IJRIT International Journal of Research in Information Technology, Volume 3, Issue 4, April 2015, Pg. 482-488

International Journal of Research in Information Technology (IJRIT) www.ijrit.com

ISSN 2001-5569

A Study Of Various Techniques For The Brain Tumor Segmentation And Detection From Mr Images : A Review Pooja Thakur Mtech (E.C.E.) M.M.E.C , Mullana India [email protected]

Dr. Kuldip Pahwa Proff. ( E.C.E.) M.M.E.C., Mullana India [email protected]

Dr. Rajat Gupta Proff. (E.C.E.) M.M.E.C., Mullana India [email protected]

ABSTRACT Brain tumor is an abnormal mass of tissue in which some cells grow and multiply uncontrollably, apparently unregulated by the mechanisms that control normal cells. The growth of a tumor takes up space within the skull and interferes with normal brain activity. So detection of the tumor is very important in earlier stages. The most common imaging technique for brain is MR imaging it is a non-invasive method. Brain tumors are mainly classified as benign or malignant tumors depending on their growth pattern. The manual analysis of brain tumor on MRI is time consuming and subjective Intensity inhomogeneity is very challenging task image segmentation to avoid thus type of problem, in this paper describe the very efficient and accurate segmentation techniques. This paper presents a comprehensive review of the methods and techniques used to detect brain tumor through MRI image segmentation. KEYWORDS: Brain tumor , MRI, Segmentation, Morphological operation. I. INTRODUCTION The brain is a delicate, sensitive, non-replaceable and spongy mass of tissue. A tumor is fundamentally a mass of tissue that develops crazy of the ordinary powers that manages its development. Brain tumor is a gathering of anomalous cells that develops either inside the cerebrum then again around the cerebrum. The term brain tumor is used to describe any tumor growing within the skull, though a more accurate term might be intracranial tumor. Only some of these growths arise directly from brain tissue. Others grow from the other tissues inside the skull, such as pituitary tumors.

In contrast to these primary brain tumors, which arise within the skull, another group consists of tumors that spread to the head from another source, such as lung or breast cancer; these are secondary brain tumors, and they are much more common. There are a great many different types of brain tumors, each with its own specific biology and treatment, but all cause similar symptoms. Both primary and secondary tumors exist on a spectrum, from high to low grade. In most high-grade tumors, also called malignant, or cancerous, the cells are very different from normal cells, grow relatively quickly, and can spread (metastasize) easily to other locations [1]. A.Types of TUMOR Tumor : The word tumor is a synonym for a word neoplasm which is formed by an abnormal growth of cells Tumor is something totally different from cancer. There are three common types of tumor: 1) Benign

Pooja Thakur, IJRIT-482

IJRIT International Journal of Research in Information Technology, Volume 3, Issue 4, April 2015, Pg. 482-488

2) Pre-Malignant 3) Malignant Benign Tumor: A benign tumor is a tumor is the one that does not expand in an abrupt way; it doesn’t affect its neighboring healthy tissues and also does not expand to non-adjacent tissues. Moles are the common example of benign tumors. Pre-Malignant Tumor: Premalignant Tumor is a precancerous stage, considered as a disease, if not properly treated it may lead to cancer. Malignant Tumor: Malignancy (mal- = "bad" and -ignis = "fire") is the type of tumor, that grows worse with the passage of time and ultimately results in the death of a person. Malignant is basically a medical term that describes a severe progressing.There are different sorts of malignant tumors, for example, astrocytoma, meningioma, glioma, medulloblastoma and metastatic, which fluctuate enormously in appearance — shape, size and location. Magnetic resonance (MR) sequences for example, T1-weighted, T2-weighted and differentiation upgraded T1-weighted sweeps give distinctive data about tumor. On these images, brain tumors show up either hypointense (darker than brain tissue), or isointense (same force as brain tissue), or hyperintense (brighter than brain tissue) [2]. ASTROCYTOMA

