Comparative Study of Different Methods for Brain Tumor Extraction from MRI Images using Image Processing

Background/Objectives: The objective of this paper is to study various segmentation methods implemented using MATLAB and to compare accuracy of each. Statistical Analysis/Findings: Preprocessing is required for better segmentation, as it removes noise and makes images having equal attribute so that accuracy to segment can be increased. Segmentation using Thresholding, region based segmentation and watershed segmentation, all the methods are performed and Comparison of accuracy of all the methods has been calculated on basis of actual tumor part and segmented tumor part. Morphological operations are used in all the methods in order to avoid noise part of segmented image and to have higher accuracy. Accuracy of the three methods which are region based, thresholding and watershed are 87.48, 91.34 and 92.76 respectively. Here we have used all T2-weighted Magnetic Resonance Imaging (MRI) images as it is noninvasive technique and having high contrast between tumor and normal part. Application/Improvement: Segmented tumor with higher efficiency leads to help doctor in anatomy and pathology to classify tumor type so that treatment could be started accordingly as soon as possible.