AI Paper Atlas

Research map / Vision-language and multimodal / CLIP and zero-shot

LLM-Guided Diagnostic Evidence Alignment for Medical Vision–Language Pretraining under Limited Pairing

Huimin Yan, Liang Bai, Xian Yang, Long Chen

ICML 2026 · Vision-language and multimodal › CLIP and zero-shot · approach: Language models for zero-shot reasoning · 0 citations (OpenAlex)

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 changedWhereMeasuredEffect
use LGDEA with only 10% paired data instead of fully paired local-alignment baselinesLGDEA · MS-CXR phrase grounding (8 categories)CNR: Atelectasis 0.8312 (MedAligner) -> 0.8449Improved
increase paired data from 5% to 10% (ablation)LGDEA · MS-CXR phrase grounding, Pleural EffusionCNR: 0.9742 -> 1.0928Improved
reduce paired data for baselines to 50% (ablation)MedAligner, GLoRIA, MGCA · MS-CXR phrase groundingCNR: MedAligner Atelectasis 0.8312 -> 0.6346Hurt
add unpaired CheXpert images (cross-domain pretraining) (ablation)LGDEA 10% pair · MS-CXR phrase grounding, Pleural EffusionCNR: 1.0928 -> 0.7519Hurt
use LGDEA with 10% paired data vs MedAligner with full pairsLGDEA · MIMIC-5x200 retrievalPrec@2: 54.08 -> 53.85No clear effect
use LGDEA with 10% paired data vs MedAligner with full pairsLGDEA · RSNA Pneumonia classificationaccuracy (%): 80.05 -> 79.26Hurt
add evidence prototype learning L_rec (ablation)LGDEA · downstream benchmarksperformanceImproved
replace Spark-Desk with Qwen-7B or LLaMA-8B for evidence extraction (ablation)LGDEA · zero-shot classificationaccuracy (%): NIH 87.06 -> 86.14 (Qwen), 85.80 (LLaMA)No clear effect
increase diagnostic prototypes K from 32 to 64 (ablation)LGDEA · COVID classificationaccuracy (%): 87.14 -> 90.47Improved
increase diagnostic prototypes K beyond 64 (ablation)LGDEA · COVID classificationaccuracy (%): 90.47 -> 84.72 (K=128)Hurt
increase lesion queries L from 32 to 64 (ablation)LGDEA · COVID classificationaccuracy (%): 85.10 -> 90.47Improved
increase lesion queries L beyond 64 (ablation)LGDEA · COVID classificationaccuracy (%): 90.47 -> 85.57 (L=80)Hurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy