Brain tumors rank among the most critical and life-threatening conditions, characterized by elevated mortality rates resulting from their aggressive and infiltrative properties. The prompt identification and precise delineation of brain tumors are essential for optimal treatment strategies and enhancing patient prognoses. The integration of attention and residual blocks into CNN designs, especially for multi-modality MRI data, is a viable approach to overcoming existing obstacles in brain tumor segmentation. Processing several MRI modalities through distinct routes enables a more thorough examination of brain tumors, resulting in enhanced accuracy in segmenting intricate tumor regions. This research enhances CNN-based models by using these techniques to establish a robust framework adept at managing real clinical data, providing practical applications in the medical domain.