Research map / Vision-language and multimodal / CLIP and zero-shot
Joint Learning of Localized Representations from Medical Images and Reports
Philip Müller, Georgios Kaissis, Congyu Zou, Daniel Rueckert
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 local (region and sentence) weighting (ablation) | LoVT · RSNA YOLOv3 Frozen 10% (ablation, pre-train 30%) | mAP (%): 16.9 -> 17.9 | Improved |
| add attention pooling instead of avg/max pooling (ablation) | LoVT · RSNA YOLOv3 Frozen 10% (ablation, pre-train 30%) | mAP (%): 15.7 -> 16.9 | Improved |
| add global loss to local-only model (ablation) | LoVT · RSNA YOLOv3 Frozen 10% (ablation, pre-train 30%) | mAP (%): 12.3 -> 15.7 | Improved |
| use smooth positiveness in local image loss (ablation) | LoVT · RSNA YOLOv3 Frozen 10% (ablation, pre-train 30%) | mAP (%): 15.6 -> 17.9 | Improved |
| add local report loss to global plus local image loss (ablation) | LoVT · RSNA YOLOv3 Frozen 10% (ablation, pre-train 30%) | mAP (%): 13.7 -> 17.9 | Improved |
| add local image loss to global plus local report loss (ablation) | LoVT · RSNA YOLOv3 Frozen 10% (ablation, pre-train 30%) | mAP (%): 16.2 -> 17.9 | Improved |
| add local losses to global-loss model with attention pooling (ablation) | LoVT · RSNA YOLOv3 Frozen 10% (ablation, pre-train 30%) | mAP (%): 17.4 -> 17.9 | Improved |
| add local losses to global-loss model with avg/max pooling (ablation) | LoVT · RSNA YOLOv3 Frozen 10% (ablation, pre-train 30%) | mAP (%): 14.2 -> 15.7 | Improved |
| use attention pooling with only global loss (ablation) | LoVT · RSNA YOLOv3 Frozen 10% (ablation, pre-train 30%) | mAP (%): 14.2 -> 17.4 | Improved |
| use LoVT instead of ConVIRT | LoVT · RSNA YOLOv3 Frozen 10% (pre-train 30% MIMIC-CXR) | mAP (%): 14.8 -> 17.9 | Improved |
| pre-train on 100% instead of 30% of MIMIC-CXR | LoVT · RSNA YOLOv3 Frozen 10% | mAP (%): 17.9 -> 16.4 | Mixed |
| use LoVT instead of ConVIRT on NIH CXR | LoVT · NIH CXR Linear (pre-train 100%) | Avg Dice (%): 18.6 -> 9.4 | Hurt |
Closest papers by meaning
- Large-scale and Fine-grained Vision-language Pre-training for Enhanced CT Image Understanding (ICLR 2025 · similarity 0.88)
- Bootstrapping Chest CT Image Understanding by Distilling Knowledge from X-ray Expert Models (CVPR 2024 · similarity 0.87)
- MLIP: Enhancing Medical Visual Representation with Divergence Encoder and Knowledge-guided Contrastive Learning (CVPR 2024 · similarity 0.87)
- Unlocking the Power of Spatial and Temporal Information in Medical Multimodal Pre-training (ICML 2024 · similarity 0.87)
- BenchX: A Unified Benchmark Framework for Medical Vision-Language Pretraining on Chest X-Rays (NeurIPS 2024 · similarity 0.86)
- Enhanced Contrastive Learning with Multi-view Longitudinal Data for Chest X-ray Report Generation (CVPR 2025 · similarity 0.85)
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.
