Research map / Vision-language and multimodal / Human and embodied modeling
AttTok: Marrying Attribute Tokens with Generative Pre-trained Vision-Language Models towards Medical Image Understanding
Hualiang Wang, Xinyue Xu, Lehan Wang, Bin Pu, Xiaomeng 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 |
|---|---|---|---|
| add AttTok attribute tokens | Qwen2.5-VL-7B · 5-modality diagnosis (open-end avg) | accuracy: 65.8 -> 69.6 | Improved |
| add AttTok attribute tokens | Lingshu-7B · 5-modality diagnosis (open-end avg) | accuracy: 67.2 -> 71.8 | Improved |
| add AttTok attribute tokens | Lingshu-7B · Medical VQA avg | accuracy: 73.2 -> 74.0 | Improved |
| add AttTok attribute tokens | Qwen2.5-VL-7B · Medical VQA avg | accuracy: 71.6 -> 72.5 | Improved |
| add ACC adapter and ACM loss jointly rather than either alone (ablation) | AttTok · five modalities | accuracy | Improved |
| add ACM loss alone vs ACC alone (ablation) | AttTok · five modalities | accuracy | Improved |
| add visual-only clustering alignment (ablation) | Lingshu-7B · Fundus and OCT (open-end) | accuracy: 63.7 -> 63.2; 74.7 -> 74.2 | Hurt |
| add vision-text alignment (ablation) | Lingshu-7B · Dermatology (open-end) | accuracy: 72.8 -> 73.6 | Improved |
| use AttTok alignment instead of best alternative (ablation) | Lingshu-7B · Dermatology, Fundus, OCT | accuracy: +3.3 Dermatoscopy, +3.2 Fundus, +2.7 OCT | Improved |
| add AttTok | Lingshu-7B · Fitzpatrick17k | accuracy: 0.27 -> 0.41 | Improved |
| add AttTok | Lingshu-7B · RFMiD | mAP: 0.25 -> 0.36 | Improved |
| add AttTok without pre-balancing | Lingshu-7B · ISIC-2018 imbalanced, DF class | F1: 11.8 -> 31.4 | Improved |
Closest papers by meaning
- OpenMedReason: Scientific Reasoning Supervision for Medical Vision–Language Models (NeurIPS 2026 · similarity 0.87)
- Falcon Medical Visual Question Answering (AAAI 2025 · similarity 0.87)
- Bringing CLIP to the Clinic: Dynamic Soft Labels and Negation-Aware Learning for Medical Analysis (CVPR 2025 · similarity 0.87)
- MedSIGHT: Towards Grounded Visual Comprehension in Medical Large Vision-Language Models (ICML 2026 · similarity 0.86)
- Alignment, Mining and Fusion: Representation Alignment with Hard Negative Mining and Selective Knowledge Fusion for Medical Visual Question Answering (CVPR 2025 · similarity 0.86)
- Prompt as Knowledge Bank: Boost Vision-language model via Structural Representation for zero-shot medical detection (ICLR 2025 · similarity 0.86)
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.
