Accurate segmentation of brain tumors is essential for precise diagnosis, effective treatment planning, and predicting patient outcomes. Manually identifying and segmenting brain tumors from MRI scans is challenging and prone to errors. This paper introduces an enhanced Gated Attention U-Net Model for brain tumor segmentation utilizing the BraTS 2020 dataset. By incorporating multiple pre-surgery MRI scans, the primary objective is to improve segmentation accuracy. Through image preprocessing, data generation, and gated attention mechanisms, this research presents a comprehensive method for detecting and segmenting brain tumors. Implemented with TensorFlow and Keras, the model uses gated attention mechanisms to emphasize relevant features while suppressing irrelevant ones. This attention-based technique significantly enhances accuracy, achieving over 99%, and optimizes several critical parameters for precise tumor segmentation. These advancements make the model suitable for real-time application in brain tumor segmentation, aiding early diagnosis and intervention for brain cancer patients.