Research map / Vision-language and multimodal / CLIP and zero-shot
Boosting Vision Semantic Density with Anatomy Normality Modeling for Medical Vision-language Pre-training
Weiwei Cao, Jianpeng Zhang, Zhongyi Shui, Sinuo Wang, Zeli Chen, Xi Li, Le Lu, Xianghua Ye, Qi Zhang, Tingbo Liang, Ling Zhang
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 anatomy-wise fine-grained image-report alignment (ViT encoder) (ablation) | ViSD-Boost · MedVL-CT69K validation zero-shot | AUC: 70.9 -> 76.5 | Improved |
| use a CNN instead of ViT as vision encoder (ablation) | ViSD-Boost · MedVL-CT69K validation zero-shot | AUC: 76.5 -> 78.7 | Improved |
| add instance-level contrastive learning for visual semantic enhancement (ablation) | ViSD-Boost · MedVL-CT69K validation zero-shot | AUC: 78.7 -> 79.7 | Improved |
| use disease-level rather than instance-level contrastive learning (ablation) | ViSD-Boost · MedVL-CT69K validation zero-shot | AUC: 79.7 -> 80.7 | Improved |
| add vision semantic density boosting (VQ-VAE normality modeling) (ablation) | ViSD-Boost · MedVL-CT69K validation zero-shot | AUC: 80.7 -> 82.5 | Improved |
| initialize encoder with disease-level contrastive pretraining instead of random (ablation) | ViSD-Boost · MedVL-CT69K validation zero-shot | AUC: 78.7 -> 80.7 | Improved |
| initialize encoder with disease-level contrastive pretraining instead of supervised classification (ablation) | ViSD-Boost · MedVL-CT69K validation zero-shot | AUC: 79.4 -> 80.7 | Improved |
| use larger ResNet34 or ResNet50 instead of ResNet18 (ablation) | ViSD-Boost · MedVL-CT69K validation zero-shot | AUC: 78.7 -> 79.0 | Improved |
| compare ViSD-Boost with base model on local diseases | ViSD-Boost · MedVL-CT69K zero-shot, local diseases | AUC: 76.3 -> 80.3 | Improved |
| compare ViSD-Boost with base model on diffuse diseases | ViSD-Boost · MedVL-CT69K zero-shot, diffuse diseases | AUC: 85.2 -> 88.0 | Improved |
| replace fVLM with ViSD-Boost on MedVL-CT69K | ViSD-Boost · MedVL-CT69K zero-shot (54 diseases, 15 organs) | average AUC: 81.3 -> 84.9 | Improved |
| replace fVLM with ViSD-Boost on Rad-ChestCT precision | ViSD-Boost · Rad-ChestCT zero-shot (external) | precision: 37.4 -> 34.2 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Merlin — Merlin: Empowering Multimodal LLMs with Foresight Minds (ECCV 2024) · 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
- Large-scale and Fine-grained Vision-language Pre-training for Enhanced CT Image Understanding (ICLR 2025 · similarity 0.90)
- Better Tokens for Better 3D: Advancing Vision-Language Modeling in 3D Medical Imaging (NeurIPS 2025 · similarity 0.87)
- BenchX: A Unified Benchmark Framework for Medical Vision-Language Pretraining on Chest X-Rays (NeurIPS 2024 · similarity 0.87)
- Enhancing Medical Large Vision-Language Models via Alignment Distillation (AAAI 2026 · similarity 0.87)
- Unleashing the Potential of Vision-Language Pre-Training for 3D Zero-Shot Lesion Segmentation via Mask-Attribute Alignment (ICLR 2025 · similarity 0.86)
- MedUnifier: Unifying Vision-and-Language Pre-training on Medical Data with Vision Generation Task using Discrete Visual Representations (CVPR 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.
