Brain image segmentation and analysis of different parts of brain is very important research issue. In this paper we proposed a method for segmenting brain tumors by combining Fuzzy C-Means thresholding with watershed segmentation method. We also present a comparative study of different existing segmentation approaches like global thresholding using Otsu method, histogram thresholding and watershed method with addition of some pre- and post-processing methods. Among all the methods Fuzzy C-Means clustering with watershed method gives the best result. These methods allow the segmentation of tumor tissue with accuracy and efficiency as compared to manual segmentation. The tumor is extracted from the brain image and its exact position is also determined. This process calculates the area of tumor for each of the algorithm used. A performance wise comparative study using four number of defected 2D MRI brain images are considered. Several cases under hazy and bad imaging conditions are also studied and good result is obtained.
Edge Detection Method, Fuzzy C-Means Clustering, Thresholding Method, Tumor Extraction, Watershed Method.