifier methods clustering algorithms do not directly incorporate spatial modelling and can therefore be sensitive to noise and intensity in homogeneities. This lack of spatial modelling, however, can provide significant advantages for fast computation. Work on improving the robustness of clustering algorithms to intensity in homogeneities in MR images has demonstrated excellent success.

V. K-MEANS ALGORITHM K-means clustering is a key technique in pixel-based methods. Because pixel-based methods based on K-means clustering are simple and the computational complexity is relatively low compared with other region-based or edge-based methods, the application is more practicable. Furthermore, K-means clustering is suitable for biomedical image segmentation as the number of clusters is usually known for images of particular regions of the human anatomy. Many researchers have proposed related research into K-means clustering segmentation. The improvements achieved have been

International Journal of Computer Information Systems, Vol. 3, No.3, 2011 remarkable, but more computational complexity and extra software functionality are required. The K-means algorithm is an iterative technique that is used to partition an image into K -clusters. The basic algorithm is: 1. Pick K cluster centres, either randomly or based on some heuristic. 2. Assign each pixel in the image to the cluster that minimizes the variance between the pixel and the centre 3. Re-compute the centres by averaging all of the pixels in the cluster 4. Repeat steps 2 and 3 until convergence is attained (eg., no pixels change clusters) In this case,variancies the squared or absolute difference between a pixel and a cluster centre. The difference is typically based on pixel colour, intensity, texture, and location, or a weighted combination of these factors. K can be selected manually, randomly, o r b y a heuristic. This algorithm is guaranteed to converge, but it may not return the optimal solution. The quality of the solution depends on the initial set of clusters and the value of K. The other approach to partition an image into K cluster is the statistical hierarchical agglomerative clusterization technique for identification of images regions by the colour similarity. This method uses a binary mask and ranks the colour components of the clusters central components. The basic algorithm is: 1. Each pixel is a separate cluster. 2. The clusters with the same masks joins into new clusters. 3. New clusters a r e s e t u p b y t h e cluster integration with minimum distance. The stage may occur until the condition for clusters comparab ility is being obeyed. This is condition based on the binary mask of correlation and ranks. Histogram Based Segmentation In an image processing context, the histogram of an image normally refers to a histogram of the pixel intensity values. The histogram is a graph showing the number of pixels in an image at each different intensity value found in that image. .For an 8-bitgrayscale image there are 256 different possible intensities and so the histogram will be graphically display 256 numbers showing the distribution of pixels amongst those grey scale values. Histograms can also be taken of colour imageseither individual histogram of red, green and blue channels can be taken\, or a 3-Dhistogram can be produced, with the three axes representing the red, blue and green channels and brightness at each point representing the pixel count. The exact output from the operation depends upon the implementation it may simply be a picture of the required histogram in a suitable image format, or it may be a data file of some sort representing the histogram statistics. The operation is very simple. The image is scanned in a single pass and a running count of the number of pixels found at each intensity value is kept. This is then used to construct a suitable histogram. Histograms have many uses, one of the more common is to decide what value of threshold is to be used when converting a grey scale image to a binary one by thresholding. If the image is suitable for

September Issue

Page 36 of 75

ISSN 2229 5208

thresholding then the histogram will be bi-model i.e., the pixels intensities will be clustered around two well separated values. A suitable threshold for separating these two groups will be found somewhere in between the two peaks in the histogram. If the distribution is not like this then it is unlikely that a good segmentation can be produced by thresholding. Thresholding is used to segment an image by setting all pixels whose intensity values are above a threshold to a foreground value and all the remaining pixels to a background value. Simply speaking k-means clustering is an algorithm to classify or to group your objects based on attributes/features into K-number of group K is a positive integer number. The grouping is done by minimizing the sum of squares of distances between data and corresponding cluster centroid. Thus the purpose of K-mean clustering is to classify the data. The basic step of k-means clustering is simple. In the beginning we determine number of cluster K and we assume the centroid or centre of these clusters. We can take any random objects as the initial centroids or the first K objects in sequence can also serve as the initial centroids or the first K objects in sequence can also serve as the initial centroids. The K-means algorithm will do the four steps below until convergence. 1. Determine the centroid co-ordinate. 2. Determine the distance of each object to the centroid. 3. Group the object based on minimum distance. Step 1 Begin with a decision on the value of k= number of clusters Step 2 Put any initial partition that classifies the data into K clusters. You may assign the training samples randomly or systematically. Step 3 Take each sample in sequence and compute its distance from the centroid of each of the clusters. If a sample is not currently in the cluster with the closest centroid switch this sample to that cluster and update the centroid of that cluster gaining the new sample and the cluster losing the sample. Step 4 . Repeat step 3 until convergence is achieved, that is until a pass through the training sample causes no new assignments. We combine histogram statistics and K-means clustering to track the tumour objects in MR brain images. Basically, feature space selection is a key issue in K-means clustering segmentation. The original MR brain image is rendered as a gray-level image that is insufficient to support fine features. To obtain more useful feature and enhance the visual density, the proposed method applies pseudo-colour transformation, a mapping function that maps a gray-level pixel to a colourlevel pixel by a lookup table in a predefined colour map. An RGB colour map contains R, G, and B values for each item. Each gray value maps to an RGB item. The proposed method has adopted the standard RGB colour map, which gradually maps gray-level values 0 to 255 into blue-to-green-to-red colour. To retrieve important features to benefit the clustering

