LOWER GRADE GLIOMA DETECTION USING MRI IMAGE

Brain tumor is one of the most serious disease which is formed by abnormal cells and should be identified before successive aggression. Glioma is the type of brain tumor which originates inside the brain glial cells and gets proliferated in malignant cases. Detection of lower-grade glioma at the early stage is complex and challenging task. Magnetic Resonance Imaging is a very important modality as it provides information about abnormal tissues developed in soft tissues. The existing classification techniques relies on algorithms such as Decision Tree (DT), Artificial neural Network, Backward propagation Network and Support Vector Machine (SVM). This paper develops a novel algorithm to detect the type of brain tumor called lower grade glioma by extending the basic novel SVM. Initially the DICOM Brain tumor images are retrieved from the database and they are preprocessed to remove the speckles present at the background. Secondly the features of the abnormal and normal brain images are extracted. Thirdly the dual step SVM is implemented. In the first step of dual SVM, the features are extracted from abnormal and normal brain images to train the SVM. In the second step, the SVM model is trained to classify lower grade glioma and other tumor type. Lower grade glioma is the initial tumor stage and it can be identified with higher accuracy using the above said algorithm which can definitely help the surgeons to preplan the surgery before aggression.