Segmenting Internal Brain Nuclei in MRI Brain Image using Morphological Operators R.Dhanasekaran2 D.Selvaraj1 1Research Scholar, Department of Electronics & Communication Engg., Sathyabama University Chennai, India 1mails2selvaraj@yahoo.com 2Principal, Syed Ammal Engineering College, Ramanathapuram, India 2rdhanasekar@yahoo.com Abstract— We present a new technique for segmenting brain nuclei from MRI brain images. Our method performs the segmentation using a combination approach of thresholding with morphological operators. The MRI brain image contains skull and noisy background. The latter have to be removed for further analysis. Elimination of any obstacles and noise from the image is the primary function of the morphological operators. We use simple morphological operators like dilation, erosion, opening and closing to the binarized MRI brain image. The results of skull stripped MR image with the use of disk shaped structuring elements are presented in the paper. The proposed method has been applied to a large number of MR images showing promising results for various image qualities, encouraging for future. Keywords— Image segmentation, Image processing, skull stripping, Morphological operator, brain segmentation I. INTRODUCTION Magnetic resonance imaging (MRI) of the human brain is the most common type of medical imaging used in the medical diagnosis among a variety of imaging modadilities such as computer tomography (CT), positron emission tomography, ultrasound, mammography and radiography. So, MR images are widely used not only for detecting tissue deformities such as cancer and injuries but also for studying brain pathology [8]. Also, many neurological diseases and conditions alter the normal volume and regional distribution of brain parenchyma (Gray and white matter), cerebrospinal fluid. Such abnormalities are commonly related to the conditions of hydrocephalus, cystic formation, brain atrophy and tumour growth. for the brain parenchyma, CSF volume and shape is segmentation. Image segmentation is to divide the image into disjoint homogenous regions, where all the pixels in the same class must have some common characteristics but the major problems that affect the quality of MRI segmentation are noise, inhomogeneous pixel intensity distribution and blunt boundaries in the medical MR images caused by MR data acquisition process [2, 3, and 4]. These problems do make manual quantitative analysis of brain imaging data a tedious and time consuming procedure, prone to observer variability [2]. Due to the characteristics of brain MRI, development of automated segmentation algorithms require pre-processing which includes denoising, stripping of skull. reliable measurement of The basis This paper presents a method for skull segmentation using a sequence of mathematical morphological operations: erosion and dilation, and their compositions i.e., opening and closing. The operators of morphological processing are particularly useful for the analysis of binary images so that MRI images need to be previously binarized. The background and brain mask of the image are obtained by applying a combination approach of thresholding with morphology. The next section presents some basics on morphological operations. Section 3 describes our methodology. Finally we show some results in section 4 and draw some conclusions and future work perspectives in section 5. II. MATHEMATICAL MORPHOLOGY CONCEPTS that extracts image objects Mathematical morphology is a non-linear image analysis technique information by describing its geometrical structure in a formal way [7]. Mathematical morphology has been largely used in several practical image processing and analysis problems. Here we briefly review some mathematical morphology operators and the corresponding operations used in this work. Mathematical operators take two data as an input: an image to be processed and a structuring element, which is a matrix used for defining a neighbourhood shape and size [1]. By choosing the shape and size of the element, we can influence the morphological operations sensitivity to specific shapes appearing in the processed image. The elementary shapes of symmetrical structuring elements used in the following processing are shown in Fig. 1. The erosion of binary image I by structuring element S is The dilation of binary image I by structuring element S is defined by the formula [1]: (2) I ⊕ S = {x,y : Sxy ∩ I≠∅} Let f: D⊂ Rn → R is an image function and g: G⊂ Rn → R is a structuring function. The two fundamental operations of gray-scale morphology, erosion and dilation, are defined as: Definition 1 (Dilation) The dilation of the function f(x) by the structuring function g(x), (f⊕ g)(x), is given by: (f ⊕ g)(x) = max {f(z) + (gx)(z) : z ∈ D[gx]} (3) Definition 2 (Erosion) The erosion of the function f(x) by the structuring function g(x), (f Θ g) (x), is given by: (f Θ g)(x) = min {f (z) − (gx) (z): z ∈ D[gx]} Where gx indicates the translation by x (gx) (z) = g(z − x), (4) and D[gx] denotes the domain of the translated structuring function. defined by the formula [1]: I ⊗ S = {x,y : Sxy ⊆ I} (1) 978-1-4244-5392-4/10/$26.00 ©2010 IEEE 

The operations of closing and opening are the combinations of erosion and dilation, both using the same structuring element. Morphological opening is erosion followed by dilation and morphological closing is dilation followed by erosion. The Fig.3 shows that in a binarized image there are some remaining pixels that represent the noise. To remove the left-over pixels the opening operation was used. (a) (b) Figure 1. Disk shape structuring elements: (a) 2-pixel radius, (b) 5-pixel radius III. PROPOSED METHODOLOGY FOR STRIPPING SKULL TO SEGMENT BRAIN This section presents the proposed methodology for segmenting brain MRI images. The fundamental task in brain MRI segmentation is the classification of volumetric data into grey matter, white matter and cerebrospinal fluid but it is not easy as there are some inherent difficulties associated with image segmentation; among them are RF coil in homogeneity, brain tissue susceptibility and other systematic artifacts. Various preprocessing steps have been proposed to deal with some or all of these difficulties. Skull stripping is the first processing step in the segmentation of brain tissueas shown in Fig 2. Figure 2. Overview of proposed methodology In the proposed method for skull stripping, we see the brain surface as a smooth manifold with relatively low curvature that separates brain from non-brain regions. Also, the brain cortex can be visualized as a distinct dark ring surrounding the brain tissues in the T1-weighted axial MR images. The steps involved in the proposed methodology for skull stripping and brain extraction consists of three steps. 1. Binarization of every image. 2. Opening operation and closing operation on every image in the sequence using the structuring element. 3. Applying the binary mask to the received MRI input image. A. Binarization Binarization is the process that converts a grey level image into a binary image. The binarization process involves examining the grey-level value of each pixel in the enhanced image, and if the value is greater than the global threshold, then the pixel value is set to a binary value one; otherwise it is set to zero. The binarized image is shown in Fig. 3. (a) (b) Figure 3. (a) Input Image, (b) Binarized Image B. Morphological Operation The binary morphological operators are then applied on the binarized image. Elimination of any obstacles and noise from the image is the primary function of the morphological operators. The morphological operators namely, opening and closing are being employed in the proposed method. 1) Opening: An opening operation consists of erosion followed by dilation with the same structuring element. The Fig. 4 shows the image after applying the opening operator. Figure 4. Binarized image after applying opening operator 2) Closing: A closing operation consists of a dilation followed by an erosion with the same structuring element. The Fig. 5 shows the image after applying the closing operator. 3) Erosion: Erosion operation on an image I containing labels 0 and 1, with a structuring element S, changes the value of pixel i in I from 1 to 0, if the result of convolving S with I, centered at i, is less than some predetermined value. We have 

