Deep learning, particularly using U-Net architecture, has shown remarkable performance in various image segmentation tasks, including medical and non-medical applications. This versatile approach enables automated analysis of complex images, which is crucial for improving diagnostic accuracy and efficiency. For medical applications, breast cancer detection serves as a prominent example, where deep learning models have demonstrated superior performance over traditional methods. We examine various techniques used to enhance U-Net's ability to detect breast cancer, Moreover, we review the most commonly used datasets for medical image segmentation tasks effectiveness in a range of applications. Our proposed custom U-Net model extends the standard U-Net architecture by incorporating advanced techniques to enhance its ability to handle segmentation tasks. These improvements result in improved accuracy, Intersection over Union (IOU) scores, and dice coefficient scores, setting a new benchmark forsegmentation models.