American International Journal of Research in Formal, Applied & Natural Sciences Available online at http://www.iasir.net ISSN (Print): 2328-3777, ISSN (Online): 2328-3785, ISSN (CD-ROM): 2328-3793 AIJRFANS is a refereed, indexed, peer-reviewed, multidisciplinary and open access journal published by International Association of Scientific Innovation and Research (IASIR), USA (An Association Unifying the Sciences, Engineering, and Applied Research) Survey of brain tumor detection techniques through MRI images Megha A Joshi 1, Prof. D.H.Shah 2 1 PG Student, Instrumentation and Control, Associate Professor 2 1,2 L.D College of Engineering, Navrangpura, Ahmedabad, Gujarat 380015, INDIA. Abstract: The images from the medical imaging technologies like MRI, US, CT are more complex to understand and noisy. Here, the area of interest is tumor detection in brain MRI Images. Today s modern medical imaging research faces the challenge of detecting brain tumor through Magnetic Resonance Images (MRI). For tumor detection in brain MRI image segmentation is an important part. Segmentation is the important tool in medical image processing which helps to make a simple format of medical image which is easier and meaningful to analyse. Hence, it is highly necessary that segmentation of the MRI images must be done accurately before asking the computer to do the exact diagnosis. Earlier, a variety of algorithms were developed for segmentation of MRI images by using different tools and techniques. However, this paper presents a comprehensive review of the methods and techniques used to detect brain tumor through MRI image segmentation. Keywords: Brain tumor detection, Magnetic resonance image, edema, image segmentation. I. Introduction In the present days, for the human body anatomical study and for the treatment planning medical science is very much depend on the medical imaging technology and medical images[1]. Magnetic resonance (MR) imaging and computer tomography (CT) scanning of the brain are the two most common tests undertaken to confirm the presence of brain tumor and to identify its location for selected specialist treatment options. Specifically for the human brain, MRI widely using. But by nature medical images are complex and noisy [1]. A tumor is a mass of tissue that's formed by an accumulation of abnormal cells [10]. Brain tumor is an abnormal mass of tissue in which cells grow and multiply uncontrollably, seemingly unchecked by the mechanisms that control normal cells [2]. So, it is very hard to detect tumor in early stage, since accurate measurements in brain tumor diagnosis are quite difficult because of diverse shapes, sizes, appearances of tumor, position of tumor in the brain but once it gets identified the treatment can be done and is curable with technique like chemotherapy, radiotherapy[5]. During the last few years brain tumor segmentation in MRI has become an emergent research field of medical image processing. MRI is an effective tool that provides detailed information about the targeted brain tumor anatomy, which in turn enables effective diagnosis and treatment [5]. This paper presents a review of the methods and techniques used during brain tumor detection through MRI image segmentation. II. Literature Review A. Seed region growing method: It is a simple region-based image segmentation method. Seed based region growing performs a segmentation of an image with respect to a point, known as seed [4]. This approach to segmentation examines neighbouring pixels of initial seed points and determines whether the pixel neighbours should be added to the region. Figure 1 segmented tumor[1] AIJRFANS 15-210; 2015, AIJRFANS All Rights Reserved Page 9
The basic formulation or mathematical description for Region-Based Segmentation is as following [1]. Where, is connected region,. It requires seed points in a region for all.this indicates that the regions must be disjoint for all For example: if all pixels in have the same gray level for any adjacent region & is a logical predicate defined over the points in set and is the null set. This region growing segmentation method without pre-processing require three or four seed point. Manual indication of the seed point with a marker is imprecise in this case [2]. In the case of improved segmentation method one seed point is enough for the appropriate segmentation [1]. B. Thresholding method: Thresholding is the simplest method. This method is on gray level intensity value of pixels [5]. Thresholding is the procedure to determine an intensity value, called the threshold, which separates the desired classes. And segmentation is done by grouping all pixels with intensity greater than the threshold into one class, and all other pixels into another class. It selects a proper threshold and then divide image pixels into many regions and spate objects form the background. Any pixel (x, y) is taken as the part of the object and provided that its intensity is greater than or equal to the threshold value. More precisely, image segmentation is the process of assigning a label to every pixel in an image such that pixels with the same label share certain visual characteristics. Generally, the threshold selection is done interactively. To get the exact shape and size of tumor, after segmentation the morphological operation and subtraction are used. Figure 2 Final tumor detected region[5] C. Watershade segmentation: The basic principle is to transform the gradient of a grey level image in a topographic surface. Where the values of f (m, n) are interpreted as heights and each local minima embedded in an image is referred as catchments basins. Figure 3 Watershed Segmentation of Input [6] AIJRFANS 15-210; 2015, AIJRFANS All Rights Reserved Page 10
