<![CDATA[Breast cancer remains one of the most common and deadly diseases among women globally. Early detection significantly increases the chances of patient recovery. The main objective of this research is to evaluate the performance of three Convolutional Neural Network (CNN) architectures, namely ResNet50, VGG16, and DenseNet201, for breast cancer image classification. In this study, there are two classification stages used: the first is to differentiate between normal and abnormal images, and the second is to distinguish between benign and malignant tumors. The dataset was obtained through the Kaggle website. It was then pre-processed using normalization and augmentation through flipping and rotation. After each CNN model was trained using transfer learning, its performance was evaluated using accuracy, precision, recall, and F1 score. In the Normal and Abnormal classification task, the DenseNet201 model outperformed other models with an accuracy of 91%. Meanwhile, ResNet50 showed the most optimal results in the Benign and Malignant classification with an accuracy of 83%.]]>
Copyrights © 2025