MENINGIOMA

TUMOR TYPE

GLIOMA

MEDULLOBLAS TOMA

METASTATIC Figure 1 Types of Tumour

In general, MRI is superior to CT, but in many instances a CT scan can accurately reveal the underlying problem. For the most part, imaging is the only test necessary to diagnose a brain tumor.. . MRI doesn't influence the human body as it doesn't utilize any radiation. It is taking into account the attractive field and radio waves. Diverse sorts of algorithm were developed for brain tumor detection. They may have some drawback in detection and extraction B. CAUSES Apart from exposure to vinyl chloride or ionizing radiation, there are no known environmental factors associated with brain tumors. Mutations and deletions of tumor suppressor genes known are considered responsible for some forms of brain tumors. Patients with various inherited diseases, such as von Hippel-Lindau syndrome, multiple endocrine neoplasia, neurofibromatosis type 2 are at high risk of developing brain tumors. II. MAGNETIC RESONANCE IMAGING Imaging plays a central role in the diagnosis of brain tumors. Imaging early invasive and sometimes dangerous, as pneumonic paleography and cerebral angiography have been abandoned in recent times in favour of noninvasive high-resolution modalities, such as computed tomography (CT) and especially magnetic resonance imaging (MRI). Magnetic resonances imaging (MRI), nuclear magnetic resonance imaging (NMRI), or magnetic resonance tomography (MRT) are medical imaging methods utilized as a part of radiology to image inside structures of the body in point of interest. MRI makes utilization of the property of nuclear magnetic resonance (NMR) to image of nuclei of atoms inside the body.

Pooja Thakur, IJRIT-483

IJRIT International Journal of Research in Information Technology, Volume 3, Issue 4, April 2015, Pg. 482-488

Types, Characteristics and Use of Weighted Image Image type

Contrast

Tissue appearance

Best for

T1Weight ed

Mainly Due to Difference s in T1 recovery Times

1.High-fat-content tissues appear as bright areas of high signal intensity (hyperintense). 2.High-watercontent tissues appear as darkareas of low signal intensity (hypointense).

Anato my and, if used with contrast enhanc ement, also patholo gy

T2Weight ed Pathol ogy

Mainly due to Difference s in T2 decay times

1. High-fatcontent tissues appear as dark areas of low signal intensity (hypointense). 2. High-watercontent tissues appear as bright areas of high signal intensity (hyperintense).

Patholo gy

Proton density Weight ed

Anatomy, and Mainly due to Difference s in proton density.

1. Low-protondensity tissues appear as dark areas of low signal intensity (hypointense). 2. High-protondensity tissues appear as bright areas of high signal intensity (hyperintense).

patholo gy

Table 1 MRI Types and Characteristics A MRI scanner is a gadget in which the patient falsehoods inside a vast, intense magnet where the attractive field is utilized to adjust the polarization of some nuclear cores in the body, and radio frequency fields to methodicallly modify the arrangement of this magnetization. MRI is of basically 2 sorts: • •

T1-weighted MRI Spin-lattice relaxation time T2-weighted MRI Spin-spin relaxation time

Another kind of MRI is: •

T*2-weighted MRI (Contrast Enhance)

Pooja Thakur, IJRIT-484

IJRIT International Journal of Research in Information Technology, Volume 3, Issue 4, April 2015, Pg. 482-488

