Hybrid GRU-Down-U-Net Deep Learning Model for Breast Invasive Ductal Carcinoma Classification

Breast cancer(BC) is a type of tumor that start to proliferate in the breast cells and, in recent years, it has gradually become a major issue for women's death. Because the early detection and diagnosis of breast cancer can increase the cure rates, scientists keep finding ways to detect breast cancer earlier. However, the traditional artificial detecting method is a time-consuming task and both its efficiency and precision is relatively low. In order to improve the efficiency and precision, researchers have begun using machine learning models to detect breast cancer and got great results. In our paper, we propose the GRU-Down-U-Net model, which is modified by the GRUU-Net model, for the automatic classification of breast invasive ductal carcinoma(BC-IDC). We trained, validated and tested the model on the dataset, and calculated the accuracy, specificity, sensitivity and F1 score during the testing process to detect the effectiveness of the hybrid model. The results showed that our model performed better than previous models of other researches, which indicates that our model is more robust than previous models during BC classification.