ResMobileNet: A Deep Ensemble Approach for Classification of Breast Cancer Using Transfer Learning

Breast cancer remains one of the leading causes of death in women worldwide, and thus there is a great need for reliable diagnostic tools. This paper proposes an improved ensemble approach for breast cancer classification using mammography images, which integrates the strengths of two advanced transfer learning architectures: ResNet and MobileNetV2. The ensemble model combines the feature extraction capabilities of ResNet with the computational efficiency of MobileNetV2 to achieve higher diagnostic accuracy without compromising operational effectiveness. This approach was applied to an open mammography dataset with the use of preprocessing and data augmentation techniques for the better generalization of models. The metrics such as accuracy, precision, recall, and F1-score depicted that ensemble methods perform better than the single model, hence being a good approach toward the early detection of breast cancer. This work holds promise toward further integration of deep learning frameworks for advancing computer-aided diagnosis systems in the field of medical imaging.