Breast cancer continues to pose a significant global health challenge, emphasizing the need for advancements in early detection methods. This study explores the application of transfer learning techniques, specifically utilizing EfficientNet, to enhance the accuracy of breast cancer detection through medical imaging. Leveraging a dataset of mammography images from the Digital Database for Screening Mammography (DDSM), the research implements various data preprocessing methods, including median filtering, contrast enhancement, and artifact removal, to ensure the quality of input data. The EfficientNet model, trained with these preprocessed images, is evaluated against other transfer learning architectures, such as DenseNet and ResNeXt50, using metrics like accuracy, AUC, precision, and F1-score. The results demonstrate that EfficientNet outperforms other models, achieving an accuracy of 95.23%, with a sensitivity of 96.67% and specificity of 93.82%. These findings suggest that transfer learning, particularly with EfficientNet, can significantly improve the predictive accuracy of breast cancer detection, offering a reliable tool for early diagnosis and personalized treatment planning. The study also discusses the potential integration of these models into clinical workflows, addressing challenges such as data privacy, model generalizability, and clinical applicability. Future research will focus on expanding the dataset and exploring the use of other advanced deep learning techniques to further enhance detection accuracy and robustness.