One of the most challenging tasks in the field of medical image analysis is segmenting a brain tumour. The segmentation method for brain tumours has as its goal the development of an accurate delineation of brain tumour locations. Deep learning algorithms have recently demonstrated performance that is encouraging in addressing a variety of computer vision difficulties. These difficulties include semantic segmentation, object recognition, and picture categorization. To segregate brain tumours, researchers have tried several approaches that are based on deep learning, and the findings have been encouraging. Within the scope of this paper is a comprehensive investigation of newly developed deep learning-based algorithms for the segmentation of brain tumours. We have presented this information considering the remarkable developments made possible by cutting-edge technology. This study made compressive analyses of many state-of-the-art models, delving thoroughly into technical areas including network architecture designing, segmentation in unbalanced scenarios, and multi-model process. In addition, we encourage thoughtful discussions on potential future growth directions.