Total: 1
To address the issues of limited samples, time-consuming feature design, and low accuracy in detection and classification of breast cancer pathological images, a breast cancer image classification model algorithm combining deep learning and transfer learning is proposed. This algorithm is based on the DenseNet structure of deep neural networks, and constructs a network model by introducing attention mechanisms, and trains the enhanced dataset using multi-level transfer learning. Experimental results demonstrate that the algorithm achieves an efficiency of over 84.0\% in the test set, with a significantly improved classification accuracy compared to previous models, making it applicable to medical breast cancer detection tasks.