**6. Location of GSCs in the tumor and their detection by neuro-imaging**

If GSCs were considered as a subset of special cells, they should be located somewhere in the tumor and therefore their search *in vivo* could be justified. According to the hypothe‐ sis that they represent a functional status, they should appear in the tumor when and where, as the consequence of the transformation process, tumor cells reach the threshold of stemness. In some experience, NS would be generated from whatever part of GBM [126], whereas in some other experiences [48], different subsets of GSCs arise from regions of GBM with different malignancy potential, showing different tumorigenic potential and genetic abnormalities, even though originating from the same ancestor cell. Since GSCs reside in niches, their distribution in the tumor should follow that of niches which in turn with their microenvironments develop where malignancy occurs [52]. On the other hand, it has been observed that GSCs occur in the hypoxic area between the central necrosis and the proliferating zone of GBM [127].

Until today, the only mean to detect GSCs is to apply the neurosphere assay to the surgical samples removed from different parts of the tumor. Their recognition can be therefore achieved only after surgery. It would be highly useful to know in advance where in tumor GSCs are located or generated in order to try to annihilate them without surgery and to cure the patient. Can they be detected by MRI or other procedures *in vivo*?

Animal *in vivo* imaging techniques have been applied to some stem cell populations – hema‐ topoietic and leukemic stem cells – but the application to solid tissues has been limited [41]. Using intravital microscopy, labelled GSCs could be followed in their propagation and responsibility in producing glioma heterogeneity [41], but data are not available by MRI techniques. Bone marrow-derived endothelial precursors, labelled by super-paramagnetic iron oxide nanoparticles, could be demonstrated in glioma-bearing immunodeficient SCID mice by MRI [128], but no similar procedure has been adopted for GSCs. The only possibility is to use the spatial relationship between MRI variables and tumor phenotypes [33] including into the phenotype the expression of GSC stemness status.
