U-Net for Breast Tumor Segmentation

Early detection of cancer will reduce the mortality rate among patients. The study aims to identify the area of breast cancer. To provide patients with appropriate therapy, it is necessary to accurately observe their symptoms and have an automatic prediction system in place to determine whether a tumor is benign or malignant. An architecture based on UNETs has been proposed in this work to segment the tumor region in histopathology images. The network relies on a fully convolutional architecture, which has been updated and enhanced to perform well with fewer training images and produce more accurate segmentations.