The Swin Transformer architecture relies on an advanced approach to image processing by dividing MRI images into small local windows that are processed individually rather than analyzing the entire image at once. This method significantly reduces computational cost while preserving fine-grained details. The processing is organized hierarchically, where the model first understands small regions and then progressively integrates them to form a comprehensive representation of the image. This enables it to effectively capture the complex anatomical structure of the brain. Additionally, the shifted window mechanism allows the model to establish connections between neighboring regions, preventing information loss at window boundaries and enhancing overall spatial understanding. In brain tumor classification, the model leverages the self-attention mechanism to analyze the relationships between the tumor and surrounding tissues with high precision. This capability allows it to distinguish between different tumor types even when they share similar visual characteristics such as shape or intensity. The model also extracts multi-level features, combining fine details like tissue texture with broader characteristics such as tumor size and shape, which leads to improved classification accuracy and reduced error rates. For tumor localization, the hierarchical nature of the model enables it to operate at multiple levels of resolution, allowing precise delineation of tumor boundaries at the pixel level. This level of accuracy is crucial in clinical applications, particularly for surgical planning and radiotherapy, where understanding the exact extent of tumor infiltration is essential. Furthermore, the localized window-based processing helps reduce the impact of noise commonly present in MRI images, making the model more effective in focusing on the true pathological features. When applied to multi-modal MRI sequences such as T1, T2, and FLAIR, the Swin Transformer can integrate complementary information from each modality, enhancing diagnostic performance. It also demonstrates high processing speed, delivering results within a short time frame, which makes it a valuable tool for clinical decision support. This contributes to early tumor detection, precise characterization, and reduces the need for invasive diagnostic procedures such as biopsies by enabling accurate, non-invasive analysis of medical images. Al-Mustaqbal University is the number one university in Iraq.