Classification of Brain Tumor on Magnetic Resonance Imaging Using Support Vector Machine

Early diagnosis of brain tumors is a crucial process for detecting human brain tumors. Various ways are conducted for the early detection of brain tumors, one with the anatomy of a digital image. A digital image through magnetic resonance imaging (MRI) is one of the methods that help doctors both analyze and classify the type of brain tumor. However, tumor detection on MRI images is usually done through manual by the doctors. In this study, a system was created that could assist doctors in detecting brain tumors based on MR images. If a tumor is identified appropriately at an early stage, the chances of survival can be improved. The methodology consists of the following steps: pre-processing by using median filters, skull stripping to removes non-cerebral tissues, segmentation of the image is performed by thresholding, the feature extraction from the detected tumor is realized by using statistical features which is first order features are extracted from the histogram of the image and the Gray Level Co-occurrence Matrix (GLCM) is used to extract second order features, classification techniques based on Support Vector Machines (SVM) are proposed and applied to brain image classification. From the results of the tests carried out on 48 dates, the system achieved a total accuracy, precision, sensitivity, and specificity is 95,83%, 94,08%, 93,33%, and 96,87% for classification of meningioma, glioma, and pituitary tumors respectively. This system was tested on T1-weighted CE-MRI.