The abstract Early detection and treatment of brain tumors are essential in the field of medical imaging. Brain tumor detection plays a critical role in identifying and addressing the condition promptly. Convolutional Neural Networks (CNNs), such as U-Net, have demonstrated their effectiveness in segmenting images. In this study, a brain tumor detection method utilizing U-Net is proposed, which consists of three key steps: preprocessing, segmentation, and post-processing. The first step, preprocessing, involves normalizing the input images. This process ensures consistency in brightness and contrast, enhancing the accuracy of subsequent analysis. By optimizing the quality of the input images, a solid foundation for tumor identification is used. The segmentation step employs U-Net, a powerful CNN architecture, to identify the tumor region within the brain MRI images. U-Net leverages convolutional and pooling layers to extract features, enabling accurate tumor region detection. This step serves as the core component of the proposed method, enabling precise tumor identification. In the post-processing step, false positives and false negatives are addressed. By removing areas.