A Novel Scheme for Detection & Feature Extraction of Brain Tumor by Magnetic Resonance Modality Using DWT & SVM

A Tumor in the brain is considered as a frightening disease. It is some blob that emerges in the brain from anomalous cell growth. The detection and classification of tumor infected part of the brain through Magnetic Resonance Modality is a tiresome and protracted task. With the aid of various techniques of Image Processing, the diverse composition of the human body can be envisioned. However it is quite complicated to detect anomalous brain structures with normal imaging modalities. Modality like Magnetic resonance imaging (MRI) discriminates and illuminate the human neural design. Theirs exist various imaging techniques that examine the inner composition of brain. In the paper presented an investigative procedure for brain tumor (BT) identification is developed. The paper concentrates on the removal of noise, gray-level co-occurrence matrix (GLCM) based features extraction and Discrete Wavelet Transform (DWT) based BT segmentation in order to enhance the performance and shrink the intricacy. Morphological operation is performed to eradicate the noise emerged due to segmentation. Support Vector Machine (SVM) based classifier is used to access the accuracy of BT detection. Experimental results exhibit a classification accuracy of 98.87% that shows the proposed system efficacy.