The major player in the revolution of early detection and diagnosis of brain tumors, with great implications for patient outcomes, is medical image processing. It is an inherently difficult and time-consuming task to manually classify brain tumors by experienced experts, even though manual classification has proven effective. A promising avenue has emerged as the integration of automatic segmentation techniques, which promises improved efficiency and performance in response to these challenges. This long work aims to provide an in-depth and critical analysis of MRI-based brain tumor segmentation techniques, with a critical eye toward the most recent developments in automatic segmentation techniques. Our analysis explores the rapidly changing field of completely automatic segmentation approaches, which diverges from the evaluations that mostly focus on traditional methodologies. The discussion opens with a broad summary that emphasizes how important brain tumor segmentation is to medical image processing as a whole. Here, we highlight how crucial precise segmentation is to facilitating early detection and guiding treatment choices later on. We recognize the difficulties that come with manual segmentation procedures and explain why automation segmentation techniques are necessary to reduce these difficulties and bring about increased productivity. The central section of the work navigates the complex terrain of cutting-edge algorithms, enabling a thorough investigation of the most recent developments in completely autonomous segmentation techniques. This thorough explanation highlights the growing acceptance and increased effectiveness of modern methods while addressing the complexities and difficulties present in the field of brain tumor segmentation. Using specially crafted neural networks, our research is unique in that it concentrates on the paradigm shift toward fully autonomous segmentation. Brain tumor segmentation has been transformed by the incorporation of deep learning techniques, which enable complex pattern recognition and nuanced analysis using medical imaging data. Our efforts have resulted in the creation of a unique neural network model specifically intended for the automated identification of brain malignancies. The talk highlights how deep learning techniques can have a revolutionary effect, and it ends with the creation of a sophisticated custom neural network model. Our model demonstrates its ability to accurately and automatically detect brain tumor boundaries by achieving a remarkable level of accuracy.