International Journal of Computer Applications (0975 – 8887) Volume 125 – No.6, September 2015 5 A Comparative Analysis of MRI Brain Tumor Segmentation Technique Anubha Lakra ECE, Hindu College of Engg. Sonepat, India R.B. Dubey ECE, Hindu College of Engg. Sonepat, India ABSTRACT Magnetic Resonance Imaging (MRI) is a powerful visualization tool that permits to acquire images of internal anatomy of human body in a secure and non-invasive manner. The important task in the diagnosis of brain tumor is to determine the exact location, orientation and area of the abnormal tissues. This paper presents a performance analysis of image segmentation techniques, viz., Genetic algorithm, K- Means Clustering and Fuzzy C-Means clustering for detection of brain tumor from brain MRI images. The performance evaluation of these techniques is carried out on the real time database on the basis of error percentage compared to ground truth. General Terms Segmentation algorithms, Brain tumor Keywords MRI brain tumor, segmentation, Genetic algorithm, K-means clustering and Fuzzy C-means clustering. 1. INTRODUCTION Modern medical imaging modalities generate larger and larger images which simply cannot be examined manually. This drives the development of more efficient and robust image analysis methods, tailored to the problems encountered in medical images. However, the generality of the problem can lead to potential impacts also in other areas of image analysis. Image segmentation is defined as a technique which partitions an image into different regions having high degree of similarity with objects of significance in the image. Depending on different properties of an image, the techniques for image segmentation can be categorized into discontinuity based segmentation and similarity based segmentation [1]. Human brain is the most complex organ present in the human body [2]. Segmentation subdivides an image into its constituent regions or objects [3]. The level of detail to which the subdivision is carried depends on the problem being solved. Means segmentation should stop when the objects or regions of interest in an application have been detected. Computer technology is having a tremendous impact on medical imaging. Computer-Aided Diagnosis (CAD) is an interdisciplinary technology combining digital image processing and medical images obtained from X-ray, MRI and ultrasound, to extract information that would normally require the intervention and analytical reasoning of radiologist. The interpretation of medical images, however, is still almost exclusively the work of radiologist but this is expected to change. Computers will be used more often for image interpretation. This kind of research area is called Computer- Aided Diagnosis (CAD). The role of CAD is very vital in diagnosis. The use of CAD not only assists the pathologists to reduce rate of error but in addition it helps in distribution and sharing of the digitally processed images [2]. Image segmentation plays a significant role in image processing as it helps in the extraction of suspicious regions from the medical images. The combination of fuzzy C-means algorithm with watershed algorithm was used to minimize the error in the process of image segmentation and to improve edge detection in brain tumor MR images [4]. A K-means clustering algorithm followed by morphological filtering was used for segmentation of brain MR image which avoids the misclustered regions that can inevitably be formed after segmentation of the brain MRI image for detection of tumor location [5]. A latest survey of different technologies used in medical image segmentation using fuzzy C -means (FCM) is presented in [6]. An optimized fuzzy logic method for MRI brain images segmentation is presented in [7] which is based on a modified FCM clustering algorithm. The FCM algorithm that incorporates spatial information into the membership function is used for clustering, while a conventional FCM algorithm does not fully utilize the spatial information in the image. The advantages of the algorithm are that it is less sensitive to noise than other techniques, and it yields regions more homogeneous than those of other methods [7]. A novel algorithm for fuzzy segmentation of MRI data is realized by modifying the objective function in the conventional FCM algorithm using a kernel-induced distance metric and a spatial penalty on the membership functions [8]. Experimental results on both synthetic and real MR images show that the proposed algorithms have better performance when noise and other artifacts are present than the standard algorithms. An alternative fuzzy C-mean (AFCM) was used for MRI segmentation in ophthalmology. These unsupervised segmentation algorithms can help ophthalmologists to reduce the medical imaging noise effects originating from low resolution sensors and the structures that move during the data acquisition [9]. An approach that combine region based fuzzy clustering called Enhanced Possibilistic Fuzzy C-means (EPFCM) and Gradient vector flow (GVF) snake model for segmenting tumor region on MR images was introduced in [10]. Region based fuzzy clustering is used for initial segmentation of tumor then result of this is used to provide initial contour for GVF snake model, which then determines the final contour for exact tumor boundary for final segmentation. An adaptive spatial fuzzy C-means clustering algorithm for the segmentation of three-dimensional (3-D) MR images is presented in [11]. The efficiency of the this algorithm is demonstrated by extensive segmentation experiments using both simulated and real MR images and by comparison with other published algorithms. Critical appraisals of the current status of semi-automated and automated methods for the segmentation of anatomical medical images are reviewed in [12]. A Grey level co- occurrence matrix (GLCM) for texture feature extraction, ANFIS (adaptive network fuzzy inference system) plus Genetic algorithm for feature selection and FCM (fuzzy C- means) for segmentation of astrocytoma with all four Grades