Reproduce “MExD: An Expert-Infused Diffusion Model for Whole-Slide 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
Jianwei Zhao, Xin Li, Fan Yang, Qiang Zhai, Ao Luo, Yang Zhao, Hong Cheng, Huazhu Fu
MExD, an Expert-Infused Diffusion Model, presents the first generative strategy for Whole Slide Image (WSI) classification. This approach achieved state-of-the-art performance on Camelyon16, TCGA-NSCLC, and BRACS datasets, demonstrating robust prediction and enhanced uncertainty estimation.
Key insights
- Generative diffusion models, when integrated with expert-infused learning, can provide a robust and interpretable alternative to discriminative methods for complex WSI classification.
- Dynamically routing instances to specialized experts and applying top-k selection is effective in mitigating noise and data imbalance within gigapixel-scale images.
- Conditioning a 1-D diffusion process with aggregated expert insights and a prior prediction enables accurate and efficient class distribution inference, reducing the number of required denoising steps.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.