부산대학교 도서관

Glioblastoma delineation and its related active region specification are still a real challenge and so difficult essentially due to their multiform aspect. in fact, this type of tumors is very invasive and appears as non-enhancing region and with various forms on magnetic resonance imaging modalities. Thus, Glioblastoma segmentation is challenging especially in differentiating between white matter and edema, necrosis and gray matter due to their homogeneity in intensity and texture. An accurate delineation of the tumor is necessary for the tumor progress evaluation and medical treatment efficacy assessment. in addition, a precise limitation of the tumor is mandatory in surgical and Radio Therapies. Manual segmentation methods have been always used and require radiologist intervention and could be also used as reference. our attention was for the benefits to extract from the semi-Automatic segmentation Methods and the Fully Automatic segmentation Methods, and this would yield a real complementarity giving hence one complete and rich convivial clinical aided tool. This paper presents therefore a useful review of these methods proposed for the Glioblastoma MRI segmentation.