Research map / Vision-language and multimodal / Cross-modal learning
SeekFlow: Synergizing Radiology and Pathology Foundation Models for Precision Oncology via Knowledge-Guided Evidence Flow
Peixiang Huang, Yanyan Huang, Yihang Chen, Maximus Yeung, Yuming Jiang, Lequan Yu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add routed evidence anchors (ablation) | SeekFlow · Overall (macro-avg over tasks) | C-index: 0.641 -> 0.699 | Improved |
| use shared plus modality-specific prototypes instead of all shared (ablation) | SeekFlow · Overall (macro-avg over tasks) | AUC: 0.833 -> 0.884 | Improved |
| combine knowledge prior with data-driven routing instead of full knowledge (ablation) | SeekFlow · Overall (macro-avg over tasks) | C-index: 0.617 -> 0.699 | Improved |
| use knowledge-guided flow matching instead of discrete OT assignment (ablation) | SeekFlow · Overall (macro-avg over tasks) | AUC: 0.567 -> 0.884 | Improved |
| add density-aligned target cost (ablation) | SeekFlow · Overall (macro-avg over tasks) | C-index: 0.668 -> 0.699 | Improved |
| use hierarchical evidential synthesis instead of flat aggregation (ablation) | SeekFlow · Overall (macro-avg over tasks) | C-index: 0.577 -> 0.699 | Improved |
| use uncertainty-biased cross-modal consensus instead of multi-head attention (ablation) | SeekFlow · Overall (macro-avg over tasks) | AUC: 0.837 -> 0.884 | Improved |
| use router-guided intra-modal gating instead of standard gating (ablation) | SeekFlow · Overall (macro-avg over tasks) | AUC: 0.878 -> 0.884 | Improved |
| swap CONCH+BiomedCLIP for CONCH+GenMedCLIP (ablation) | SeekFlow · Overall (macro-avg over tasks) | AUC: 0.884 -> 0.895 (C-index 0.699 -> 0.661) | Mixed |
| swap pathology encoder CONCH for MUSK (ablation) | SeekFlow · Overall (macro-avg over tasks) | AUC: 0.884 -> 0.840 | Hurt |
| change knowledge LLM from Gemini 3.0 Pro to GPT-5.5 (ablation) | SeekFlow · Overall (macro-avg over tasks) | C-index: 0.699 -> 0.684 | No clear effect |
| SeekFlow vs best baseline on classification | SeekFlow · Chondrosarcoma staging | AUC: 0.776 -> 0.820 | Improved |
Closest papers by meaning
- TumorChain: Interleaved Multimodal Chain-of-Thought Reasoning for Traceable Clinical Tumor Analysis (ICLR 2026 · similarity 0.87)
- A Textbook Remedy for Domain Shifts: Knowledge Priors for Medical Image Analysis (NeurIPS 2024 · similarity 0.85)
- Modeling Dense Multimodal Interactions Between Biological Pathways and Histology for Survival Prediction (CVPR 2024 · similarity 0.84)
- HEALNet: Multimodal Fusion for Heterogeneous Biomedical Data (NeurIPS 2024 · similarity 0.84)
- Leveraging Tumor Heterogeneity: Heterogeneous Graph Representation Learning for Cancer Survival Prediction in Whole Slide Images (NeurIPS 2024 · similarity 0.84)
- Gastric-X: A Multimodal Multi-Phase Benchmark Dataset for Advancing Vision-Language Models in Gastric Cancer Analysis (CVPR 2026 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
