Research map / Vision-language and multimodal / Cross-modal learning
Unified Mixture-of-Experts Framework for Joint Cardiac and Vascular Ultrasound Analysis and Report Generation
Bin Pu, Jiewen Yang, Xingguo Lv, Kai Xu, Kenli Li
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 |
|---|---|---|---|
| joint cardiac and vascular training instead of cardiac only (ablation) | ECV · CVU parameter measurement | RMSE: 42.30 -> 12.43 | Improved |
| joint cardiac and vascular training instead of cardiac only (ablation) | ECV · CVU parameter measurement | MAE: 6.15 -> 5.72 | Improved |
| add cardiac data to vascular-only training (ablation) | ECV · CVU parameter measurement | MAE: 6.89 -> 5.72 | Improved |
| add vascular data to cardiac-only training for vascular reports (ablation) | ECV · CVU vascular report generation | BertScore: 78.64 -> 81.98 | Improved |
| fine-tune vision encoder together with LLM decoder instead of decoder only (ablation) | ECV · CVU parameter measurement | MAE: 7.48 -> 5.72 | Improved |
| fine-tune LLM decoder together with vision encoder instead of encoder only (ablation) | ECV · CVU cardiac report generation | BertScore: 78.05 -> 87.37 | Improved |
| train jointly on measurement and report generation instead of parameter-only (ablation) | ECV · CVU cardiac report generation | BertScore: 63.67 -> 87.37 | Improved |
| train jointly instead of parameter-only (ablation) | ECV · CVU parameter measurement | MAE: 6.01 -> 5.72 | Improved |
| train jointly instead of report-only on cardiac reports (ablation) | ECV · CVU cardiac report generation | BertScore: 88.43 -> 87.37 | No clear effect |
| train jointly instead of report-only on vascular reports (ablation) | ECV · CVU vascular report generation | BLEU-1: 8.46 -> 24.58 | Improved |
| use ECV with MoE vision encoder and measurement module instead of HuaTuo-34B | HuaTuo-34B · CVU parameter measurement | MAE: 13.25 -> 5.72 | Improved |
| use ECV instead of HuaTuo-34B for cardiac report BLEU-4 | HuaTuo-34B · CVU cardiac report generation | BLEU-4: 13.05 -> 5.82 | Hurt |
Closest papers by meaning
- EchoXFlow: A Beamspace Echocardiography Dataset for Cardiac Motion, Flow, and Function (NeurIPS 2026 · similarity 0.87)
- U2-BENCH: Benchmarking Large Vision-Language Models on Ultrasound Understanding (ICLR 2026 · similarity 0.85)
- Self-supervised Learning of Echocardiographic Video Representations via Online Cluster Distillation (NeurIPS 2025 · similarity 0.84)
- FETAL-GAUGE: A BENCHMARK FOR ASSESSING VISION-LANGUAGE MODELS IN FETAL ULTRASOUND (ICLR 2026 · similarity 0.83)
- ECHOPulse: ECG Controlled Echocardio-gram Video Generation (ICLR 2025 · similarity 0.83)
- CardioLens: Revealing the Clinical Reality Gap of MLLMs via Multi-Sequence Cardiac MRI Evaluations (NeurIPS 2026 · similarity 0.83)
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.
