Research map / Image generation / Autoregressive and latent generation
Auto-Encoding Knowledge Graph for Unsupervised Medical Report Generation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 shared mapping function F in the encoder (ablation) | KGAE unsupervised · IU X-ray | BLEU-4: 0.086 -> 0.109 | Helped |
| add shared mapping function F in the encoder (ablation) | KGAE unsupervised · MIMIC-CXR | BLEU-4: 0.032 -> 0.045 | Helped |
| add knowledge bank (10,000) to decoder (ablation) | KGAE unsupervised · IU X-ray | BLEU-4: 0.109 -> 0.126 | Helped |
| add knowledge bank (10,000) to decoder (ablation) | KGAE unsupervised · MIMIC-CXR | BLEU-4: 0.045 -> 0.062 | Helped |
| add shared mapping function F in the encoder (ablation) | KGAE-Supervised · IU X-ray | BLEU-4: 0.166 -> 0.162 | Hurt |
| add shared mapping function F in the encoder (ablation) | KGAE-Supervised · MIMIC-CXR | BLEU-4: 0.104 -> 0.101 | No clear effect |
| add knowledge bank (10,000) to decoder (ablation) | KGAE-Supervised · MIMIC-CXR | BLEU-4: 0.101 -> 0.118 | Helped |
| increase knowledge bank from 10,000 to 15,000 (ablation) | KGAE-Supervised · IU X-ray | BLEU-4: 0.179 -> 0.170 | Hurt |
| increase knowledge bank from 10,000 to 15,000 (ablation) | KGAE unsupervised · MIMIC-CXR | BLEU-4: 0.062 -> 0.059 | Hurt |
| build knowledge graph from IU X-Ray only instead of MIMIC-CXR (ablation) | KGAE unsupervised · IU X-ray | BLEU-4: 0.126 -> 0.114 | Hurt |
| fine-tune unsupervised KGAE with 100% image-report pairs | KGAE · MIMIC-CXR | BLEU-4: 0.062 -> 0.118 | Helped |
| fine-tune with 60% pairs instead of unsupervised only | KGAE · MIMIC-CXR | clinical efficacy F1: 0.156 -> 0.307 | Helped |
Closest papers by meaning
- DAMPER: A Dual-Stage Medical Report Generation Framework with Coarse-Grained MeSH Alignment and Fine-Grained Hypergraph Matching (AAAI 2025 · similarity 0.85)
- KPL: Training-Free Medical Knowledge Mining of Vision-Language Models (AAAI 2025 · similarity 0.85)
- Instance-level Expert Knowledge and Aggregate Discriminative Attention for Radiology Report Generation (CVPR 2024 · similarity 0.85)
- LLaVA Needs More Knowledge: Retrieval Augmented Natural Language Generation with Knowledge Graph for Explaining Thoracic Pathologies (AAAI 2025 · similarity 0.85)
- PromptMRG: Diagnosis-Driven Prompts for Medical Report Generation (AAAI 2024 · similarity 0.84)
- MEPNet: Medical Entity-Balanced Prompting Network for Brain CT Report Generation (AAAI 2025 · similarity 0.84)
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.