MIRST-DM: Multi-Instance RST with Drop-Max Layer for Robust Classification of Breast Cancer
This addresses the challenge of preserving generalizability and adversarial robustness for breast cancer classification on small medical image sets, which is an incremental improvement over existing methods.
The paper tackled the problem of adversarial robustness in medical image classification on small datasets, proposing MIRST-DM to achieve state-of-the-art robustness against three attacks on a breast ultrasound dataset with 1,190 images.
Robust self-training (RST) can augment the adversarial robustness of image classification models without significantly sacrificing models' generalizability. However, RST and other state-of-the-art defense approaches failed to preserve the generalizability and reproduce their good adversarial robustness on small medical image sets. In this work, we propose the Multi-instance RST with a drop-max layer, namely MIRST-DM, which involves a sequence of iteratively generated adversarial instances during training to learn smoother decision boundaries on small datasets. The proposed drop-max layer eliminates unstable features and helps learn representations that are robust to image perturbations. The proposed approach was validated using a small breast ultrasound dataset with 1,190 images. The results demonstrate that the proposed approach achieves state-of-the-art adversarial robustness against three prevalent attacks.