Biomarker Inference
Cross-source consensus on Biomarker Inference from 1 sources and 5 claims.
1 sources · 5 claims
How it works
Benefits
Comparisons
Highlighted claims
- Fusion retrieval combined H&E and metadata-only text similarity to retrieve mIF gallery patches. — Linking spatial biology and clinical histology via Haiku
- Biomarker inference removed explicit biomarker content from text and used a similarity-weighted average over retrieved mIF patches. — Linking spatial biology and clinical histology via Haiku
- H&E plus text fusion improved zero-shot biomarker inference over H&E-only Haiku. — Linking spatial biology and clinical histology via Haiku
- Clinical context contributed complementary information beyond morphology for biomarker inference and mIF retrieval. — Linking spatial biology and clinical histology via Haiku
- The poor MUSK biomarker inference performance was attributed to collapsing high-dimensional mIF channels into three-channel input. — Linking spatial biology and clinical histology via Haiku