Reproduce “HADS-Net:A Hybrid Attention-Augmented Dual-Stream Network with Physics-Informed Augmentation for Breast Ultrasound Image Classification” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Chinedu Emmanuel Mbonu, Blessing Nwamaka Iduh, Joseph Ikechukwu Odo, Doris Chinedu Asogwa
HADS-Net, a hybrid attention-augmented dual-stream network, was developed to classify breast ultrasound images into benign, malignant, and normal categories using physics-informed augmentation and a cross-attention fusion module. This method achieved an overall accuracy of 96.58% and a Macro ROC-AUC of 0.9978 on the BUSI dataset, notably avoiding misclassification of malignant cases as normal.
Key insights
- Simulating ultrasound-specific artifacts like speckle noise, acoustic shadowing, and gain variation through physics-informed augmentation improves model robustness and generalization.
- Explicitly processing both global texture and local lesion boundary features in a dual-stream architecture provides a more comprehensive diagnostic representation.
- A cross-attention mechanism enables dynamic and context-aware integration of complementary features, which is superior to simple concatenation for classification.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.