In recent years, there has been significant attention towards deep neural networks and they are increasingly being applied to clinical practices. Medical imaging technology, biomedical data analysis, computer-aided diagnosis, and healthcare in general are all to gain immensely from these breakthroughs which are just now becoming apparent. Our study focuses on Magnetic Resonance Imaging (MRI) scans, an area of growing interest in medical research. The availability of sophisticated, low-cost computer hardware over the past decade has been the primary driving force in the advancement of computer vision in medical research, which has led to immense developments in digital MRI image analysis that ranges from simple qualitative analysis of disease detection to acquiring more insight into its nature. As deep learning techniques evolve and more data becomes available, new opportunities arise to further explore and optimize their utilization in MRI analysis. This study aims to contribute to this dynamic area of research by providing a broad overview of the latest applications and identifying potential areas for continued growth and improvement. The paper is an attempt to address the research gap by exploring how deep learning techniques and architectures can be effectively employed in various stages of the classical medical image processing workflow for MRI scans, from image acquisition and retrieval to preprocessing, segmentation, feature extraction and disease detection or prediction. The intended outcome of this study is an up-to-date, comprehensive summary of the state-of-the-art deep learning techniques employed in MRI-based research. With an understanding of the present landscape, researchers and medical practitioners can garner valuable insights into the recent advancements and identify prospective avenues for future investigations. By emphasizing the potential of deep learning in medical imaging and providing a clear perspective on its applications and limitations, t...