If rain falls on the defined topographical surface, then water would be collected equally in all the catchments basins. The watershed transformation can be built up by flooding process on a gray tone image [6]. However, a major problem with the watershed transformation is the over-segmentation. And this over-segmentation is overcome by applying marker based watershade segmentation [6]. Here, watershade internal marker is the only allowed regional minima. External markers found by finding pixels that are exactly midway between the internal markers and is associated with the background[6]. The gradient image is then modified. The next step involves the computation of the watershed transformation of the Marker modified gradient image to produce watershed ridge lines. Finally resulting watershed ridge lines are superimposed on the original image and produce the final segmentation [6]. D. Cohesion self means algorithm: In this algorithm first K-mean algorithm is applied and after that cohesion self-merging algorithm is applied. It is an unsupervised learning technique. Simply speaking it is an algorithm to classify or to group your objects based on attributes/features into K number of group [8]. In the beginning, we determine number of cluster K and we assume the centroid or center of these clusters [8].Then Calculate distance from all the selected initial centroids to all existing points inside the data set then depending upon the minimum distance criterion the clusters have been formed [7]. Next step is new centroids inside the newly formed clusters calculated. Repeat above steps with respect to newly generated new cluster centroids and algorithm continued until the convergence is reached. Although this technique has showed promising results for a few data sets, it needs to prove its potential in practical applications [9].To identify tumor region CSM algorithm has been applied as the measurement of the inter cluster similarity [7]. Basic formulation is as follows. First calculate mean vector& co-variance matrix Estimator ( ).Consider clusters of n points with locations, the values of ( ) of the clusters calculated by: Where indicates the (pixel vector) Contains location & grey level information of that pixel In the next step we consider all the pixels in the image and here we calculate probability density function for each pixel with respect to the mean and covariance of all existing clusters. Assume the location of a point in each cluster follows a multivariate normal distribution, i. e Where d is the dimension of the space. The probability density function is calculated by using the following formula: Where, Figure 4 An illustration for the meaning of joinability [7] Where, and are the probability (density) function of the distributions in the above mentioned two clusters ( and ) is being calculated by using the formula as shown below AIJRFANS 15-210; 2015, AIJRFANS All Rights Reserved Page 11
Where, and the size of the clusters and respectively [7]. Figure 5 image wiith tumor[7] The above figure5 shows the tumor which is detected by applying cohesion self means algorithm and mathematical morphology. Here binary morphological erosion algorithm is used for the noise removal process. After erosion eroded image removes some parts of tumor. Therefore to get back its original size the opposite algorithm of erosion, dilation has been applied. The dilated output represents the exact size and location of tumor in original image. But, cohesion self means algorithm gives not accurate result when tumor is surrounded by edema. Table 1: MRI brain tumor method comparison Technique Advantage/Disadvantage Seed region growing Advantage Correctly separate the regions that have the same properties we define Determine the seed points and the criteria we want to make It requires manual interaction to obtain seed point Thresholding Advantage Simple method Tumor is diagnosed at advanced stage Also tumor growth can be analysed by plotting graph, which can be obtained by studying sequential images of tumor affected patient. Markerbsed Watreshade Advantage segmentation It removes the over segmentation problem, which occurs in watershade segmentation Cohesion self-merging Advantage It is a simple method It has a less computational complexity. Performance depends highly on initial cluster centers. III. Conclusion For accurate diagnosis of brain tumor proper segmentation method is required for MR images to carry out an improved diagnosis and treatment. In this paper we attempted to review some of the worthwhile recent research works done on brain tumor detection and segmentation. Merits and demerits of different segmentation algorithm is discussed. The discussion showed that few methods are working effectively and accurately in regard of brain image analysis but still there exists need for more effective and precise work. But, these methods give not accurate result when tumor is surrounded by edema. IV. References [1] Praveen Kumar E, Manoj kumar V, Sumithra M G Tumour Detection In Brain Mri Using Improved Segmentation Algorithm IEEE 31661, 4th ICCCNT 20. [2] Ewelina Piekar, Paweł Szwarc, Aleksander Sobotnicki, Michał Momot Application Of Region Growing Method-To Brain Tumor Segmentation Preliminary Results Journal Of Medical Informatics & Technologies Vol. 22/20, Issn 1642-6037. [3] Mukesh Kumar, Kamal Mehta, A Modified Method To Segment Sharp And Unsharp Edged Brain Tumors In 2 D MRI Using Automatic Seeded Region Growing Method, International Journal Of Soft Computing And Engineering (IJSCE) ISSN: 2231-2307, Volume-1, Issue-2, May 2011. [4] Aminah Abdul Malek, Wan Eny Zarina Wan Abdul Rahman, Arsmah Ibrahim, Rozi Mahmud, Siti Salmah Yasiran, Abdul Kadir Jumaat Region and Boundary Segmentation of Microcalcifications using Seed-Based Region Growing and Mathematical Morphology International Conference on Mathematics Education Research 2010 (ICMER 2010). [5] Natarajan P1, Krishnan.N, Natasha Sandeep Kenkre, Shraiya Nancy, Bhuvanesh Pratap Singh, Tumor Detection Using Threshold Operation In MRI Brain Images 978-1-4673-44-5/12/2012 IEEE. AIJRFANS 15-210; 2015, AIJRFANS All Rights Reserved Page 12
[6] Pratik P. Singhai, Siddharth A. Ladhake Brain Tumor Detection Using Marker Based Watershed Segmentation from Digital MR Images International Journal of Innovative Technology and Exploring Engineering (IJITEE) ISSN: 2278-3075, Volume-2, Issue-5, April 20. [7] Subhranil Koley, Aurpan Majumder, Brain MRI Segmentation for Tumor Detection using Cohesion based Self Merging Algorithm 978-1-61284-486-2/2011 IEEE. [8] By Kardi Teknomo,PhD, K-Means Clustering Tutorial July 2007. [9] D T Pham_, S S Dimov, and C D Nguyen, Selection of K in K-means clustering C09304 IMechE 2005 Proc. IMechE Vol. 219 Part C: J. Mechanical Engineering Science. [10] www.webmd.com/cancer/brain-cancer/brain-tumors-in-adult AIJRFANS 15-210; 2015, AIJRFANS All Rights Reserved Page