The qualities of the aforementioned sorts of MRI are given in table 1. In clinical practice, MRI is utilized to recognize pathologic tissue, (for example, a brain tumor) from typical tissue. One focal point of a MRI scan is that it is safe to the patient. It utilizes solid magnetic fields and nonionizing electromagnetic fields in the radio frequency range, dissimilar to CT scan and conventional X-ray, which both utilization ionizing radiation. While CT gives great spatial resolution (the capacity to recognize two different structures a subjectively little separation from one another), MRI gives equivalent resolution with obviously better contrast resolution (the capacity to recognize the contrasts between two subjectively comparative however not indistinguishable tissues). III. SEGMENTATION METHODS A. Seed-based region growing The principal region growing technique was the seeded region growing strategy. This technique takes an arrangement of seeds as information alongside the image. The seeds check each of the articles to be sectioned. The locales are iteratively developed by contrasting all unallocated neighbouring pixels to the areas. The contrast between a pixel's intensity value and the region's mean, δ, is utilized as a measure of likeness. The pixel with the littlest contrast measured thusly is apportioned to the particular region. Seed based region growing (SBRG) performs a segmentation of a image regarding a point, known as seed. Beginning with a seed point, the locale will develop by attaching to every seed whose neighbouring pixels have properties like the seed. In the district developing division, the first point is to focus the introductory seed focuses. A seed point is the beginning stage for district developing and its determination is critical for the division result. The technique for numerical morphology is utilized keeping in mind the end goal to get a beginning seed point consequently. Morphology includes a theory for the examination of shape and spatial structures. Morphological operations like expansion, disintegration, opening, shutting and local maxima are utilized for separating, altering and controlling the highlights show in the image in light of their structuring component (SE) [4-9]. B. Level-set segmentation Utilizing a Partial differential equation (PDE)-based system and comprehending the PDE equation by a numerical plan, one can segment the image. Curve propagation is a mainstream strategy in this class, with various applications to object extraction, object tracking, and stereo remaking. There are for the most part three routines under the PDE Level Set method, Parametric Method, Fast Marching system. Level Set Method one of the developing image division systems for medical image segmentation. The level set technique is a numerical system for following interfaces and shapes. The fundamental thought of the level set technique is to speak to forms as the zero level arrangement of a certain capacity characterized in a higher measurement, typically alluded to as the level set capacity, and to develop the level set capacity as indicated by an partial differential equation (PDE). In common PDE systems, images are thought to be continuous functions sampled on a grid. Active contours were introduced in order to segment objects in images using dynamic curves. Geometric active contours form models are ordinarily determined utilizing the Euler-Lagrange mathematical equation. In level set definition of moving fronts (oractive shapes), the fronts, indicated by, are spoken to by the zero level arrangement of a level set function[1011]. C. Graph-based segmentation The primary thought behind graph based is: • • •

Convert image into a graph Vertices for the pixels Edges between the pixels

Extra vertices and edges to encode other requirements •

Manipulate the graph to portion the image.

Graph based system predominantly comprised of two stages, i.e. the graph development for mapping a image to a graph, and the merging of vertices in the graph. The graph based segmentation method went about as a

Pooja Thakur, IJRIT-485

IJRIT International Journal of Research in Information Technology, Volume 3, Issue 4, April 2015, Pg. 482-488

bunching strategy and extended (or combined) locales as per the nearby spatial, notwithstanding the worldwide data. In this way, the areas with comparative power levels however distinctive areas could be decently partitioned into diverse sections D. Split and Merge-based segmentation Completely merging methods are, generally; computationally extravagant in light of the fact that the beginning purpose of such technique is small regions (individual focuses). This technique can be more productive by recursively part the image into littler and littler region until all individual areas are reasonable, then recursively consolidating these to create bigger reasonable regions. First, we must part the image. Begin by considering the whole image as one region. 1. If the entire region is coherent (i.e., if all pixels in the locale have sufficient likeness), abandon it unmodified. 2. If the entire region is not sufficiently coherent, part it into four quadrants and recursively apply these ventures to every new region. Split and Merge-based segmentation is in view of a quadtree segment of a image. It is now and again called quadtree division. This technique begins at the base of the tree that speaks to the entire image. On the off chance that it is discovered nonuniform (not homogeneous), then it is part into four sonsquares (the part process), thus on so forward. Alternately, if four child squares are homogeneous, they can be converged as a few joined parts (the consolidating methodology) [1]. R1

R2

R3

R41

R42

R43

R44

Figure 3 Partitioned image

R

R1

R2

R41

R4

R3

R42

R43

R44

Figure 4 Quadtree Image

E. Edge Based Segmentation Region boundaries and edges are nearly related, subsequent to there is regularly a sharp change in intensity at the locale limits. Edge detection procedure have along these lines been utilized as the base of another division method. The edges recognized by edge detection are regularly detached. To segment an article from a image then again, one needs closed region boundaries. The wanted edges are the limits between such object. Division strategies can likewise be connected to edges gotten from edge detectors. Since a (binary) object is completely represented by its edges, the division of an image into separate items can be attained to by discovering the edges of those items. A common way to deal with division utilizing edges is (1) process an edge image, containing all edges of an unique image, (2) process the edge image so that just closed object limits remain, and (3) change the result to a customary divided image by filling in the object limits.

