Semi-Supervised Approach Based Brain Tumor Detection with Noise Removal

Abstract

Brain tumor detection and segmentation is the most important challenging and time consuming task in the medical field. In this paper, Magnetic Resonance Imaging (MRI) sample image is considered and it is very useful to detect the Tumor growth. It is mainly used by the radiologist for visualization process of an internal structure of the human body without any surgery. Generally, the Tumor is classified into two types such as malignant and benign. There are many variations in tumor tissue characteristics like its shape, size, gray level intensities and its locations. In this paper, we propose a new cooperative scheme that applies a semi-supervised fuzzy clustering algorithm. Specifically, the Otsu (Oral Tracheal Stylet Unit) method is used to remove the Background area from a Magnetic Resonance Image. Finally, Semi-supervised Entropy Regularized Fuzzy Clustering algorithm (SER-FCM) is applied to improve the quality level. The intensity, shape deformation, symmetry and texture features were extracted from each image. The usefulness and significance of this research are fully demonstrated within the extent of real-life application.