Advancements in Breast Cancer Detection: A Comprehensive Review of Deep Learning Techniques for Mammogram Analysis

Biomedical advances in breast cancer detection are playing an important role in reducing the global death rate. Early and accurate identification of breast cancer not only increases life expectancy for patients but also fostering a paradigm shift to medical care and cancer management. Mammography is the primary screening tool for breast cancer, offering non-invasive imaging of breast tissue to identify potential abnormalities. However, mammogram interpretation is challenging due to its complex nature, and the risk of false negatives and false positives remains a significant concern. In recent years, learning-based methods have revolutionized mammogram analysis, showing tremendous potential in early breast cancer detection. This review paper presents a comprehensive analysis of various learning-based network architectures used for breast cancer detection and classification. Some models are very accurate, almost perfect, like ResNet18, VGG-16, and Inception V4. Some models may not be as accurate but are still useful, like RESNET-50 and MobileNet. The paper will also compare the strengths and limitations of these architectures, emphasizing the need for careful selection and potential combination of their best features to design more effective and robust models.