Pooja Thakur, IJRIT-486

IJRIT International Journal of Research in Information Technology, Volume 3, Issue 4, April 2015, Pg. 482-488

The edge based division comprises of these steps: 1.

Register an edgeness image ∇f from f . Any favored angle administrator can be used for this.

2.

Edge ∇f to a image (∇f )t, so we have a paired image indicating edge pixels.

Figure a Laplacian image ∇f from f . Any favoured discrete or continuous Laplacian administrator may be utilized. 4. Figure the image g=(∇f ) ‫ݐ‬.sgn(∇f ). The sgn operator gives back the indication of its contention. The result image g will therefore contain just three values: 0 at non-edge pixels of f, 1 at edge pixels on the splendid side of an edge, and −1 at edge pixels on the dim side of an edge. The image g contains the limits of the articles to be sectioned. The Laplacian is used to encourage the last ventures of the calculation: transforming the limit image into asegmented image h containing strong items. In the event that we navigate the image g from left to right,two neighboring pixels with qualities −1 and 1 methods we move into an item, and the values1 and −1 implies we move out of one. The image h can along these lines be made by setting all pixel qualities to zero, aside from those pixels that are between the moves 1 → −1and −1 → 1 in every line of g, which are situated to 1. If remarkable qualities are wanted for each separate portion, a labelling algorithm can be run on h. There are several methods for edge based segmentation such as: Sobel, Prewitt, Roberts, Canny etc. 3.

F. Morphological Operations Morphological image processing is gathering of non linear operations identified with the shape or morphology (shape, structure) of features in a image. As per Wikipedia, morphological operations infer just on the relative requesting of pixel values , not on their numerical values. Morphological operations are more suited to the handling of parallel images. Morphological operations can likewise be connected to greyscale images such that their light exchange capacities are obscure and consequently .Their absolute pixel qualities are of no or minor hobby. Morphological methods test a image with a little shape or layout called an organizing component. The organizing component is situated at all conceivable areas in the image and it is contrasted and the comparing neighbourhood of pixels. A few operations test whether the component "fits" within the neighbourhood, while others test whether it "hits" or intersects the area: A morphological operation on a parallel image creates a new binary image in which the pixel has a non-zero worth just if the test is fruitful at that area in the data image [1214]. IV. CONCLUSION Image division is the most common and most recent research territory in the field of image preparing for the last decade. Disregarding the accessibility of a huge mixed bag of condition of workmanship strategies for cerebrum MRI division, it is still an intense assignment and there is a need and wide degree for future examination to enhance the exactness and precision of division techniques. Presenting new strategies and joining distinctive techniques can be the future outline for making change in cerebrum division strategies. In view of the today's exploration in natural world, In this paper, we display a similar study (audit) of diverse methodologies utilized for therapeutic image division. The technique exhibited in this paper is utilized with new methodologies of image division for the better exactness and accuracy of results. There are a few methodologies utilized for Brain tumor division from MR images, for example, Fuzzy C-mean and K-mean bunching [3] [15-17], Neural system based division [6] [18]. REFERENCES [1] S. K. Bandhyopadhya and T. U. Paul, ―Segmentation of Brain MRI Image – A ReviewǁInternational Journal of Advanced Research in Computer Science and Software Engineering, Vol. 2, No. 3, issue 3, 2012. [2] Romesh Laishram, W.Kanan Kumar Singh, N.Ajit Kumar, Robindro.K, S.Jimriff― A Novel MRI Brain Edge Detection Using PSOFCM Segmentation and Canny Algorithm vol.1, pp. 398 – 401, 2014

