Comparative Analysis of Segmentation of Tumor from Brain MRI Images Using Fuzzy C-Means and K-Means

Background/Objectives: Main aim of this study is to test the advantages and failure of each under varying conditions and finally discover which algorithm is excellent in segmentation of tumor. Statistical Analysis/Findings: In this research work, FCM (Fuzzy C-Means) which is representative object based method and centroid based K-Means, the two important clustering algorithms clustering algorithms are compared, both the methods are clustering based methods. Comparison of both methods has been done in respect of computational cost and accuracy of segmented tumor. Here, testing of both the techniques over 35 images, and got accuracies as: 79.15% by FCM and 94.72% by k-means technique and calculated time elapsed by each technique which is same for all the images and that is: 0.019 second for k-mean and 0.027 second for FCM. It is the challenging task to accurately segment tumor region because of its unpredictable shape and appearance. Application/Improvement: The segmented image contains less but effective information, so ultimately for analysis one may need less memory space and time to process on image. Hence segmented image having high accuracy is considered to direct classification task instead of original brain MRI (Magnetic Resonance Imaging).