Researchers have developed DeCo-MIL, a novel approach to address the challenges of long-tailed distributions in whole slide image (WSI) analysis. This method tackles both inter-slide and intra-slide long tails by employing frequency-debiased counterfactual reasoning. DeCo-MIL clusters patches, intervenes with normal prototypes to estimate counterfactual contributions, and uses these to preserve scarce discriminative instances. It also constructs anchor-stratified pseudo-bags and employs tail-aware oversampling to enhance supervision for rare classes. Experiments on three benchmarks show DeCo-MIL achieving state-of-the-art performance. AI
IMPACT This method could improve diagnostic accuracy in medical imaging by better handling rare conditions.
RANK_REASON The cluster contains a research paper detailing a new method for image analysis. [lever_c_demoted from research: ic=1 ai=1.0]
- alphaXiv
- arXiv
- CatalyzeX
- CORE Recommender
- DagsHub
- DeCo-MIL
- Gotit.pub
- Hugging Face
- Multiple instance learning
- ScienceCast
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
