Breast Cancer Detection Using CNNs on Mammogram Images: A Dataset-Level Comparison of CBIS-DDSM, INbreast, and MIAS

This study presents a CNN-based deep learning model for the automated diagnosis and classification of breast cancer using only mammographic images. Mammograms are a primary tool in breast cancer screening due to their accessibility and cost-effectiveness. The proposed model performs end-to-end learning from raw mammographic inputs through convolutional layers to predict benign or malignant conditions. Preprocessing techniques such as contrast enhancement and noise removal are applied to improve image quality. The model is trained and validated using publicly available mammography datasets (e.g., CBIS-DDSM), achieving high diagnostic performance in terms of accuracy, sensitivity, and specificity. Explainability techniques like Grad-CAM are incorporated to visualize regions of interest contributing to the diagnosis. DOI: https://doi.org/10.17762/ijisae.v12i22s.7632