set this value to be the area of S, which is basically the number of pixels that are 1 in the structuring element itself. The structuring element (also known as the erosion kernel) determines thins boundaries. the details of how particular erosion 4) Dilation: Dual to erosion, a dilation operation on an image I containing labels 0 and 1, with a structuring element S, changes the value of pixel i in I from 0 to 1, if the result of convolving S with I , centered at i , is more than some predetermined value. We have set this value to be zero. The structuring element (also known as the dilation kernel) determines the details of how a particular dilation grows boundaries in an image Figure. 5. Brain Mask C. Region-based binary mask extraction Step 8: Develop binary mask choosing the larger area Step 9: Apply binary mask on the original image to get skull stripped image (brain nuclei). IV. EXPERIMENTAL RESULTS The experimental results of the proposed methodology for segmenting brain MRI images are presented in this section. The proposed methodology is implemented in Matlab (7.4). The input to the proposed methodology is T1-weighted brain MRI images collected from publicly available databases. Regarding T1-weighting, every tissue in the human body has its own T1 and T2 value. This term is used to indicate an image where most of the contrast between tissues is due to differences in the T1 value. The proposed methodology is based on Intensity Thresholding (IT), which is the easiest and fastest segmentation method, often adopted for preprocessing of medical images and preregistration problems. The sample results of brain MRI segmentation obtained using the proposed methodology is shown in the following Fig.7 to Fig. 11. Region-based extraction is done by examining the properties of each block that satisfy some criteria. We have used one of two criteria. One criterion is to look at the max- min difference and the other is to determine the mean values of the blocks. The process results with a brain mask that is then applied to the original MRI data. Consequently, we attain a brain MRI image with its brain cortex stripped as shown in Fig. 6. (a) (b) Figure 7. (a) Input Image, (b) Segmented Brain Image Figure 6. Skull Stripped Brain Image (a) (b) Figure 8. (a) Input Image, (b) Segmented Brain Image D. Algorithm Step 1: Get the MRI brain image to be stripped Step 2: Find the maximum pixel value in the image Step 3: Find the limit of high and Low frequency value Step 4: Normalize the image Step 5: Calculate the gray threshold from normalized image Step 6: Construct binary image using gray threshold value Step 7: Apply disk shaped morphological operator to eliminate the obstacle and noise from the image (a) (b) Figure 9. (a) Input Image, (b) Segmented Brain Image 

(a) (b) Figure 10. (a) Input Image, (b) Segmented Brain Image (a) Figure 11. (a) Input Image, (b) Segmented Brain Image (b) V. CONCLUSION In this paper, an automated, simple and efficient brain MRI segmentation method for classifying brain tissues has been presented. Initially, the cortex present in the brain MRI images is extracted by combining preprocessing techniques and incorporating mathematical morphology. Experimental results have showed that the proposed method does a reasonably good job in terms of segmentation. In future from this segmented brain image we can segment grey matter, White matter and cerebrospinal fluid. REFERENCES [2] [1] Renata Kalicka, Anna Pietrenko-Dabrowska, “Efficiency of new method of removing the noisy background from the sequence of MRI scans depending on structuring elements used to morphological operations, Proceedings of the 2008 1st international conference on information Technology, IT 2008, May 2008 Zijdenbos, A., Forghani, R., Evans, A.: “Automatic pipeline analysis of 3D MRI data for clinical trials: application to multiple sclerosis”, IEEE transactions on medical imaging, Vol: 21, No: 10, pp: 1280–1291, 2002. [3] Van-Leemput, K.: “Probabilistic brain atlas encoding using bayesian inference”, Book Medical Image Computing and Computer-Assisted Intervention (MICCAI), vol:1, pp:704–711, 2006. [4] D. L. Pham, C. Xu, J. L. Prince, " A survey of current methods in medical image segmentation", In Annual Review of Biomedical Engineering, Vol. 2 , pp. 315-338, 2000. Legal-Ayala, H.A. Facon, J., "Automatic segmentation of brain MRI through learning by example", proceedings of International Conference on Image Processing, (ICIP '04.), Vol.2, pp. 917- 920, 24-27 Oct. 2004. Pham, D., Xu, C., Prince, J., ”Current methods in medical image segmentation”, Annual Review of Biomedical Engineering Vol:2, pp: 315–337, 2000. Leyza aldo Dorini, Rodrigo Minetto, “White Blood cell Segmentation using morphological operators and scale-space analysis”A. Karnik, “Performance of TCP congestion control with rate feedback: TCP/ABR and rate adaptive TCP/IP,” IEEE computer Society, 294- 304, 2007. [8] Atkins M.S., Mackiewich B.T., Fully automatic segmentation of the brain in MRI, IEEE Transactions on Medical Imaging 17(1), 1998, pp. 98-107. [7] [5] [6]