Glioblastoma is the most aggressive type of brain cancer with high levels of intra- and inter-tumour heterogeneity. However, a spatial characterization of gene signatures and the cell types expressing these in different tumour locations is still lacking. We have used deep convolutional neural networks (DCNN) as a semantic segmentation model to segment tumour regions in glioblastoma histopathological slides. We combined these results with RNA gene expression data to characterize the cellular composition of the tumour microenvironment in different tumour regions. Validation with single-cell RNA sequencing data from resected glioblastoma tissue samples further confirms that different cells in the tumour microenvironment drive gene signatures that are involved in tumour-stromal interactions, which were correlated with survival. These results pointed to a key role for interactions between stromal cells and tumour cells, which may contribute to poor survival in glioblastoma.