<![CDATA[Breast cancer remains one of the most common and deadly diseases affecting women worldwide, highlighting the importance of early and accurate diagnosis to improve treatment outcomes and survival rates. However, traditional mammography techniques often fall short, failing to detect up to 20% of cases, especially in women with dense breast tissue, which makes detection more difficult. In response to these limitations, this study explores the use of neural networks to enhance diagnostic accuracy in breast cancer detection, focusing on the Convolutional Neural Network (CNN) and Multilayer Perceptron (MLP). Utilizing the Wisconsin Diagnostic Breast Cancer (WDBC) dataset, a baseline CNN model is compared against an optimized CNN refined through hyperparameter tuning using randomized search, as well as two MLP models implemented via Keras and Scikit-learn, along with their optimized versions. Each model is evaluated using key classification metrics, including accuracy, precision, recall, F1-score, and AUC, with an emphasis on minimizing false negatives, as this is critical in medical diagnosis to avoid missed malignancies. The results indicate that the optimized CNN model achieved near-perfect scores across all metrics and demonstrated the best balance between training and testing data. Therefore, it outperforms the baseline CNN and MLP models in significantly reducing false negatives, showcasing the potential of a well-tuned CNN to enhance the automation and reliability of breast cancer diagnostic processes.]]>