International Journal of Computer Information Systems, Vol. 3, No.3, 2011 process, the RGB colour space is further converted to a CIELab colour model (L*a*b*) [2]. The L*a*b* space consists of a luminosity layer L*, a chromaticity-layer a*, which indicates where colour falls along the red-green axis, and a chromaticity-layer b*, which indicates where the colour falls along the blue-yellow axis. The translating formula calculates the tri-stimulus coefficients first as: W=0.4303R+0.341G+0.1784B, Y=0.2219R+0.7068G+0.0713B, Z=0.0202R+0.1296G+0.9393B. The CIELab colour model is calculated as L*=116(h(Y/Ys))-16, a*=500(h(W/Ws))-h(Y/Ys), b*=200(h(Y/Ys)-h(Z/Zs)) ,h (q)={3sqrt(q) q>0.0088567 {, 787q+16/116 q<=0.008856 Where YS, WS, and ZS are the standard stimulus coefficients. Both the a* and b* layers contain all required colour information. Therefore, the proposed method then classifies the colours in the a*b* space using K-means clustering. After the clustering process, the cluster containing an area of interest is selected as the primary segment. To eliminate the pixels which are not related to the interest in the selected cluster, histogram clustering is applied by luminosity feature L* to derive the final segmented result.

VI. EXPERIMENTAL EVALUTION We convert a gray-level MR brain image into an RGB colour image first and then convert the RBG colour image into a CIELab colour model. Therefore, colours in both the a* and b* spaces are feature vectors for K-means clustering. To demonstrate the detection performance of the proposed method, an MR brain image (288 x 288) containing the pathological change area shown in Fig. 2(a) served as a test image. The RGB colour image converted from the gray-level image is shown in Fig. 2(b). To prove that the feature vectors proposed by our method really can provide better segmentation performance, two different data sets were prepared: the gray feature vectors of the original MR brain image and the RGB features derived from the converted RGB colour image. Generally, an MR brain image consists of regions that represent the bone, soft tissue, fat and background. In these two different data sources, visual judgments from the gray and colour test images suggest three primary clusters in the test image shown in Fig. 2(a) when k=3. Figs. 3(a)(c) show the image labelled by cluster index from the K-means process for different kinds of feature vectors. Using index labels, we can separate objects in the brain image by three colours: white, gray, and black. The final segmentation results generated by histogram clustering are shown in Figs. 3(d)(f). By combining Fig. 3(a) and Fig. 3(d), we can see that not only a tumour (in the right half of the image) is identified but also the white matter, cerebrospinal fluid, and the ventricles are. In other words, the segmentation result cannot exactly identify

September Issue

Page 37 of 75

ISSN 2229 5208

International Journal of Computer Information Systems, Vol. 3, No.3, 2011 the position of the tumour shown in Fig. 3(d). The same phenomenon exists in Fig. 3(e). However, the segmentation result generated by the proposed method can ignore most of the white matter, cerebrospinal fluid, and ventricles while exactly locating the position of the tumour. Therefore, the segmentation performance of the proposed features derived from the CIELab colour model and the proposed method is confirmed.

(e)

(f)

Fig. 3. Comparison of segmentation results. (a) image labelled by cluster index of a gray image; (b) image labelled by cluster index of an RGB image; (c) image labelled by cluster index of the proposed method; (d) final Segmentation of (a); (e) final segmentation of (b); (f) final segmentation of (c).

(a)

(b)

Fig.2 Original MR brain images. (a) gray -level image, (b) colour image.

(a)

(b)

(c)

(d)

In this project three approaches for segmentation of Brain tissue in MR images is presented. The results show that this method can successively segment a tumour provided the parameters are chosen properly. The visualization and detective valuations of the results of the segmentation show the success of the approaches. In this study, the tumour identification and the investigation are carried out for the potential use of MRI data for improving the tumour shape and 2D visualization of the surgical planning. In this project, the problem of the region segmentation in MR images due to the brain image complexity has been addressed using seed point and region growing algorithm. Conventionally, for brain MR image segmentation, there are various structural and statistical approaches. Structural methods are based on the spatial properties of the image, such as edges and regions. Region growing is a structural approach that is used for accurate segmentation of tumour. Future scope: Future research in MRI segmentation should strive toward improving the accuracy, precision and computation speed of the segmentation algorithm, while reducing the amount of manual interactions needed. This is particularly important as MR imaging is become a routine diagnostic procedure in clinical practice. It is also important that any practical segmentation algorithm should deal with 3D volume segmentation instead of 2D slice by slice segmentation, since MRI data is 3D in nature. Volume segmentation ensures continuity of the 3D boundaries of the segmented images where as slice by slice segmentation doesnt guarantee continuation of the boundaries of the tissue regions between slices. Future work will focus on the construction of more accurate brain atlas using anon rigid registration technique and on the investigation of the partial volume effect.

