Research map / Video / Video-language understanding
Clinically-Grounded Counterfactual Reasoning for Medical Video Diagnosis
Jianzhe Gao, Churan Wang, Weiyi Zhang, Jianghua Li, Li-An Li, Wenguan Wang, Yixin Zhu, Yizhou Wang
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 clinical rules (CRs) to the baseline learner (ablation) | I3D medical video learner · Colposcopy biopsy-site localization (in-house, fully supervised) | Recall@1: 77.9 -> 89.4 | Improved |
| add clinical rules (CRs) to the baseline learner (ablation) | I3D medical video learner · Colonoscopy polyp-frame detection (Kvasir + LDPolypVideo, weakly supervised) | AP: 82.8 -> 91.6 | Improved |
| add dual diagnostic prediction (DDP) to the baseline learner (ablation) | I3D medical video learner · Colposcopy biopsy-site localization (in-house, fully supervised) | Recall@1: 77.9 -> 80.2 | Improved |
| add dual diagnostic prediction (DDP) to the baseline learner (ablation) | I3D medical video learner · Colonoscopy polyp-frame detection (Kvasir + LDPolypVideo, weakly supervised) | AP: 82.8 -> 85.5 | Improved |
| add DDP on top of clinical rules (ablation) | MedVCR · Colposcopy biopsy-site localization (in-house, fully supervised) | Recall@1: 89.4 -> 93.0 | Improved |
| add DDP on top of clinical rules (ablation) | MedVCR · Colonoscopy polyp-frame detection (Kvasir + LDPolypVideo, weakly supervised) | AP: 91.6 -> 94.8 | Improved |
| add temporal consistency rule (ablation) | MedVCR with DDP · Colposcopy biopsy-site localization (in-house, fully supervised) | Recall@1: 80.2 -> 84.1 | Improved |
| add pathological separability rule (ablation) | MedVCR with DDP · Colposcopy biopsy-site localization (in-house, fully supervised) | Recall@1: 80.2 -> 87.5 | Improved |
| add counterfactual alignment rule (ablation) | MedVCR with DDP · Colposcopy biopsy-site localization (in-house, fully supervised) | Recall@1: 80.2 -> 90.8 | Improved |
| add counterfactual alignment rule (ablation) | MedVCR with DDP · Colonoscopy polyp-frame detection (Kvasir + LDPolypVideo, weakly supervised) | AP: 85.5 -> 93.0 | Improved |
| use I3D visual encoder instead of CLIP-ViT/B (ablation) | MedVCR · Colposcopy biopsy-site localization (in-house, fully supervised) | Recall@1: 84.5 -> 93.0 | Improved |
| use I3D visual encoder instead of ResNet101 (ablation) | MedVCR · Colposcopy biopsy-site localization (in-house, fully supervised) | Recall@1: 89.1 -> 93.0 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- STDDNet — STDDNet: Harnessing Mamba for Video Polyp Segmentation via Spatial-aligned Temporal Modeling and Discriminative Dynamic Representation Learning (ICCV 2025) · better on 1 of 1 shared result
- VideoMAEv2 — VideoMAE V2: Scaling Video Masked Autoencoders With Dual Masking (CVPR 2023) · better on 1 of 1 shared result
- MIST — MIST: Multiple Instance Self-Training Framework for Video Anomaly Detection (CVPR 2021) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Medical thinking with multiple images (ICLR 2026 · similarity 0.85)
- Patho-R1: A Multimodal Reinforcement Learning-Based Pathology Expert Reasoner (AAAI 2026 · similarity 0.85)
- MedHorizon: Towards Long-context Medical Video Understanding in the Wild (NeurIPS 2026 · similarity 0.85)
- MedEyes: Learning Dynamic Visual Focus for Medical Progressive Diagnosis (AAAI 2026 · similarity 0.84)
- Med-CMR: A Fine-Grained Benchmark Integrating Visual Evidence and Clinical Logic for Medical Complex Multimodal Reasoning (CVPR 2026 · similarity 0.84)
- M3CoTBench: Benchmark Chain-of-Thought of MLLMs in Medical Image Understanding (ICLR 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.
