Because of high heterogeneity and invasiveness, treatment of GlioBlastoma Multiform (GBM) still remains a complex challenge. Several recent advanced therapies have improved precision of treatment deliverance. Multimodality imaging plays an increasingly important role in this process and images segmentation has become an essential part of the pipeline of standard treatment planning system. With the sophistication of multimodality information, the development of reliable and robust segmentation algorithms to overcome manual segmentation and optimize targeted treatment is highly expected.

In this paper, we first introduce targeted therapies applied in the GBM clinical care, from routine or research. Different segmentation methods from state of the art are highlighted to achieve GBM delineation. New trends in GBM segmentation such as machine learning and multimodal features are discussed. These additional frameworks may achieve segmentation with refining capacities, active tumour probability mapping and, even, tumour relapse prediction capacities.