Pooja Thakur, IJRIT-487

IJRIT International Journal of Research in Information Technology, Volume 3, Issue 4, April 2015, Pg. 482-488

[3] J.selvakumar, A.Lakshmi and T.Arivoli ―Brain Tumour Segmentation and Its Area Calculation in Brain MR Images using K-Mean Clustering and Fuzzy C-Mean Algorithmǁ.IEEEInternational Conference on Advances in Engineering, Science and Management, vol. 31, pp. 978-81-909042-2-3, 2012. [4] A. K. Jumaat, R. Mahmud, and S. S. Yasiran, ―Region and boundary segmentation of microcalcifications using seed-based region growing and mathematical morphologyǁ, International Conference on Mathematics Education Research, Vol. 8, pp. 634– 639, 2010. [5] Q-H. Huang, Su-Y. Lee, L-Z. Liu, M-H. Lu, L-W. Jin and A-H. Li ―A robust graph-based segmentation method for breast tumours in ultrasound imagesǁ, Ultrasonics, vol. 52, pp. 266–275, 2011. [6] M. Jafari and S. Kasaei ―Automatic brain tissue detection in Mri images using seeded region growing segmentation and neural network classificationǁ,Australian Journal of Basic and Applied Sciences, vol. 5, No. 8, pp.1066-1079, 2011. [7] N. M. Saad, S.A.R. A.-Bakar, S. Muda, M. Mokji and A.R. Abdullah ―Fully automated region growing segmentation of brain lesion in diffusion-weighted MRIǁ,IAENG International Journal of Computer Science, vol. 39, No. 2, 2012 [8] S. TIWARI, A. BANSAL and R. SAGAR ―Identification of brain tumours in 2 D MRI using automatic seeded Region Growing Methodǁ,International Journal of Electronics Communication and Computer Engineering Vol. 2, No. 1, pp. 41-43, 2012. [9] T. Węgliński and A. Fabijańska ―Brain tumours segmentation from MRI data sets using region growing approachǁ, MEMSTECH, 2011. [10] P. C. Barman, S Miah, B. C. Singh and T. Khatun―MRI Image segmentation using level set method and implement an medical diagnosis systemǁ,An International Journal (CSEIJ), Vol.1, No.5, 2011. [11] S. Taheri , S.H. Ong and V.F.H. Chong ―Level-set segmentation of brain tumours using a threshold-based speed functionǁ, Image and Vision Computing,vol. 28, pp. 26–37, 2009. [12] K. Thapaliya andG.-R. K won ―Extraction of Brain Tumour Based on Morphological Operationsǁ.,Vol. 1,pp. 515 – 520,2012. [13] M. U. Akram and A. Usman ―Computer aided system for brain tumours detection and segmentationǁ, IEEE transaction, 978-1- 61284-941-6/11, 2011. [14] R. C. Patil and A. S. Bhalchandra ―Brain Tumour Extraction from MRI Images Using MATLABǁ, International Journal of Electronics, Communication & Soft Computing Science and Engineering Vol. 2, pp. 2277-9477,2012. [15] A.Rajendran and R. Dhanasekaran―Fuzzy Clustering and Deformable Model for Tumour Segmentation on MRI Brain Image: A Combined Approachǁ,Procedia Engineering,Vol. 30, pp. 327 – 333, 2012. [16] L-H Juang and M-N Wu―MRI brain lesion image detection based on color-converted K-means clustering segmentationǁ.Measurement, Vol. 43, pp. 941–949, 2010. [17] S. J. Hussain, T. S. Savithri and P.V. S. Devi ―Segmentation of Tissues in Brain MRI Images using Dynamic Neuro-Fuzzy Techniqueǁ, International Journal of Soft Computing and Engineering (IJSCE), Vol. 1, No. 6, pp. 2231-2307, 2012. [18 ] A. Kothari ―Detection and classification of brain cancer using artificial neural network in MRI imagesǁ, World Journal of Science and Technology, Vol. 2, No. 5, pp. 2231 – 2587, 2012.

Pooja Thakur, IJRIT-488