I. Introduction
Brain glioma tumor could be considered as among serious pathologies threatening people. An estimated 37,890 Americans were diagnosed with glioma since2011. Refering to the statistics published by the American society of clinical oncology, the five year survival rate for patients with glioma is 2.9% [1]. Hence, early detection and classification of this pathology becomes important. Magnetic resonance imaging is the main modality used to brain glioma diagnosis. We could notice the scarcity of CAD system dedicated to brain glioma tumor classification based on their radiological appearance. Based on their radiologic appearance, we can classify brain glioma tumor into for classes: non-enhanced, full-enhanced without edema, full-enhanced with edema and ring-enhanced tumors. Computer aided diagnosis (CAD) is then highly recommended since it would be able to detect and even identify brain glioma tumor at an earlier stage [2], [3] and classify them so as to decide correctly about adequate action to conduct. Moreover, this clinical tool offer clinicians the opportunity to process a large dataset in a reduced time in a clearer and accurate process. During our research, we focused on the four key steps: preprocessing step, segmentation step, 3D reconstruction and classification.