Breast cancer is the leading cause of death among women globally. Early detection of abnormalities is crucial for enhancing treatment effectiveness. Mammography, the most common screening tool for breast cancer, presents challenges due to the manual interpretation of mammograms and variability among radiologists. In this paper, we propose an automatic breast mass detection system. This system utilizes fuzzy logic for image preprocessing, in conjunction with deep learning network models like U-Net and Mask R-CNN for mass segmentation. Our experimental results indicate that this method improves the quality of input images, efficiently retains crucial features, and aids in accurately localizing masses. It achieved a high accuracy rate of 92.5% on the data set collected from several hospitals. These results serve as a significant foundation for our ongoing development of fuzzy learning techniques in classifying and identifying masses in mammograms.