Breast cancer is the most prevalent cancer among women globally, with various types of lesions identified, including microcalcifications, which can be benign, malignant, or benign without a callback.
The manuscript introduces an automated detection pipeline utilizing convolutional neural networks (CNNs) to classify different categories of microcalcifications, testing four optimization algorithms.
The classification system demonstrated superior performance over existing deep learning and traditional machine learning methods, verified through evaluation metrics such as sensitivity, specificity, accuracy, and area under the curve (AUC) on the CBIS-DDSM mammogram dataset.