September Issue

Page 38 of 75

ISSN 2229 5208

International Journal of Computer Information Systems, Vol. 3, No.3, 2011 VII. CONCLUSION A colour-based segmentation method based on K-means clustering for tracking tumour in the MRI brain image is proposed. A preliminary experiment conducted on the MRI brain image demonstrates encouraging results. By using the features derived from CIELab colour model can provide good segmentation performance with the proposed method and the location of a tumour or lesion can be exactly separated from the colored image. In addition, the proposed method simply combines colour translation, K-means clustering and histogram clustering, thus making it efficient and very easy to implement.
8. F. Dabek, M.F. Kaashoek, D. Karger, R. Morris, and I. Stoica, Wide-Area Cooperative Storage with CFS, Proc. 18th ACM Symp. Operating System Principles (SOSP 01), Oct. 2001. 9. The Gnutella Home Page, Gnutella, http://gnutella.wego.com/, 2008. 10. E.J. Goh, H. Shacham, N. Modadugu, and D. Boneh, SiRiUS: Securing Remote Untrusted Storage, Proc. 10th Ann. Network and Distributed System Security Symp. (NDSS), 2003. 11. J. Kubiatowics, D. Bindel, Y. Chen, S. Czerwinski, P. Eaton, D. Geels, R. Gummadi, S. Rhea, H. Weatherspoon, W. Weimer, C. Wells, and B. Zhao, OceanStore: An Architecture for Global-Scale Persistent Storage, Proc. Ninth Intl Conf. Architectural Support for Programming Languages and Operating Systems (ASPLOS 00), Nov. 2000. 12. S. Ratnasamy, P. Francis, M. Handley, R. Karp, and S. Shenker, A Scalable Content-Addressable Network, Proc. ACM SIGCOMM 01, Aug. 2001. 13. A. Rowstron and P. Druschel, Pastry: Scalable, Distributed Object Location and Routing for Large-Scale Peer-to-Peer Systems, Proc. 18th IFIP/ACM Intl Conf. Distributed Systems Platforms (Middleware 01), Nov. 2001. 14. I. Stoica, R. Morris, D. Karger, M. Kaashoek, and H. Balakrishnan, Chord: A Scalable Peer-to-Peer Lookup Service for Internet Applications, Proc. ACM SIGCOMM 01, Aug. 2001

REFERENCES
1. Chen, C. W, Luo, J., Parker, K. J., Image Segmentation via Adaptive Kmean Clustering and Knowledge-based Morphological Operations with Biomedical Applications, IEEE Transactions on Image Processing, Vol. 7, No, 12, Dec. 1998, pp. 1673-1683. 2. Christine, C., Thomas, F., A Study of Efficiency and Accuracy in the Transformation from RGB to CIELAB Colour Space, IEEE Transactions on Image Processing, Vol. 6, No. 7, July 1997, pp. 1046-1048. 3. Dhawan, A. P., A Review on Biomedical Image Processing and Future Trends, Computer Methods and Programs in Biomedicine, Vol. 31, No.3-4, 1990, pp.141-183. 4. Gonzalez, R. C.; Woods, R. E., Digital Image Processing, 2nd ed., Prentice-Hall, Englewood Cliffs, NJ, 2002. 5. Ng, H. P., Ong, H. H., Foong, K. W. C., Goh, P. S., Nowinski, W. L., Medical Image Segmentation Using KMeans Clustering and Improved Watershed Algorithm, IEEE Southwest Symposium on Image Analysis and Interpretation, 2006, pp. 61-65. 6. Tsai, C. S., Chang, C. C., An Improvement to Image Segment Based on Human Visual System for Object-based Coding, Fundamental Informaticae, Vol. 58, No. 2, 2004, pp. 167-178.

Author : Mr.P.D.Ratna Raju has completed his post graduation in MCA., M.Tech. He contributed to research work in computer science thru his paper in M.Phil. At present he is working as Associate Prof, CSE Dept. in Priyadarshini Institute of Technology & Science, Tenali. He has vast teaching experience and more enthusiastic to contribute his knowledge and work in research. He is actively participating in research work in Ph.D from Dravidian University, Kuppam.

7. A. Adya, W. Bolosky, M. Castro, G. Cermak, R. Chaiken, J.R. Douceur, J. Howell, J.R. Lorch, M. Theimer, and R.P. Wattenhofer, Farsite: Federated, Available and Reliable Storage for an Incompletely Trusted Environment, Proc. Fifth Symp. Operating Systems Design and Implementation (OSDI), 2002.

September Issue

Page 39 of 75

ISSN 2229 5208